Primary 6 vocabulary for charts, tables and data interpretation helps a learner explain what a number means before deciding what it proves. This Top 100 Primary 6 Vocabulary List develops Grade 6 data vocabulary through clear meanings, original example sentences and carefully explained comparisons. Words such as frequency, proportion, median, fluctuate and qualify become useful when a student can distinguish a count from a percentage, an average from every individual value, and a missing observation from a recorded zero.

This advanced Grade 6 vocabulary collection connects 6th grade statistics vocabulary, reading tables and graphs, and describing charts in English. Its 100 entries include mathematical terms, language for describing trends, and verbs for writing evidence-based explanations. The teaching moves beyond definitions into original data-reading exercises, worked answers, sentence repairs and an editorial story. The aim is not to make an ordinary chart sound impressive. It is to make the comparison understandable, accurate and appropriately limited.

Charts and tables vocabulary with meanings and examples is especially valuable when a learner can calculate but struggles to explain the result. “The number rose by twenty per cent” does not mean the same thing as “the share rose by twenty percentage points”. “The mean was four minutes” does not establish that everyone waited four minutes. This Primary 6 data interpretation and writing manual teaches those distinctions through fully specified fictional records. Primary 6, Grade 6, sixth grade and Year 6 are different school-system labels, not claims of identical curricula or expectations.

Primary 6 Advanced Vocabulary Collection · Volume 6. Alicia, Tricia and Kai Kai are fictional learners. All clubs, surveys, counts, waiting times, statements and incidents created for the exercises are fictional teaching material. They are not reports about real pupils or measured tuition outcomes. This is an editorial selection of useful words and phrases, not an official prescribed list or a ranking established by word-frequency research. Some entries are suitable for independent use; others are optional extensions to discuss with a teacher.

Choose the reading problem, not a vocabulary quota

Begin with entries 1–10 when the recorded information is unclear, or entries 11–20 when a claim changes from respondents to everybody. Use entries 21–30 for rows, columns and chart conventions; entries 31–40 for displays and frequencies; and entries 41–50 for counts, rates and percentages. The remaining groups cover averages and variation, trends, comparison language, data-quality boundaries and the actions of an accountable writer.

For a complete practice task, choose a data laboratory. All values and display conventions needed for the questions are provided in text; no unseen graph must be guessed. The original editorial story shows why a small missing word can change a whole claim. The writing clinic and teaching sequence then move from supported discussion to a fresh independent response.

Before the calculation, ask what was counted

A fictional library records thirty loans. A pupil writes that thirty children borrowed books. That could be true, but it is not established by the transaction count: one child might borrow several books. The missing distinction is not difficult arithmetic. It is the meaning of the unit being counted. A correct sum of loan transactions can become an incorrect sentence about distinct borrowers when the noun changes. Vocabulary control begins by protecting that noun.

The Common Core Grade 6 statistics standards provide one reference point for describing numerical data with attention to context, observations, units, centre and variability. They do not prescribe this collection or determine every country’s curriculum. Here, numerical work serves an English-writing question: can another person understand precisely what the evidence says? A student may know the calculation while still needing language to explain its scope.

Read a display as a set of choices. Someone chose the population, the question, the counting rule, the period, the categories and the representation. Those choices need not be suspicious; they make a manageable investigation possible. They also determine what can be concluded. A table about completed forms cannot automatically describe people who never submitted a form. A line drawn between two observations cannot turn an unobserved intermediate value into a measured fact.

Keep three stages separate: the supplied record, the calculation made from it, and the interpretation expressed in a sentence. A calculation may be exact within a fictional packet while the wider interpretation remains unsupported. Twenty divided by forty is one half. Whether those forty people represent a larger community is a different question. Checking the division cannot answer it. Each stage needs the kind of evidence appropriate to its own claim.

Entries 1–10: identify the information before interpreting it

The working distinctions in this group are supported by the Australian Bureau of Statistics explanation of variables and Statistics Canada’s introduction to data types. The examples and practice decisions are original. The important move is to decide what a value represents, not merely whether its printed form contains digits.

1. Data

Data are recorded information used for a purpose, such as counts, measurements, categories or responses. A class might record the material of each paper model, its measured length and whether it meets a stated requirement. The data include more than the numbers. A material label such as card or paper is information even though adding those labels would make no mathematical sense.

Recorded information is not automatically a complete or correct account of the subject. A value may have been copied wrongly, or the collection may exclude relevant cases. Calling something data does not remove the need to inspect how it was obtained. Equally, a small classroom dataset can answer a narrow question well without being suitable for a claim about every learner in the world.

Use the data record, data about and the available data. Alicia writes what was observed before offering a conclusion. For practice, separate five statements into supplied information and interpretations added by a reader. Keep an exact count in the first group; place a proposed explanation of why the count occurred in the second unless the packet independently supplies that explanation.

2. Dataset

A dataset is an organised collection of related data. A table recording each model’s identifier, material and length is one dataset when those records belong to the same stated investigation. Its boundaries matter. Ten measurements from Monday and ten from Friday are not automatically twenty different models; the same models may have been measured again.

Combining datasets requires a check of definitions and identities. One file may record length in centimetres and another in millimetres. A shared column heading does not guarantee compatible units. Conversely, two differently named columns may represent the same quantity. Read the accompanying definitions before merging, comparing or declaring that a larger file contains more independent evidence.

Natural phrases include the supplied dataset, combine compatible datasets and the dataset’s coverage. Tricia keeps the source and observation date attached to each collection. For practice, explain what must be checked before adding a second file of returns to the first. Identifying repeated records is different from removing genuine repeated events, so the definition of one record must be established first.

3. Observation

An observation is a recorded instance of the information being studied. In one investigation, an observation might be one model’s measured length. In another, it might be one day’s total loans. The word does not always refer to one person or one moment of looking. Its unit depends on the way the investigation defines a record.

Repeated observations of the same object can be useful, but they do not create additional distinct objects. Measuring one model five times can reveal variation in the readings; it does not show the range of lengths among five independently made models. A writer should preserve whether the record concerns repetition on one subject or comparisons across different subjects.

Use five recorded observations, one observation per day and repeated observations of the same model. Kai Kai names the object and occasion together. For practice, compare a five-row file from one model with a five-row file from five models. Explain a question each file could help answer and a question that the row count alone does not settle.

4. Variable

A variable is a characteristic whose recorded value can differ between observations, such as length, chosen material or number of books borrowed. A variable is not the same as one particular value. Length is the variable; twelve centimetres is a value. A table heading usually names the characteristic, while the entries underneath record its values for particular cases.

A variable can happen to have the same value in every observed case. All five models may be made from card without making material cease to be the characteristic being recorded. Also, a statistical variable need not be something a researcher deliberately changes. It may simply be observed. Do not import the experimental idea of a manipulated variable into every table.

Natural phrases include the variable measured, values of the variable and a categorical variable. Alicia asks which question each column answers. For practice, distinguish a characteristic, its measurement unit and one recorded value. The three belong together but cannot substitute for each other: centimetres is not the model’s length, and twelve does not identify the characteristic without context.

5. Value

A value is the entry recorded for a variable in a particular observation. A model’s material value can be paper; its length value can be twelve centimetres. In ordinary speech, value can mean importance or usefulness. In the table-reading sense here, it describes what is entered, not whether the object or person is valuable.

The same printed value can have different meanings under different headings. Twelve may identify a model, state a measurement or record a count. Only the latter two necessarily support arithmetic as quantities. An identifier of twelve does not mean that model twelve contains twelve objects or is larger than model nine. Read the heading and the record definition together.

Use the recorded value, a missing value and the value in this cell. Tricia repeats the unit in her answer when the table supplies one. For practice, place the number twelve under three different headings and explain the resulting meanings. Then identify which comparisons are justified. Equal digits do not guarantee equal quantities when the variables or units differ.

6. Category

A category is a group defined by a shared characteristic or rule. A survey of one preferred activity might offer reading, drawing and music as categories. The category is meaningful only with its question. Choosing reading as a preferred activity does not establish that the respondent never draws or listens to music at other times.

Categories should be interpreted through their definitions, not through a reader’s assumptions. A category called other may combine several different answers. An unanswered question is not automatically an other response. Some surveys permit more than one selection, creating overlapping memberships; others require a single choice. These designs produce different totals and different appropriate descriptions.

Natural phrases include belongs to the category, a category defined by and the other category. Kai Kai reads whether more than one answer is permitted. For practice, compare favourite activity with activities enjoyed. Explain why the same person might supply one category for the first question and several for the second without contradicting themselves.

7. Categorical

Categorical describes information recorded as types or groups rather than numerical amounts. Material, activity choice and room name can be categorical variables. Numbers can serve as category codes: code 1 may mean paper and code 2 may mean card. The numerical appearance of those labels does not turn material into a measurable amount.

Some categories have a meaningful order, such as low, medium and high. Others do not, such as the names of materials. An order does not automatically establish equal distances between categories. The difference between low and medium cannot be assumed equal to the difference between medium and high unless a measurement rule supplies that relationship.

Use categorical data, a categorical response and counts within categories. Alicia avoids calculating an average material code as though it described a physical mixture. For practice, explain why counting five paper models is meaningful while calling the mean of paper’s code and card’s code a halfway material is not. The calculation can be possible on digits without answering a sensible question.

8. Numerical

Numerical describes data whose values express quantities, such as counts or measurements. The number of models completed and their lengths are numerical variables. Their units and meanings still differ. Adding completed-model counts across disjoint groups can give a total count, whereas adding their lengths answers a different question about combined length.

Not every arithmetic operation is useful for every numerical variable. Averaging a set of durations can describe a typical amount of time in one sense, but it does not identify the longest wait. Numerical data still require a well-chosen question. A spreadsheet’s ability to calculate a result does not establish that the result represents what the writer claims.

Natural phrases include numerical measurements, a numerical variable and numerical rather than coded categorical values. Tricia names the quantity before selecting an operation. For practice, distinguish pupil identification numbers from the number of books each pupil borrowed. Both columns contain digits, but only one describes the borrowing quantity relevant to a total or mean.

9. Discrete

Discrete numerical values take distinct, separated possibilities rather than every value in an interval. A count of completed models can be zero, one, two and so on. A count of two and a half completed models would not fit a rule requiring whole completed objects. Partially completed work would need a separate quantity or an explicitly different counting convention.

Discrete does not simply mean that the written number has no decimal point. A variable taking only quarter-unit steps is also discrete under that definition. Conversely, recording a length to the nearest whole centimetre does not make the underlying length incapable of lying between centimetre marks. The possible values of the characteristic and the recording resolution are different questions.

Use a discrete count, discrete possible values and recorded in fixed increments. Kai Kai distinguishes whole-object counts from measured quantities. For practice, explain why a mean of 2.5 books per pupil can arise from individual whole-number counts. A summary statistic need not be one of the possible values for a single observation.

10. Continuous

Continuous describes a numerical characteristic modelled as capable of taking any value within a relevant interval. Length and elapsed time are familiar examples. A paper strip can be slightly longer than twelve centimetres without reaching thirteen. Recording its length as twelve may reflect the instrument or rounding rule rather than a claim of perfect exactness.

Continuous measurement is not the same as continuous observation through time. A duration is a continuous quantity even when only one reading is taken. Likewise, a count recorded every second remains a count. Do not confuse how often information is collected with the kind of values the variable can take. The adjective’s meaning depends on the phrase it modifies.

Natural phrases include a continuous variable, continuous measurements and recorded to the nearest unit. Alicia compares the underlying characteristic with its recorded form. For practice, describe a timing task whose results are rounded to whole seconds. Explain what the displayed integers establish and why adding invented decimal places would not recover the unrecorded timing detail.

Entries 11–20: preserve who and what the evidence covers

The ABS guide to censuses and samples distinguishes whole-population collection from collection involving a subset. Its metadata guidance explains the value of information about how data were produced. In the following fictional cases, these ideas help prevent a sentence about returned forms from quietly becoming a sentence about every eligible person.

11. Population

A population is the complete group defined for the question under investigation. It could be all sixty members of a fictional club, all eligible models made in a workshop, or all recorded loans during one month. The word does not refer only to people living in a country. Its boundaries come from the investigation’s stated purpose.

A population needs enough definition to decide whether a case belongs. All club members could mean members on a particular date, anyone who belonged during a year, or everyone currently receiving notices. Those groups may differ. Before using all, every or the whole club, check that the dataset covers the population named in the sentence.

Use the target population, the defined population and members of the population. Tricia keeps the membership date beside the count. For practice, distinguish all enrolled members from all visitors during a session. Explain why one group cannot stand in for the other merely because both are associated with the same club and appear in similarly titled records.

12. Sample

A sample is a subset of a defined population selected or observed for an investigation. Ten responding members among sixty eligible members provide information about those ten. Whether their answers support a wider estimate depends on how they were selected and on other features of the collection. The word sample does not itself guarantee representativeness.

A large sample can still systematically omit a relevant part of the population. Asking only people present at an early session may miss members who cannot attend early. Increasing that same convenience sample does not automatically correct its coverage problem. A sample’s size is one consideration, not a substitute for examining selection and response.

Natural phrases include the observed sample, a sample drawn from and the sample’s selection method. Kai Kai reports eight of the ten respondents rather than eight of every ten members. For practice, state one conclusion directly supported by the responses and one proposed population conclusion requiring additional justification. Keep the useful local result while identifying the wider claim’s extra burden.

13. Census

A census aims to collect information about every member of a defined population. A complete inventory of all eligible models in one specified cupboard can be a census of that cupboard’s contents. It is not automatically a census of every model in the school. The completeness claim remains attached to its stated boundary.

Attempting to contact everyone does not establish that complete usable information was obtained from everyone. Some cases can be missed, entered twice or recorded incorrectly. A census avoids the particular uncertainty arising from selecting only a sample when it fully enumerates the population, but it can still contain collection and processing errors. Completeness and accuracy require separate attention.

Use a census of the defined group, complete enumeration and an attempted census. Alicia distinguishes invitations sent from valid responses received. For practice, describe an investigation that contacted all sixty members but obtained fifty answers. Explain why all invited and all answered are different findings, even when the organiser intended to collect the whole population’s views.

14. Respondent

A respondent is a person or reporting unit that provides an answer in a survey or similar collection. In the club examples, each respondent is an eligible member who submits a usable answer to the specified question. A respondent is not the same as a response: one person may answer several questions or accidentally submit the same form twice.

The respondent count can differ between questions when some people skip particular items. A survey with twenty participating members does not necessarily have twenty valid answers to every question. A percentage about one question should use its appropriate denominator or state how missing responses were handled. Otherwise, a seemingly small omission can change the represented share.

Natural phrases include among respondents, eligible respondents and respondents to this question. Tricia checks unique member identifiers before counting people. For practice, explain how twelve submitted forms can correspond to eleven respondents. Then distinguish an accidental duplicate from a legitimate follow-up response collected at another time under a different rule.

15. Response

A response is an answer given to a question or request. A member’s choice of drawing for a preferred activity is one response to that item. In a question allowing several selections, one completed response can contain several chosen categories. Read the question’s format before deciding whether category counts should sum to the number of people.

A response records what was answered, not automatically the underlying behaviour. A statement that someone prefers reading does not prove how many minutes they read yesterday. Their answer may be sincere while the writer’s behavioural conclusion is unsupported. Do not make a preference survey carry a measurement that it was not designed to collect.

Use a response to the item, valid responses and multiple-response questions. Kai Kai keeps selections separate from people. For practice, compare a form asking for one favourite activity with a form asking for every activity enjoyed. Explain why their category totals can behave differently even if the same five members complete both forms accurately.

16. Unit of analysis

The unit of analysis is the kind of entity about which an analysis makes its comparisons or conclusions. An account can examine individual pupils, loan transactions, daily totals or whole clubs. These are different units. Thirty loan records can support analysis of transactions without identifying thirty distinct borrowers or thirty separate days.

Moving between units requires a justified transformation. To describe borrowers from loan records, the writer needs a reliable way to group transactions by distinct people. To describe days, the writer needs dates and a rule for daily totals. A row count answers a question about rows only until the meaning of one row has been established.

Natural phrases include the analysis treats each club as one unit and the unit changes from transactions to people. Alicia asks what receives one count. For practice, explain how the same library records can support a transaction total and a borrower total. State the additional identity information needed for the second result rather than pretending both totals must be equal.

17. Eligibility

Eligibility is whether a case meets the stated conditions for inclusion. A survey may include members registered before a particular date and exclude visitors. The rule concerns which cases belong in this collection. It is not a judgement of anyone’s worth, and exclusion from this survey does not mean exclusion from every club activity.

Eligibility and participation are different. An eligible member may not respond; a submitted answer may come from someone outside the eligible group. Counting every submitted form without checking the inclusion rule can therefore distort the result. Any decision about an ambiguous case should follow a declared rule rather than a preference for the answer that improves the headline.

Use eligible members, meets the eligibility rule and eligibility for this survey. Tricia compares registration dates with the stated cutoff. For practice, explain why a late registrant’s answer might be preserved separately but not added to this particular denominator. Preserving a record and including it in a defined calculation are not the same decision.

18. Coverage

Coverage is the people, places, periods or categories included in a collection. A record covering weekday loans does not cover weekend borrowing. A table covering public workshops does not necessarily include private sessions. The title may summarise the topic broadly, while a method note supplies the limits needed for an accurate statement.

A complete count within narrow coverage should not be mistaken for incomplete work. It may answer its intended question exactly. The mistake occurs when a writer extends it beyond that boundary. Conversely, a record can aim at broad coverage and still miss eligible cases. Intended scope and achieved coverage should be distinguished whenever the evidence makes that difference important.

Natural phrases include coverage of weekdays only, outside the survey’s coverage and the coverage note. Kai Kai adds weekday before a total drawn only from Monday to Friday. For practice, identify the shortest qualification that stops a limited dataset from becoming a universal claim. A single accurate modifier can preserve the finding without requiring a paragraph of vague warnings.

19. Nonresponse

Nonresponse is the absence of a requested answer from an eligible person or to a particular item. A blank preference box is not automatically a vote for none. The respondent may have skipped it, misunderstood it or chosen not to answer. The available record may not establish why the answer is missing.

Nonresponse can affect a whole form or only one question. It can also matter to population conclusions when respondents differ systematically from people who did not respond. The presence of missing answers does not prove a particular direction of bias. State what is unknown and avoid inventing either favourable or unfavourable answers to complete a desired total.

Use item nonresponse, nonresponse among eligible members and the unanswered items. Alicia separates missing answers from explicit none responses in the record. For practice, compare ten invitations, eight returned forms and seven valid answers to one question. Identify the denominator for describing those seven answers and the different denominator for reporting the form-return rate.

20. Metadata

Metadata are information explaining a dataset’s contents, origins and conventions. A note may state when observations were collected, which people were eligible, what each column means and which symbol represents missing data. These details help the reader interpret the values. They are not merely technical decorations added after the important numbers have been printed.

The dataset’s publication date may differ from the observation date. A new table can reproduce an older survey without collecting anything new. Similarly, a file named final may contain a version that later receives a correction. Metadata help identify these relationships, but their presence alone does not guarantee correctness. Read the actual definitions rather than trusting reassuring labels.

Natural phrases include the metadata specify, consult the metadata and the observation-date note. Tricia reads the unit and missing-value convention before calculating. For practice, write a four-sentence metadata note for a fictional class table. Include its population, variable, collection period and missing-value rule. Each sentence should prevent a plausible misunderstanding of the entries.

Entries 21–30: read the structure carrying the number

A correct number taken from the wrong column is still the wrong answer. Read the row label, column heading and unit as one combined address. For charts, inspect how positions and lengths encode values. The Office for National Statistics guidance on axes and gridlines distinguishes zero-based bar lengths from the different choices possible for line charts. The exercises below use that distinction without treating every nonzero axis as dishonest.

21. Table

A table organises information into rows and columns so that related entries can be located and compared. A library table might give each club’s member count, distinct borrowers and loan transactions. The arrangement puts several variables beside the same club. It does not make those variables interchangeable just because their values appear in neighbouring cells.

A table may contain totals and subtotals as well as individual records. Adding every numerical row can therefore double-count information. A grand total may already include the group totals above it. Identify the structure before calculating. Blank spaces, dashes and symbols also need their stated meanings; none should automatically be treated as the number zero.

Use according to the table, the table records and the relevant table entry. Alicia reads an entry as a full sentence before using it. For practice, explain what thirty means in the Moon Club loan column and why the same value cannot be reported as its number of borrowers without checking that different column.

22. Row

A row runs horizontally across a table. Its label often identifies the person, group, date or category to which its entries belong. Reading along the Sun Club row keeps several variables connected to Sun Club. Moving down into Moon Club’s row changes the subject, even if the column remains the same.

Not every row is one person. A row may represent a whole month or contain a total for several groups. Nor does a row necessarily describe an independent observation when the same subject appears on several dates. The meaning of a row comes from the table’s design, not from its visual shape alone.

Natural phrases include the row labelled, read across the row and the total row. Tricia follows the label before taking a number. For practice, deliberately take the correct column’s value from the neighbouring row and identify the changed claim. This tests whether the learner checks the subject of the answer, not only whether its arithmetic looks plausible.

23. Column

A column runs vertically through a table. Its heading usually identifies the variable recorded in its cells. The loan-transactions column allows comparison of transactions across clubs, while the distinct-borrowers column answers a question about people. A writer who shifts columns without changing the noun creates a meaning error before any calculation begins.

Some tables use several levels of headings. A year heading may cover separate columns for count and percentage. Read both levels. A percentage under one year is not the same quantity as a count under another. Units can appear in a shared heading rather than beside every number, so the lack of a repeated unit in each cell does not mean there is none.

Use the column headed, compare within the column and the column’s unit. Kai Kai states the complete heading in his working. For practice, explain why a total should not combine a member count and a percentage merely because both are numerical entries in the same row. Their column meanings determine which operations make sense.

24. Cell

A cell is one position at the intersection of a row and a column. Its meaning depends on both headings. The cell where Sun Club meets distinct borrowers records the relevant borrower count for that club. Referring to the cell as the number on the left is less dependable because the table’s layout may change.

A cell can contain a value, a label or a missing-data symbol. Its emptiness does not establish an observed absence. A footnote marker beside a value may also modify its meaning, indicating an estimate or a changed definition. Read the marker before copying only the digits into a sentence.

Natural phrases include the cell at the intersection, the marked cell and a cell containing missing data. Alicia names the row and column instead of relying on page position. For practice, turn a single cell into a full statement including group, variable, value and period. The exercise reveals information that the surrounding structure supplies silently.

25. Heading

A heading labels a section, row, column or whole display. It tells the reader what the following information concerns. A heading such as loans completed during March differs from people who borrowed during March. Those labels can accompany different counts even when they describe the same library activity.

A shortened heading can create ambiguity. Average time may need to specify whether it means the arithmetic mean, median or another summary. Number of users may need to distinguish distinct people from visits. A useful heading is not necessarily long, but it must retain the terms that determine the quantity. More attractive wording is not an improvement when it changes the measure.

Use the column heading specifies, an informative heading and a misleading heading. Tricia checks the heading against the data definition. For practice, repair a title that says children when the table counts visits. Explain why changing the title is a factual correction rather than merely a matter of tone or presentation.

26. Legend

A legend explains symbols, colours or line styles used in a display. One line might represent completed loans and another returned books. Without the legend or direct labels, a reader may assign the values to the wrong series. A familiar colour does not have one universal statistical meaning across every chart.

The legend can also distinguish measured values from estimates or actual results from projections. A dashed line is not automatically missing data; its meaning depends on the display’s convention. Likewise, one icon in a pictogram might represent five items rather than one. The symbol’s interpretation must come from the stated rule, not its resemblance to a real object.

Natural phrases include the legend identifies, according to the legend and a separately labelled estimate. Kai Kai reads the legend before following a line. For practice, give the same two line styles different legends and explain how the supported caption changes. The marks may remain identical while the variable represented by each changes completely.

27. Axis

An axis is a reference line along which positions, categories or values are organised on a graph. The plural is axes. One axis may show weeks and the other completed loans. Their labels and scales explain how to read a plotted point. A point’s position is not a numerical answer until the reference system is understood.

The horizontal axis does not always show time, and the vertical axis does not always show a count. A horizontal bar chart can place categories vertically and quantities horizontally. Some graphs show two numerical variables. Read the actual labels rather than applying a memorised page-direction rule that belongs only to one chart type.

Use the horizontal axis, the quantity axis and the axes are labelled. Alicia names each axis before interpreting a point. For practice, explain why moving from one chart to another requires checking both labels again. The same printed position may represent a different quantity when the scale, unit or variable has changed.

28. Scale

A scale specifies how positions or intervals on an axis correspond to numerical values. Equally spaced labels of zero, ten, twenty and thirty indicate ten units per labelled interval on that linear scale. A point halfway between twenty and thirty represents twenty-five if the scale is continuous and linear there, not automatically twenty-one.

Compare scale before comparing apparent heights across separate charts. A tall bar on one chart may represent a smaller value than a short bar on another if their axes use different ranges. The width of a line chart also affects how steep a trend looks. Read the values and time intervals instead of making a numerical comparison solely from the angle of a line.

Natural phrases include the axis scale, equal intervals on a linear scale and different scales. Tricia records the unit represented by one interval. For practice, compare two displays of the same values using different vertical ranges. Explain which appearance changes and which recorded differences remain unchanged.

29. Baseline

A baseline is a reference from which a quantity or change is compared. In a bar chart, the baseline may be the zero line from which bar lengths represent values. In a before-and-after comparison, baseline may instead refer to the initial measurement. State the intended sense: a chart’s zero line and a study’s starting result are different references.

Changing the baseline can change a visual or percentage comparison. Bars for eighty and one hundred drawn upward from sixty have visible lengths of twenty and forty. Their visible lengths double, but the values themselves do not. In a percentage-change calculation, changing the starting quantity similarly changes the denominator against which the difference is assessed.

Use a zero baseline, the baseline value and relative to the stated baseline. Kai Kai checks what the chart’s lengths actually encode. For practice, explain the difference between a twenty-unit increase and a doubling in the eighty-to-one-hundred comparison. The first follows from the values; the second can arise only from misreading the cropped bar lengths.

30. Annotation

An annotation is a note attached to a display to clarify a feature, event or limit. A note saying Wednesday’s record was lost prevents a blank position from being read as zero activity. Another note might identify where a counting definition changed. The annotation connects contextual information with a particular part of the chart.

An annotation can explain an event’s timing without proving its causal effect. Marking the date a new notice appeared does not establish that the notice caused a later rise in responses. The note should report what is known. A reader must still distinguish temporal coincidence from evidence of a mechanism or a supported causal comparison.

Natural phrases include an annotation explains, annotate the missing period and a contextual annotation. Alicia writes the observation period beside an older survey result. For practice, add one note that prevents a specific misunderstanding. Avoid a vague phrase such as use carefully when the real issue can be named precisely as a missing day or a changed denominator.

Entries 31–40: distinguish a representation from the values it represents

The ONS bar-chart guide discusses categorical comparison, while its line-chart guide addresses change over time and gaps in observations. The NIST handbook’s histogram explanation provides a reference for grouped numerical distributions. The following entries teach the question each display can answer without treating one chart type as universally best.

31. Bar chart

A bar chart represents values for categories using bar lengths. It can compare completed loans across named clubs or responses across preferred activities. The categories need not have a numerical distance between them. Moving a category’s position for a clear ordering does not change its count, provided the label moves with its bar.

When ordinary bar lengths represent absolute values, they should begin at zero so the lengths preserve those comparisons. A bar chart can also represent a derived difference, including negative differences; its axis then needs to identify that different variable. Do not describe every negative bar as an impossible negative count when the graph actually shows change rather than a total.

Use a bar representing, compare the bar values and a zero-based bar chart. Tricia names the encoded quantity before commenting on height. For practice, distinguish a chart of eighty and one hundred completed forms from a chart of changes of minus five and plus ten. The latter bars describe differences, not the number of forms that exist.

32. Line chart

A line chart connects plotted values in a meaningful sequence, commonly to show change over time. Four weekly counts can be plotted in their date order so a reader can see a rise, a fall and another rise. The order matters: sorting the weeks by their counts would change the temporal story while leaving the individual numbers unchanged.

A line between observations is a display choice, not evidence that every intermediate value was measured. Missing observations should remain distinguishable from measured ones. A nonzero vertical starting point can be appropriate for showing variation in a line chart when its labels make the range clear; this differs from the length-encoding issue in ordinary bars.

Natural phrases include the line connects recorded values, a gap in the series and the time axis. Kai Kai separates a visible segment from an actual intermediate observation. For practice, explain what is known between Monday’s and Thursday’s readings when no readings were made on Tuesday or Wednesday. A smooth line does not supply those missing records.

33. Pictogram

A pictogram uses repeated pictures or symbols to represent quantities. If its key states that one book symbol represents four loans, five symbols represent twenty loans. The symbol is a counting unit in the display, not necessarily an illustration of one actual borrowed book. The key determines the calculation.

Partial symbols need an explicit interpretation. A half symbol can represent two loans under the four-loans rule, but the reader should not assume every decorative half-image has that meaning. Enlarging an icon in both dimensions is another issue: its visual area changes faster than its width. Simply making one picture bigger is not equivalent to adding clearly defined repeated symbols.

Use each symbol represents, the pictogram’s key and a half symbol. Alicia converts symbol counts into the actual unit before comparing. For practice, describe twelve icons under a five-items-per-icon key and the same icons under a ten-items-per-icon key. The picture count stays twelve while the represented totals differ.

34. Pie chart

A pie chart represents a whole divided into parts, using sectors of a circle. For a single-choice question with complete, mutually exclusive categories, each respondent contributes to one part. A quarter of the whole corresponds to one quarter of the circle. The chart’s denominator must be clear before a sector can be interpreted as a number of people.

Overlapping selections cannot simply become separate parts of one pie when the same person belongs to several categories. Their percentages may legitimately total more than one hundred in a multiple-response survey. That does not mean a circle can contain more than its whole. The categories need a different representation or a new mutually exclusive grouping such as reading only, music only and both.

Natural phrases include a sector represents, a share of the stated whole and mutually exclusive parts. Tricia checks the question’s selection rule. For practice, explain why a fifty-per-cent sector does not identify a count until the total is known. Half of twenty and half of sixty describe the same share but different numbers.

35. Dot plot

A dot plot, in the elementary statistical sense used here, places a dot for each observation above its numerical value on a number line. Repeated values produce stacks. For the values two, two, three, three and ten, there are two dots at two, two at three and one at ten. The plot preserves the individual observed values.

The dots do not necessarily identify named individuals. They show observations unless another label supplies identity. A different kind of chart may also be called a dot plot, so read the display convention. In this collection, one dot means one recorded observation, not a group of five and not a geographical point.

Use a stack of dots, one dot per observation and the dot at ten. Kai Kai counts all dots before finding a summary. For practice, explain what information a dot plot of the five values retains that a single mean does not. The average alone cannot reconstruct the repeated small values and the one larger observation.

36. Histogram

A histogram groups numerical observations into intervals and represents their distribution with adjoining rectangles. In an equal-width frequency histogram, the rectangle heights can represent the counts in those intervals. This differs from a categorical bar chart: the numerical intervals have an ordered scale and meaningful widths, not merely separate category names.

Grouping loses some detail. A count of four observations between five and ten does not reveal their four exact values. Bin boundaries must also avoid ambiguous overlap. Unequal-width histograms require attention to how height and area are defined, so do not carry the equal-width frequency rule into every histogram without reading its vertical-axis label.

Natural phrases include an equal-width histogram, the interval contains and grouped numerical data. Alicia checks whether an exact value of five enters the interval beginning at five. For practice, compare a full value list with its grouped histogram counts. Identify one question still answerable and one exact-value question that the grouped representation cannot settle.

37. Interval

An interval is a span between specified values or times. A data interval might include values from zero up to but not including five. The next could include five up to but not including ten. The boundary wording places an exact five in the second interval and prevents it from being counted twice.

An axis interval, a grouping interval and a time between observations are related but different uses. A five-minute gap between readings does not tell the reader the grouping width of their values. Likewise, a five-unit bin does not imply that observations were made every five seconds. State which kind of interval is being described.

Use within the stated interval, equal-width intervals and the interval between observations. Tricia tests exact boundary cases before counting. For practice, assign values of 4.9, 5 and 9.9 to the two stated bins. Explain why rounding the values first could change their assigned intervals and therefore answer a different question.

38. Frequency

Frequency is how many times a particular value or category occurs in the observations being counted. If four of eight pupils report exactly one borrowed book, the frequency of one book is four. One is the value of the borrowing variable; four is the number of observations with that value. Both numbers need the correct noun.

The ABS explanation of frequencies distinguishes this count from a relative frequency. A table can also describe frequency over a period, such as visits per week, but that rate needs its time reference. Do not read every occurrence of frequency as an unqualified percentage or a count of distinct people.

Use the frequency of this category, occurs four times and a frequency table. Kai Kai reads a value-frequency pair aloud. For practice, explain why four pupils each borrowing one book contribute four observations to that frequency and four books to a total, whereas four pupils borrowing three each contribute twelve books but still four observations.

39. Relative frequency

A relative frequency compares a category’s frequency with the relevant total number of observations. Four occurrences among eight observations give four eighths, or one half. It can be expressed as a fraction, decimal or percentage. The denominator makes the count interpretable as a share of this particular collection.

A larger frequency does not necessarily mean a larger relative frequency. Six among twenty is a larger count but a smaller share than four among eight. The comparison changes because the totals differ. A relative frequency observed in one sample is also not automatically an exact probability for every future occasion or a known population proportion.

Natural phrases include a relative frequency of one half, the observed share and relative to the total. Alicia states both numerator and denominator before converting to a percentage. For practice, construct two groups in which the larger count has the smaller share. Explain both findings in one sentence without suggesting that either calculation contradicts the other.

40. Cumulative frequency

Cumulative frequency adds frequencies up to a stated point in an ordered variable. Suppose three pupils borrowed zero books, four borrowed one and one borrowed two. The frequency for one book is four, but the cumulative frequency for at most one book is seven. The latter includes the zero-book group as well as the one-book group.

The cutoff wording matters. Fewer than one book includes only zero in this whole-number example, while at most one includes both zero and one. Cumulative frequency is meaningful where the values can be ordered appropriately; accumulating arbitrary activity names in alphabetical order usually answers no useful question about amount.

Use cumulative frequency up to, at most and fewer than. Tricia identifies which categories belong in the running total. For practice, explain why seven is not the number who borrowed exactly one book. A correct addition becomes an incorrect statement when the word exactly replaces the inclusive cutoff that justified the calculation.

Entries 41–50: keep the whole beside the part

A comparison needs a reference. More participants, a greater participation share and a faster arrival rate are three different findings. The ONS guide to percentages and percentage points and the House of Commons Library explanation of percentage calculations support the distinction between absolute differences and relative changes. Every numerical example below is an original fictional teaching case.

41. Total

A total is the complete amount obtained for a stated collection or by adding its relevant parts. Twenty-four loans from one club and thirty from a separate club give fifty-four loan transactions when the records cover the same period and do not duplicate transactions. The word loans must remain attached: the total does not automatically describe fifty-four different people.

Before adding parts, check that they belong to the same whole and do not overlap. A table may include a subtotal alongside the categories already included in it. Adding that subtotal again counts the same information twice. Likewise, adding a monthly cumulative total to the daily values from which it was formed does not create extra activity.

Use a total of, the total for the stated period and included in the total. Alicia identifies which rows are parts and which summarise them. For practice, explain why thirty books and twenty borrowers should not be added into a total of fifty people. A valid sum requires compatible meanings, not merely available numbers.

42. Count

A count states how many qualifying items, events or people are included under a rule. Counting visits means giving each qualifying visit one place in the total. Counting distinct visitors means giving each qualifying person one place, even when that person visits repeatedly. Both can be useful measures of library activity, but they answer different questions.

A count depends on eligibility and identity. A cancelled loan may be excluded from completed transactions; two entries with the same transaction identifier may represent an accidental duplicate. Do not remove repeated names automatically, because one person can make several genuine transactions. The correct deduplication rule follows what is being counted.

Natural phrases include a count of distinct borrowers, count each transaction once and the counting rule. Tricia states the rule before comparing records. For practice, give one fictional pupil three separate loan records. Explain why the pupil contributes three to a transaction count but one to a distinct-borrower count for that period.

43. Proportion

A proportion expresses a part in relation to its whole. Twelve borrowers among twenty members give a borrower proportion of twelve twentieths, or 0.6. The numerator counts the qualifying part and the denominator identifies the whole membership being considered. A part-to-whole proportion in this sense lies between zero and one.

The whole must match the question. Twelve borrowers among twenty members is not the same comparison as twelve borrowers among twelve survey respondents. Changing the denominator changes the meaning even when the numerator stays twelve. A ratio of borrowers to non-borrowers is another comparison again; it does not use the entire membership as its second quantity.

Use the proportion of members who borrowed, a larger proportion and part of the defined whole. Kai Kai writes both groups in words before dividing. For practice, compare twelve among twenty with twelve among forty. Explain why equal borrower counts can produce different proportions without either record being inconsistent.

44. Percentage

A percentage expresses a ratio in hundredths. A part-to-whole proportion of 0.6 is sixty per cent, written 60%. This does not require a group of exactly one hundred people. Twelve of twenty and sixty of one hundred represent the same percentage while describing different counts. The percent sign changes the expression, not the underlying relationship.

Percentages can describe other ratios, including relative change, so not every percentage is a share of a group. A growth of 150% is possible because it compares an increase with an earlier value. That differs from claiming that 150% of distinct people belong to one subset of themselves. Always identify what is divided by what.

Natural phrases include sixty per cent of the members, the percentage share and expressed as a percentage. Alicia keeps of the members beside the figure. For practice, explain why 60% alone cannot tell a reader how many people responded. A count requires the size of the whole as well as its stated percentage.

45. Denominator

The denominator is the lower part of a fraction and identifies the quantity by which the numerator is divided. In twelve borrowers out of twenty members, twenty is the denominator. Its meaning matters as much as its number. Dividing by all members, responding members or opening hours produces different kinds of result.

A changed denominator can make a share fall while its count rises. Twenty participants among forty people is 50%; twenty-four among sixty is 40%. The participating count increases by four, but the whole grows by more in relative terms. Neither finding should be hidden because it complicates a preferred positive or negative headline.

Use the denominator represents, a common denominator for comparison and a changed denominator. Tricia asks out of how many before reading a percentage aloud. For practice, explain why a percentage cannot be calculated by ordinary division when the required denominator is zero. Report the undefined comparison rather than inventing a zero-per-cent result.

46. Numerator

The numerator is the upper part of a fraction. In a participation proportion, it counts the people meeting the stated condition. If twenty-four of sixty eligible members signed up, twenty-four is the numerator. It must refer to members within that sixty-person whole, not to additional people collected from an unrelated group.

A numerator can be misdefined even when the division is accurate. Counting all submitted forms as distinct participants may include a duplicate form. Counting people who expressed interest as people who actually attended changes the condition. Check the numerator’s membership rule rather than treating it as a neutral number whose meaning is already secure.

Natural phrases include included in the numerator, the numerator counts completed responses and the qualifying cases. Kai Kai checks whether every counted case satisfies the claim. For practice, distinguish twelve applications, ten approvals and eight attendances. Write three different fractions using an appropriate whole and explain why their numerators cannot be exchanged without changing the statement.

47. Ratio

A ratio compares two quantities in a specified order. If twelve members borrow and eight do not, the borrower-to-non-borrower ratio is twelve to eight, or three to two. The borrower proportion is instead twelve out of twenty. A part-to-part comparison and a part-to-whole comparison use different second quantities.

Reversing the named order reverses the ratio. Non-borrowers to borrowers is two to three, not three to two. Equivalent ratios preserve the relationship: six to four and three to two express the same comparison, although their actual group sizes can differ. A ratio alone generally does not establish the total number in either category.

Use the ratio of borrowers to non-borrowers, in the ratio three to two and an equivalent ratio. Alicia expands a ratio into a complete sentence before using it. For practice, explain why a three-to-two split gives three fifths in the first category, not three halves of the whole group.

48. Rate

A rate compares a quantity with another relevant measure, often time. Twenty-four transactions during ten opening hours give an average rate of 2.4 transactions per hour. The fractional result is a summary over the period; it does not imply that 0.4 of a transaction was completed as a separate event.

A higher total can accompany a lower rate when the observation period is longer. Thirty transactions over fifteen hours give two per hour, below the 2.4 rate despite the larger total. An average rate also does not establish an even schedule. The transactions may have clustered in particular hours, which would require more detailed timing records to examine.

Natural phrases include an average rate per hour, the rate over the stated period and a rate rather than a total. Tricia checks both count and duration. For practice, compare two equal totals observed over unequal time spans. Explain why identical counts do not establish identical rates or identical experiences within the periods.

49. Percentage point

A percentage point is the unit used for the direct difference between percentage values. A participation share moving from 50% to 40% falls by ten percentage points. This is not a ten-per-cent relative fall from the original share. The relative fall is ten divided by fifty, or 20% of the starting percentage value.

The two descriptions answer different questions and can both be correct when clearly labelled. They should not be exchanged to make a result sound larger or smaller. Naming the starting and ending shares helps the reader check the difference. The underlying participating counts may move in another direction when the population sizes change, as the earlier denominator example shows.

Use a fall of ten percentage points, rose by two percentage points and the percentage-point difference. Kai Kai keeps the whole phrase rather than writing an ambiguous points. For practice, describe a share moving from 20% to 25% in percentage points and relative percentage change. Explain why the answers are five and twenty-five, respectively.

50. Percentage change

Percentage change compares a change with its starting value. For a positive starting value, subtract the original from the new value, divide that difference by the original, and multiply by one hundred. Moving from twenty participants to twenty-four gives four divided by twenty, or a 20% increase. The new count is 120% of the original, not a 120% increase.

Successive percentage changes use successive bases. An increase from one hundred to one hundred and twenty is 20%. A subsequent decrease of 20% of one hundred and twenty removes twenty-four, leaving ninety-six rather than returning to one hundred. Matching percentage labels do not automatically cancel. The actual reference quantity at each stage controls the calculation.

Use increased by twenty per cent from the original count, the relative change and the starting value. Alicia names the base in her working. For practice, compare the twenty-to-twenty-four increase with the reverse change. The reverse fall is four out of twenty-four, not four out of twenty, because the starting point has changed.

Entries 51–60: summarise a group without inventing a typical person

The ABS guide to measures of central tendency distinguishes mean, median and mode. Its guide to measures of spread explains why a centre alone does not describe variation. In these entries, the short fictional lists are complete as stated. A summary should illuminate their structure rather than replace the values with an invented experience shared by everyone.

51. Mean

The mean here is the arithmetic mean: add the numerical values and divide by the number of observations. Waiting times of two, two, three, three and ten minutes total twenty minutes across five observations, giving a mean of four minutes. Four summarises the distribution in this way even though nobody in the list waited exactly four minutes.

The mean uses every value and can be affected strongly by a comparatively large or small one. It is not automatically the most frequent value or a maximum anyone experienced. An equal mean for two groups does not establish equal distributions. Five waits of four minutes also have a mean of four but show none of the variation in the first list.

Natural phrases include the mean waiting time, an arithmetic mean of and the mean across five observations. Tricia states the sample size beside the result. For practice, explain what the two equal-mean lists share and what they do not. Avoid calling every person average simply because the group can be summarised by one number.

52. Median

The median is a middle value of ordered numerical observations. With an odd number of observations, use the middle position after sorting. With an even number in these elementary exercises, take the mean of the two middle values. For two, two, three, three and ten, the median is three because the third ordered value occupies the middle position.

Repeated values mean it is misleading to insist that exactly half the observations are strictly below the median and exactly half strictly above. Some can equal it. The median also need not be an observed value when two middle values are averaged. Sorting is essential: the middle item in an unsorted list may be unrelated to the distribution’s centre.

Use the median of the ordered values, the two middle observations and a median waiting time. Kai Kai sorts before selecting. For practice, reorder the same five values in several ways and explain why the median should not change. The original presentation order is not part of this particular summary calculation.

53. Mode

The mode is a value or category occurring with the greatest frequency. In two, two, three, three and ten, two and three tie for the greatest frequency, so the list has two modes under the usual elementary convention. Neither occurs in a majority of the five observations. Most frequent is not the same as more than half.

A mode can describe categorical data, such as the most commonly chosen material, when a numerical mean would make no sense. Several categories can tie. Where all numerical values occur once, school conventions often describe the list as having no mode; state the convention rather than claiming a uniquely most frequent value that the frequencies do not distinguish.

Natural phrases include the modal category, two tied modes and the most frequent value. Alicia checks the frequency count instead of selecting the largest number. For practice, explain why ten is the maximum but not a mode in the five-value list. Different summary terms ask different questions about the same information.

54. Minimum

The minimum is the smallest value in the stated set or the lower limit defined by a rule. In the observed waiting-time list, two minutes is the minimum. That does not establish that nobody outside this dataset ever waited less than two minutes. The comparison set belongs to the claim.

An observed minimum differs from a required minimum. A rule saying at least two completed tasks states a threshold; an observed minimum of two reports what occurred in the supplied group. One is a condition, the other a summary. A writer should not treat the observed value as evidence that a rule existed or that every future case must remain above it.

Use the minimum observed wait, a minimum requirement and at least. Tricia checks whether the word reports a result or defines eligibility. For practice, write one sentence about a dataset’s smallest value and another about a task’s minimum standard. Explain what additional evidence would be needed to claim that all observed cases met that standard.

55. Maximum

The maximum is the largest value in a stated set or the upper limit defined by a rule. Ten minutes is the maximum of the five recorded waits. It is not the mean, the total or the most common wait. A sentence should name which of these relationships is being reported rather than describe ten as the biggest answer.

An observed maximum does not by itself establish the system’s capacity or a universal limit. A group may never reach a permitted maximum during the recorded period. Conversely, a capacity statement does not prove that the full amount was used. Maximum observed and maximum allowed should remain distinct in both tables and prose.

Natural phrases include the maximum recorded value, the permitted maximum and no more than. Kai Kai adds observed when the figure comes from a sample. For practice, compare a room with capacity for thirty and a largest recorded attendance of twenty-four. Explain why neither number should replace the other in an accurate description.

56. Range

The range, as a numerical measure of spread here, is the maximum minus the minimum. For waits from two to ten minutes, the range is eight minutes. The expression from two to ten reports the endpoints; eight reports their difference. Both are useful, but the question may request one specifically.

Range uses only the two extremes. It does not reveal how the remaining observations are arranged. Lists with the same minimum and maximum can have very different concentrations in the middle. Also, a wide range does not prove an error. It may describe genuine variation that a single average would hide.

Use a range of eight minutes, values ranging from two to ten and the observed range. Alicia distinguishes endpoints from spread. For practice, create two five-value lists with the same minimum and maximum but different medians. Explain why equal ranges cannot establish equal typical values or identical experiences among the observations.

57. Distribution

A distribution describes how observations are arranged across values or categories, including their frequencies. The five waits have four observations at two or three minutes and one at ten. That description reveals more than the mean alone. It identifies the concentration of short waits and the separate larger value.

Distribution in statistics need not mean geographical distribution. The horizontal arrangement may be a number line rather than a map. A distribution can be described by centre, spread and shape, but a small set should not be assigned an elaborate theoretical model merely because a familiar label sounds appropriate. Start with the values actually supplied.

Natural phrases include the distribution of waiting times, a concentration of values and the full observed distribution. Tricia describes the pattern before choosing a summary. For practice, compare the variable list with five equal waits of four. Both means are four, but their distributions differ in centre by median, spread and frequency pattern.

58. Variability

Variability is the extent to which values differ. Five waits of four minutes show no variation in the recorded values, whereas two, two, three, three and ten do vary. The comparison concerns these supplied observations. It does not establish that one real service is always perfectly predictable or that another will always vary in the same way.

Variability is not automatically measurement error. Different people can genuinely wait different amounts of time. Repeated measurements of one unchanged quantity can also vary because of the measurement process. Identify which situation the data describe. The same word variation can refer to real differences, recording differences or a mixture of both.

Use greater observed variability, variation among the values and a measure of variability. Kai Kai avoids interpreting every difference as a mistake. For practice, explain what extra information would help distinguish changing waiting times from inconsistent timing methods. A spread calculation describes the values but does not by itself diagnose why they differ.

59. Outlier

An outlier is an observation unusually far from the main pattern under an appropriate definition or investigation. In the small waiting-time example, ten is conspicuously larger than the values two and three. It is reasonable to investigate it as a possible outlying observation. A formal decision rule, when required, should be stated rather than invented after seeing the result.

An unusual value is not automatically incorrect. It may represent an important genuine case. Removing it solely because it raises the mean would change the question and could conceal the variation the reader needs to understand. Check the original record and explain any exclusion. A retained valid outlier can be discussed alongside the median and the other observations.

Natural phrases include a possible outlier, an unusual recorded value and investigate rather than automatically remove. Alicia checks the ten-minute record before editing the dataset. For practice, compare two scenarios: the source confirms ten, or the source shows it was mistyped for one. Explain why the appropriate repair differs.

60. Weighted mean

A weighted mean gives values different contributions according to stated weights. To combine mean results from groups of unequal sizes, each group’s contribution must reflect its number of observations when the goal is the mean across all observations. A twenty-person group and a forty-person group should not receive equal weight for that particular question.

Suppose the smaller group has a 60% borrower share and the larger group a 30% share. They contribute twelve borrowers each, so the combined share is twenty-four out of sixty, or 40%. The simple mean of the two percentages is 45%, which describes an equally weighted average of the group percentages, not the borrower share among all members.

Use weighted by group size, the combined mean and the weighting rule. Tricia explains what receives equal importance: each group or each person. For practice, change the groups to equal sizes while retaining their shares. Explain why the simple average then agrees with the combined-person proportion under compatible definitions and non-overlapping membership.

Entries 61–70: describe the sequence, not merely its endpoints

A sequence can finish above its starting value without increasing at every step. A running total can rise while the newest weekly contribution falls. Read the observations in their meaningful order and decide whether the sentence describes individual periods, a cumulative amount or a difference between endpoints. The ONS line-chart guidance provides a reference for showing time sequences and missing observations; the numerical sequences here are original classroom examples.

61. Trend

A trend is a general direction or pattern of change over a specified sequence. A series may show an overall upward tendency while containing individual falls. The description needs a time window and a named variable. “Weekly loans generally increased across the term” concerns a different sequence from “the cumulative loan total increased”, even when both refer to the same records.

Two endpoints establish their difference but provide little information about the path between them. A short series also does not establish a permanent future trend. Avoid calling every final increase sustained growth. State what the observations show and preserve consequential reversals. A prediction extending that pattern belongs to a separate claim with additional assumptions.

Use the observed trend over these weeks, an overall upward pattern and the trend is interrupted by. Alicia reads all the values before writing the summary. For practice, compare 18, 24, 12 and 30 with 18, 22, 26 and 30. Both share endpoints; only the second increases at every recorded step.

62. Increase

An increase is a movement to a greater value. A count that moves from eighteen to twenty-four increases by six and increases to twenty-four. The words by and to identify different quantities. The first gives the change; the second gives the ending value. A sentence that loses this distinction can report a correct number in the wrong role.

Increasing is not automatically improving. An increase in completed tasks may be welcome, while an increase in errors may not be. The evaluation depends on the purpose and other relevant measures. A larger count also need not mean a larger share when its denominator changes. Describe the numerical relationship before adding a judgement about whether it is desirable.

Natural phrases include an increase of six loans, increased from eighteen to twenty-four and increased by one third. Tricia identifies the starting value for the relative comparison. For practice, write three versions of the same change: its endpoints, its absolute difference and its percentage increase. Keep the unit and period constant across all three.

63. Decrease

A decrease is a movement to a smaller value. A weekly count falling from twenty-four to twelve decreases by twelve, or by half of its starting value. It ends at twelve, not at zero. The decrease and the remaining amount happen to be equal in this example, but they are different quantities and need not be equal elsewhere.

A falling weekly count does not necessarily make the accumulated total fall. Twelve additional completed loans still increase a running total when no earlier transactions are removed or reclassified. Likewise, a falling percentage does not necessarily mean fewer participating people. The writer must state which measure decreased and avoid extending the finding to another measure without calculation.

Use decreased to twelve, a decrease of fifty per cent and a smaller weekly contribution. Kai Kai names the interval being compared. For practice, explain a series in which the weekly count falls but the cumulative count rises. Both can be accurate because one describes the latest period and the other includes earlier periods as well.

64. Fluctuate

Fluctuate means to vary up and down rather than remain constant or move steadily in one direction. The counts eighteen, twenty-four, twelve and thirty fluctuate across four equal weeks. The ending count is greater than the starting count, but the sequence includes a substantial fall between the second and third observations.

Fluctuation does not by itself establish a cause or a statistical model. It does not necessarily mean that the process is random, unreliable or failing. Different weeks may have different conditions, or the variation may have another explanation. The numbers identify the changing pattern; further information is required to explain why it occurred.

Natural phrases include fluctuated between twelve and thirty, fluctuating weekly counts and rose and fell during the period. Alicia includes the direction changes in her summary. For practice, revise “The count increased steadily from eighteen to thirty” using all four observations. Preserve the net increase without deleting the intervening decline that makes steadily inaccurate.

65. Remain constant

To remain constant is to keep the same value over the specified observations or interval. Four recorded counts of six show a constant recorded count. That does not necessarily mean the same six people or objects are present each time. A stock can remain six while different items enter and leave in equal numbers.

Equal sampled readings do not establish that the value never changed between readings. If the register is checked only at the end of each day, unrecorded daytime variation may remain unknown. State whether the claim concerns recorded endpoints or continuously known conditions. A constant rounded value can also conceal small changes below the reporting resolution.

Use remained constant in the recorded observations, an unchanged reported value and constant at six. Tricia distinguishes stability of count from stability of identity. For practice, describe a shelf containing six folders each evening when two leave and two different folders arrive daily. Explain what remains constant and what does not.

66. Plateau

A plateau is a period in which a previously changing quantity levels off or varies only slightly around a level. A sequence of four, seven, ten, ten and ten can be described as reaching a plateau in its last three observations. The word identifies a local pattern rather than an explanation of why growth stopped.

A plateau does not prove that a maximum possible capacity has been reached. The observed count might be limited by demand, recording rules or another unmeasured condition. It also does not promise that future values will remain at that level. Keep the observation window attached, especially when a short flat section appears within a longer changing record.

Natural phrases include plateaued at ten during the final observations, a temporary plateau and the count levelled off. Kai Kai avoids replacing plateau with permanent limit. For practice, add a later value of fourteen to the sequence. Explain why the earlier plateau remains an accurate description of that interval but not of the entire extended record.

67. Peak

A peak is a high point in a sequence relative to the surrounding values or the whole stated period. In eighteen, twenty-four, twelve and thirty, twenty-four is a local high before a fall, while thirty is the highest recorded value of the four. The intended comparison should be clear when more than one high point appears.

A final high observation is not proof that the actual process has reached its ultimate peak. Later values may be higher, and unobserved intervals may contain values outside the recorded range. “The highest recorded weekly count was thirty” is often more exact than “The programme reached its maximum possible demand.” The latter adds a capacity claim the sequence does not support.

Use the highest recorded peak, a local peak in Week 2 and peaked within the observed period. Alicia specifies the time window. For practice, compare the largest plotted value with a stated service capacity. Explain why an observed peak and a permitted maximum can differ without either number being mistaken.

68. Trough

A trough is a low point in a sequence relative to surrounding values or the specified period. Twelve is a trough between twenty-four and thirty in the four-week example. It describes the position of that observation in the pattern. It does not necessarily describe a failure, a closure or the smallest value that could ever occur.

A missing observation is not a trough. A graph that drops to zero where data are unavailable creates a false low point. A recorded zero, by contrast, can be a genuine observation under the counting rule. The distinction between an absent value and a low value is essential before interpreting the shape of the sequence.

Natural phrases include a trough in Week 3, the lowest recorded point and followed by a recovery in the count. Tricia checks missing-value notes before describing the pattern. For practice, explain why replacing an unknown Wednesday with zero would create a visual claim not present in the source records.

69. Cumulative

Cumulative means built up by adding successive contributions from a stated starting point. Weekly counts of eighteen, twenty-four, twelve and thirty produce cumulative totals of eighteen, forty-two, fifty-four and eighty-four. Each running total includes previous weeks; it is not another separate week’s count to add to the weekly figures.

When the contributions are nonnegative and no corrections remove earlier entries, a cumulative count cannot fall. It can rise more slowly when a weekly contribution is smaller or remain unchanged when the new contribution is zero. An apparently rising cumulative line therefore does not establish increasing weekly demand. Check whether the display shows a stock accumulated through time or separate period amounts.

Use the cumulative total through Week 4, a running count and the latest contribution. Kai Kai obtains a week’s count by comparing successive cumulative values. For practice, explain why fifty-four minus forty-two gives twelve for Week 3 and why fifty-four itself cannot be called the number completed during that week.

70. Net change

Net change is the ending value minus the starting value over a stated interval. A weekly count beginning at eighteen and ending at thirty has a net increase of twelve across those endpoints. This differs from the total number of loans across all four weeks, which is eighty-four. Both calculations use the same series but answer different questions.

Net change can conceal reversals. Adding the successive changes of plus six, minus twelve and plus eighteen gives the same net plus twelve, yet those steps reveal a less steady pattern. A zero net change likewise does not mean nothing happened between endpoints. The count may rise and fall before returning to its starting value.

Natural phrases include a net increase of twelve, no net change between the endpoints and the intervening changes. Alicia reports both the endpoint difference and an important reversal. For practice, create a three-value series with zero net change but a substantial middle increase. Explain why unchanged overall is not the same as constant throughout.

Entries 71–80: select a claim whose size matches its support

Comparison words carry commitments. Majority means more than half of a named whole. Exceed means greater than, not equal to. Approximately signals limited numerical precision, while estimate identifies a value obtained under a method rather than an exact supplied observation. The ordinary words around a statistic often decide whether the final sentence accurately represents it.

71. Compare

Compare means to examine things in relation to each other using relevant dimensions. Comparing two clubs can involve their loan totals, borrower shares or transactions per opening hour. Those comparisons can rank the clubs differently. A useful response states the dimension before declaring one club larger, busier or more successful.

A fair comparison requires compatible definitions, periods and units. A full-month total should not be treated as directly equivalent to a one-week total when the question concerns average activity. Dividing by time may help for a particular question, but it does not guarantee that the periods have otherwise comparable conditions. State what the adjustment achieves and what it leaves unresolved.

Use compare the same measure, compared over equal periods and a comparison limited to these records. Tricia checks the metadata before subtracting. For practice, write a comparison in which Moon has more transactions but Sun has a higher transaction rate. Name both denominators so the apparent disagreement becomes understandable.

72. Contrast

Contrast means to draw attention to a relevant difference. “The two sets have the same mean; in contrast, their ranges differ” directs the reader from a shared feature to a distinction. A contrast should connect comparable quantities rather than merely place unrelated facts in neighbouring sentences.

Contrasting data does not require treating groups as complete opposites. Two clubs can differ in loan totals while sharing the same distinct-borrower count. Preserve the similarities that matter. Likewise, a contrast in observed outcomes does not by itself identify a cause, motive or general quality of either group.

Natural phrases include in contrast to, contrasts with and a contrast in observed spread. Kai Kai names the differing measure rather than calling one group simply better. For practice, combine a similarity sentence and a contrast sentence about the waiting-time lists. Explain why that pair is more informative than an unqualified statement that the groups are identical.

73. Exceed

Exceed means to be greater than a stated value or limit. A count of thirty exceeds twenty-four by six. A count of twenty-four does not exceed a threshold of twenty-four; it equals it. This boundary distinction matters in questions about qualification, capacity or the number of observations above a cutoff.

Exceeds five and at least five select different cases because the second includes exactly five. A learner should test the boundary value before counting. The phrase exceeds by also differs from exceeds to: write the difference after by, and use rises to when the intended number is the ending level.

Use exceeds the limit, exceeds the earlier count by six and does not exceed. Alicia checks whether equality qualifies under the actual wording. For practice, classify values of four, five and six under exceeds five, at least five and no more than five. Explain exactly which cases change across those instructions.

74. Majority

A majority is more than half of the specified whole. Six of ten respondents form a majority of those respondents. Five of ten are exactly half, not a majority. The denominator must remain explicit because a majority of respondents may be far fewer than half of all eligible people.

The largest category need not have a majority. With preferences split four, three and three among ten people, the four-person category is the largest single group but remains below half. Calling it a majority changes the finding. A writer can say the most frequently selected category without implying that most people selected it.

Natural phrases include a majority of the respondents, more than half and the largest category, but not a majority. Tricia checks the whole before using most. For practice, explain why eight favourable answers from ten respondents do not establish a favourable majority among sixty eligible members when the other fifty answers are unknown.

75. Minority

A minority, in the numerical comparison used here, is less than half of a stated whole. Four of ten responses are a minority of the responses even when four is the largest of several category counts. The word describes a share, not whether the view is important or whether its holders deserve attention.

Minority has social and political meanings beyond this elementary numerical use. Do not assign a demographic label from a classroom fraction. Also avoid using a minority as a substitute for a useful exact count when the numbers are available. Four of ten communicates more about the size than the broad category alone.

Use a numerical minority of respondents, less than half and the smaller share in this comparison. Kai Kai reports the count without dismissing its significance. For practice, describe a four-person preference group that is the largest category among ten. Explain how it can be both the largest category and a minority of the whole.

76. Respectively

Respectively matches items in the corresponding order just stated. “Sun and Moon recorded twenty-four and thirty loans, respectively” assigns twenty-four to Sun and thirty to Moon. Reversing the club names without reversing the numbers changes the reported facts. The word can save repetition when the paired lists remain easy to follow.

Respectively differs from respectfully, which describes a manner of showing respect. It also should not make a sentence unnecessarily difficult. When several groups, percentages and periods compete for correspondence, separate sentences may be clearer. A grammatically correct compressed sentence can still impose too much matching work on the reader.

Use twenty and forty members, respectively after two clearly ordered group names. Alicia expands the sentence into two statements to check it. For practice, reverse the order of Sun and Moon, reconstruct the sentence and verify both assignments. Then add a third club and decide whether respectively still improves clarity or whether another structure would be easier to read.

77. Approximately

Approximately means close to a stated value without claiming exact equality. One third is approximately 33.3% when rounded to one decimal place. The underlying fraction can be exact while the displayed decimal is approximate. The qualification concerns the expression’s precision, not necessarily uncertainty about the original count.

Do not add approximately to every numerical answer automatically. Twelve out of twenty is exactly 60% under exact supplied counts. Conversely, an estimated input does not become exact because a calculator displays many decimal places. Decide whether approximation comes from measurement, sampling, rounding or another specified limitation.

Natural phrases include approximately one third, about 33.3% and rounded to one decimal place. Tricia states the reason for approximation where it matters. For practice, compare an exact classroom fraction with a ruler estimate. Explain why their approximate-looking outputs can have different kinds of limitation despite using the same word about.

78. Estimate

An estimate is a value obtained to approximate a quantity using available information and a stated method. A reader might estimate a plotted value between labelled gridlines, or an analyst might estimate a population proportion from a sample. These are different estimation tasks. The method and its assumptions determine what the result can reasonably support.

An estimate is not an arbitrary guess, but neither is it automatically a verified observation. A midpoint calculation for a missing day can be mathematically exact under a chosen interpolation rule while the day’s actual count remains unknown. The label estimated must stay attached when the figure moves into a summary or a table.

Use an estimated value, estimated from the available observations and the estimation method. Kai Kai separates a measured value from a model-based substitute. For practice, explain why a clearly labelled estimate can be useful while silently replacing missing data with that same number would misrepresent what was actually recorded.

79. Interpolate

To interpolate is to estimate a value between known positions or observations using an assumed relationship. Under a straight-line interpolation between Tuesday’s four and Thursday’s zero, Wednesday’s midpoint estimate would be two. The calculation follows the chosen linear rule. It does not establish that anyone observed two on Wednesday.

Interpolation may be unsuitable when the process changes abruptly or does not follow the assumed form. A missing daily count could differ greatly from the midpoint even when both neighbouring counts are correct. Read the purpose before filling gaps. A display should not make estimated intermediate values look identical to directly recorded observations without explanation.

Natural phrases include interpolate under a linear assumption, an interpolated value and estimated between the known points. Alicia names the assumption beside the number. For practice, provide two possible Wednesday counts consistent with the same Tuesday and Thursday values. Their coexistence shows why those endpoints alone cannot identify the missing observation exactly.

80. Extrapolate

To extrapolate is to extend an estimated relationship beyond the observed range. A writer who sees weekly counts of ten, twelve and fourteen might predict sixteen for the next week under a continuing two-per-week increase. Sixteen is then an extrapolated prediction, not another observation already contained in the record.

A pattern can stop outside the observed interval. Capacity, demand, rules or other conditions may change. Even a useful prediction needs its assumptions and status stated clearly. Do not describe a forecast as a confirmed result merely because the arithmetic producing it is simple. The uncertainty concerns whether the relationship continues, not whether fourteen plus two equals sixteen.

Use extrapolate beyond the observed weeks, a conditional forecast and assuming the same pattern continues. Tricia keeps will become separate from was recorded. For practice, compare an interpolation inside the observation window with an extrapolation outside it. Identify the assumption each introduces and the information needed to check its resulting prediction later.

Entries 81–90: find the error without inventing the missing fact

The ABS discussion of types of error distinguishes uncertainties arising from sampling from errors in collection and processing. A wrong transcription, incomplete coverage and a misleading interpretation are not the same problem. The words below help identify which part of a claim needs repair. Accurate criticism should be as carefully bounded as the original finding.

81. Missing

Missing describes a required value that is unavailable in the record being used. A lost Wednesday register leaves Wednesday’s count missing. It does not establish that the count was zero or that the activity stopped. A dash can indicate missing information only when the table’s convention says so; another table might use that mark differently.

Different reasons for missingness can matter to interpretation. A skipped question, an ineligible item and a lost recording are not automatically equivalent. Preserve any distinction the method note supplies. Where the reason is unknown, say that it is unknown rather than assigning a convenient explanation. Filling a gap with an estimate also needs an explicit label and method.

Use a missing observation, the value is unavailable and missing rather than recorded as zero. Alicia keeps the gap visible in the record. For practice, state what can still be calculated from four known daily counts and what remains unresolved about the whole five-day period. Missing one value does not make every known value unusable.

82. Zero

Zero is a numerical value, not an automatic substitute for an empty cell. In a complete count of eligible loans during an open session, zero means no qualifying loans were recorded as occurring under that count. It does not mean nobody entered the room or no other activity happened. The category being measured still matters.

A zero may also be the result of rounding a small value, rather than an exact absence, when the source states that convention. Read the reporting rule. Likewise, zero change means equal endpoint values; it does not mean the quantity itself is zero. A stock of six that stays six has zero net change and a nonzero ending stock.

Natural phrases include a recorded zero, zero eligible transactions and zero net change. Tricia names what is absent or unchanged. For practice, compare a recorded zero for Thursday, an unknown Wednesday and a five-day net change of zero. Explain why these three statements carry different information despite their superficial association with nothing.

83. Duplicate

A duplicate is an additional copy or repeated entry of something already represented. In a one-response-per-person survey, two confirmed copies of the same person’s same submitted answer should not be counted as two respondents. Their duplicate status must be established through the record, not guessed merely because the answers match.

Repeated values are not necessarily duplicate observations. Two different pupils can both wait three minutes. One pupil can make two genuine loans on different occasions. Removing one such record because its value or name repeats would erase a valid event. The question is whether the same unit and occasion were entered twice under the counting rule.

Use a confirmed duplicate record, count each respondent once and repeated values from distinct observations. Kai Kai checks identifiers before removing a row. For practice, compare two matching timestamps with the same submission identifier against two different loan identifiers belonging to one borrower. Explain why the first may be a duplicate while the second can represent separate valid transactions.

84. Overlap

An overlap occurs when groups share members or records. Four pupils choosing art and three choosing music can involve only five distinct pupils when two choose both. Adding four and three counts selections across categories, not necessarily distinct people. The shared membership must be accounted for when the question asks how many people chose at least one activity.

Overlap is not always an error. A multiple-response question can deliberately allow several choices. Its category percentages may then total more than one hundred because the same person contributes to more than one numerator. A single-choice pie chart needs mutually exclusive parts of one whole, so the representation must match the question’s structure.

Natural phrases include overlapping categories, the two groups share two members and distinct people rather than selections. Alicia separates art only, music only and both. For practice, explain why the distinct total is four plus three minus two. The subtraction is justified by confirmed shared identities, not by a rule that every pair of categories must overlap.

85. Rounding

Rounding expresses a value at a chosen level of numerical detail according to a rule. One third of a group can be reported as approximately 33.3% to one decimal place. The exact fraction has not changed; the displayed value is a shorter approximation. The intended precision should be stated where it affects the answer.

Separately rounded category percentages may not add to exactly one hundred. Three equal shares rounded to 33.3% total 99.9%, although the exact thirds form one whole. Do not invent an extra category to fill that rounding gap. Similarly, calculating from already rounded intermediate values can produce a different result from rounding once after an exact calculation.

Use rounded to one decimal place, rounding explains the small discrepancy and retain the unrounded values for calculation. Tricia keeps the original fractions in her working. For practice, compare the sum of three rounded thirds with the rounded value of their exact sum. Explain why the two displayed totals can differ without missing a respondent.

86. Precision

Precision can concern how exactly a result is expressed or how closely repeated measurements agree, depending on context. In reporting a percentage, specifying one decimal place identifies a level of displayed detail. In repeated timing, closely grouped readings describe repeatability. The sentence should make clear which meaning is intended.

More displayed digits do not automatically mean better evidence. A duration measured only to the nearest second cannot acquire genuine millisecond detail through a longer written number. Closely repeated measurements can also share the same systematic offset. Their agreement therefore does not by itself establish accuracy against the correct reference value.

Natural phrases include reported precision, precise about the known boundary and closely repeated readings. Kai Kai distinguishes a detailed-looking answer from a justified one. For practice, explain why a bounded statement such as between twelve and twenty-two transactions can be precise about what is known even though it does not supply one exact total.

87. Accuracy

Accuracy concerns whether a value or statement correctly represents the relevant fact or quantity. A correctly copied transaction count can be numerically accurate while an accompanying sentence wrongly calls it a borrower count. In that case, the digits are preserved but the description is inaccurate. Checking the meaning is part of checking the result.

Accuracy needs a relevant reference. Comparing a transcription with its source checks whether it was copied correctly, not whether the original observation was itself correct. Repeating the same mistaken record on another page does not create independent confirmation. State what was checked and avoid converting a limited verification into a guarantee about the entire investigation.

Use accuracy of the transcription, an accurate description of the supplied data and checked against the original record. Alicia verifies both the number and its column label. For practice, describe a table whose values are correct but whose heading gives the wrong year. Identify the correction needed without altering the values that the evidence already supports.

88. Bias

Bias, in this data-reading context, is a systematic tendency that can make a result differ from the quantity or population the writer intends to describe. Asking only early-session attendees about preferred opening times may leave out people whose schedules prevent early attendance. The selection procedure, not simply the number of answers, needs attention.

Bias need not be deliberate. It can arise from coverage, question wording, measurement or processing choices. Nor does finding a possible source of bias establish its exact size or direction in every case. A careful critique identifies the mechanism of concern and the conclusion it limits without inventing the unobserved answers.

Natural phrases include possible selection bias, a systematically excluded group and the procedure may bias this comparison. Tricia distinguishes an unrepresentative method from an accusation about someone’s motives. For practice, improve the sampling plan for a scheduling question without predicting how the newly included people will answer. Better coverage is not permission to supply their preferences in advance.

89. Association

An association is a relationship observed between variables. A fictional dataset might show that sessions with longer opening hours also recorded more loans. This identifies a pattern across the observations. It does not by itself establish how much changing hours would cause loans to change, because the sessions may differ in other relevant ways.

The ABS guide to correlation and causation explains why a relationship alone does not settle causality. A small visual pattern also should not automatically be described as statistically strong without an appropriate analysis. Begin with the comparison the record actually supports and keep causal explanations separate.

Use an observed association, the variables vary together in this dataset and associated with rather than shown to cause. Kai Kai asks what other conditions differ across sessions. For practice, write a descriptive relationship and then a question for further investigation. Do not turn the question’s proposed explanation into an established answer.

90. Causation

Causation concerns whether one factor produces or contributes to a change in another. A causal claim goes beyond saying that two events occurred together or in sequence. A new poster appearing before an increase in registrations supplies timing information; the poster’s actual effect needs a justified comparison or other appropriate causal evidence.

Failure to establish causation does not prove that no causal effect exists. It means the supplied evidence is insufficient for that stronger conclusion. Nor should a learner memorise that only one research design can ever support causal reasoning. For these short classroom packets, the practical rule is to identify what was controlled, compared or otherwise established before using caused, produced or led to.

Natural phrases include causation is not established by this record alone, a proposed causal explanation and contributed to the change. Alicia retains a before-and-after finding without attributing it automatically to the poster. For practice, name one alternative explanation the packet leaves open, clearly labelling it as a possibility rather than another reported event.

Entries 91–100: make the answer accountable to its evidence

Reading a number, calculating from it and defending a conclusion are different actions. The final group gives names to those jobs. A strong answer can be brief, but it should let another reader identify the population, quantity, comparison and necessary limit. The following original writing exercises focus on those relationships rather than rewarding a quota of technical expressions.

91. Identify

Identify means to recognise and name the particular feature relevant to a question. A learner can identify the largest recorded weekly count, the correct denominator or the row containing a club’s borrower total. This is often the first step in a longer answer. It does not automatically explain why the feature occurs.

Identification should remain within the task’s scope. Finding the largest observed value does not identify the system’s maximum possible output. Finding a missing cell does not identify the cause of its absence. A correct label can be useful without carrying every explanation that a reader might attach to it.

Use identify the relevant column, identify the highest observed value and the identified comparison. Tricia names the feature before interpreting it. For practice, divide an answer into its identification and explanation. Where the question asks for both, check that merely repeating the label has not replaced the reasoning the second part requires.

92. Describe

Describe means to give an account of relevant features. A description of weekly counts might state that they rose, fell and then reached their highest recorded value. Include values or intervals that make the pattern clear. Listing every number without explaining their relationships may leave the reader to perform the description’s main work.

Description does not require inventing a cause. “The count fell in Week 3” reports a pattern. “The count fell because pupils disliked the activity” adds an explanation that the numbers alone do not establish. A writer can describe an important finding confidently while leaving its cause open for further investigation.

Natural phrases include describe the observed pattern, a description of the distribution and the main feature and its exception. Kai Kai keeps evaluative adjectives separate from measurements. For practice, write a two-sentence description of eighteen, twenty-four, twelve and thirty. Include the overall endpoint change and the intervening reversal without claiming a future result.

93. Interpret

Interpret means to explain what information means in its context. Reading a borrower share of 60% as twelve out of twenty eligible members connects the percentage with its actual whole. Interpreting the same number as sixty people would fail when the group contains only twenty. The context is part of the numerical meaning.

Interpretation should preserve the source’s definitions. A display of interest in an activity does not necessarily measure attendance. An average transaction rate does not establish an identical number every hour. These limits do not make interpretation impossible; they identify the particular meaning the result can carry without importing a different variable.

Use interpret the percentage as, the result means that and an interpretation limited to the respondents. Alicia paraphrases the statistic in ordinary language. For practice, ask a partner to reconstruct the fraction from the explanation. If the reconstructed numerator or denominator differs, identify which word in the sentence allowed the meaning to drift.

94. Infer

To infer is to reach a conclusion from information rather than simply repeat a directly stated sentence. If a complete table lists sixty eligible members and twenty-four distinct borrowers, the remaining thirty-six did not borrow during that period under the stated rule. That conclusion follows by subtraction even if the table has no non-borrower column.

An inference depends on its premises. The subtraction needs complete coverage and a borrower definition restricted to the stated membership. A table of only some members would not support the same whole-group conclusion. Inference is not a synonym for guessing what makes a good story; it is reasoning whose connecting assumptions can be examined.

Natural phrases include infer from the complete record, a supported inference and cannot be inferred from these values alone. Tricia states the completeness condition. For practice, remove that condition from the packet and explain which conclusion becomes less secure. Keep the arithmetic correct while revising the scope of the statement it can justify.

95. Substantiate

Substantiate means to support a claim with adequate relevant evidence. Complete transaction records can substantiate a stated loan total for their period. They do not automatically substantiate a claim about enjoyment, reading comprehension or the reason people borrowed. The source must measure or otherwise establish what the claim actually concerns.

More numbers are not always stronger support. Several tables copied from the same original record may repeat one source rather than provide independent observations. A large count of irrelevant events does not repair a missing denominator. Identify what each piece of evidence contributes before deciding that a statement is well supported.

Use substantiates the narrow claim, the claim remains unsubstantiated and evidence sufficient for this conclusion. Kai Kai connects a sentence to the particular cells that support it. For practice, write the strongest claim a small complete packet supports and a tempting stronger claim it does not. Explain the additional evidence required by the stronger version.

96. Qualify

Qualify means to limit or modify a statement so that it fits its evidence. “Eighty per cent supported the proposal” becomes clearer as “Eight of the ten respondents supported the proposal.” Adding among respondents prevents the result from being mistaken for the known views of all sixty eligible members.

A useful qualification names the limit rather than sprinkling perhaps over everything. The eight-out-of-ten calculation is exact within the supplied responses. The uncertain part concerns extending it to missing answers. Do not weaken a secure calculation merely because a different, broader conclusion remains unsupported.

Natural phrases include a qualified conclusion, qualify the population claim and the statement applies to. Alicia changes the unsupported phrase while retaining the finding. For practice, repair a headline by adding the smallest necessary wording. Explain why that wording changes the meaning rather than merely making the writer sound cautious.

97. Synthesise

Synthesise means to combine relevant information into a connected account. A borrower comparison can bring together membership totals, distinct borrowers, transaction counts and opening hours. Each contributes a different dimension. A synthesis explains how the findings fit rather than producing four separate summaries that never address the same question.

Synthesis should not force all measures to tell one uniformly positive or negative story. A club can have more transactions and a lower transaction rate. Preserve that distinction. If two records use incompatible periods or definitions, their information may need to remain separate until the difference is resolved. Fluent merging is not a substitute for compatibility.

Use synthesise the compatible records, a synthesis of count and rate and the findings together show. American English commonly uses synthesize. Tricia organises her paragraph by the comparison rather than by file name. For practice, connect two true findings using whereas, and add one sentence explaining the denominator that makes them compatible.

98. Verify

Verify means to check a detail or result against an appropriate reference or method. Recalculating twenty-four divided by sixty checks the pooled proportion’s arithmetic. Comparing the twenty-four borrower count with a deduplicated membership record checks a different part of the claim. Verification should identify which layer was examined.

A calculation can be correct on wrongly selected inputs. A copied table can match its source even when the original collection has limitations. Do not treat one successful check as universal verification. State what remains unchecked, such as current conditions, individual identities or the cause of a change.

Natural phrases include verify the calculation, verified against the original record and the conclusion needs a separate check. Kai Kai checks row, column and unit before pressing the calculator. For practice, match three possible errors with three checks: wrong group label, arithmetic mistake and outdated survey date. Explain why repeatedly checking only the arithmetic could miss the other two.

99. Revise

Revise means to change a text, calculation or plan after reconsidering it. A learner revises a caption when a missing denominator makes its percentage misleading. The best revision targets the problem: replace an unsupported population claim, correct the selected value or clarify the unit. Rewriting everything is not necessary when one phrase carries the error.

Revision differs from changing a result to suit a preference. The evidence should determine the correction. Keep a record of why an entry was removed or a number changed, especially when a confirmed duplicate or source correction is involved. A revised text should not conceal the distinction between a new observation and a repair to an earlier transcription.

Use revise the claim to match the data, a documented revision and the corrected version. Alicia keeps the first caption beside the repair for learning. For practice, explain what changed in the meaning, why the source required it and which other conclusions remain unchanged. The explanation of the revision tests understanding beyond the final wording.

100. Communicate

Communicate means to convey information so another person can understand the intended meaning. A statistical answer communicates well when its reader can tell what was counted, who was included, which comparison was made and what the evidence leaves open. Correct digits alone may fail that task if their labels or qualifications are missing.

Communication also requires selection. Listing every cell can bury the central finding, while removing every limit can distort it. Choose the information that answers the reader’s question and retain the words that determine its scope. An ordinary sentence about twelve borrowers out of twenty may communicate more successfully than a crowded sentence full of technical terms.

Use communicate the main comparison, communicates the limit clearly and an understandable account of the data. Tricia asks a partner to paraphrase the conclusion without hints. For the final entry’s practice, identify any difference between the intended claim and the partner’s reconstruction. Repair the missing reference or category rather than assuming that a longer answer will automatically be clearer.

Six original data-reading laboratories with worked answers

All the following records, organisations, timings and statements are fictional teaching material. Each packet supplies the information needed for its questions. A display described in words does not require an unseen image; an optional sketch should represent the stated rules rather than introduce new measurements. Treat the supplied complete counts as exact within the exercise, while preserving any missing information explicitly identified.

Choose A: borrowers and transactions, B: a chart whose bars appear to double, C: equal means and different waits, D: a rising count with a falling share, E: missing observations and cumulative totals, or F: respondents, duplicates and overlapping choices. There are five main questions in each laboratory, followed by explained responses and a changed-condition transfer task.

Attempt each question before consulting its model. An acceptable answer need not copy the model’s sentences, but its figures, conditions and interpretation must remain consistent with the packet. Start by naming what the question asks you to compare. Then locate the relevant records, calculate only where needed and write a sentence whose noun matches the quantity. Finish by checking any claim about all people, future periods or causes against the evidence actually supplied.

Laboratory A: Fifty-four loans do not mean fifty-four borrowers

A1 — the population and period. A fictional lending project keeps complete records for Sun Club and Moon Club during the same four-week period. Sun has twenty registered members and Moon has forty. No person belongs to both clubs. These memberships remain unchanged throughout the period. All recorded loans belong to those registered members; there are no visitor loans in this packet.

A2 — three different counts. A distinct borrower is a member who completes at least one qualifying loan during the period. Each borrower is counted once in that column, regardless of how many loans they complete. A loan transaction is one completed lending event with a unique transaction identifier. The records contain no duplicated transactions. The total membership includes members who make no loans.

A3 — the complete data. Sun Club records twenty members, twelve distinct borrowers, twenty-four loan transactions and ten total opening hours. Moon Club records forty members, twelve distinct borrowers, thirty loan transactions and fifteen total opening hours. Opening hours are the hours when each club’s lending desk is available during the four-week period. The packet supplies no hour-by-hour distribution of transactions.

A4 — a draft interpretation. Kai Kai writes, “Moon helped more children because it completed more loans. Together, the clubs served fifty-four borrowers. Their average borrowing share was 45%, so 45% of all sixty members borrowed.” Alicia separates the people from the transactions. Tricia checks whether each club or each member should receive equal weight in the combined share.

A5 — the limit of the record. The figures describe the defined lending activity only. They do not measure reading enjoyment, pages read, learning improvement or the reasons a member borrowed. They also do not say that a transaction took a fixed amount of time. An average transaction rate is a count divided by total opening hours, not a claim that loans occurred evenly.

Questions A

AQ1. Which club has more loan transactions, and which has more distinct borrowers? Explain why choosing the transaction figure for a borrower question changes the meaning.

AQ2. Calculate the combined membership, distinct-borrower count and transaction total. State the assumptions that permit the two club counts to be added as distinct people.

AQ3. Calculate each club’s borrower share and the share across all members. Explain why 45% is not the combined-person result.

AQ4. Calculate transactions per opening hour and transactions per distinct borrower for each club. Does either average establish that every hour or every borrower had the average value?

AQ5. Write a corrected four-sentence account of the comparison. Retain a useful finding rather than replacing the draft with the vague statement that more research is needed.

Worked answers A

AQ1. Moon has more transactions: thirty compared with Sun’s twenty-four. Both clubs have twelve distinct borrowers. The transaction count records completed lending events, whereas the borrower count records different people who made at least one loan. Because one person can contribute several transactions, more loans do not establish more borrowers. The phrase helped more children also introduces an evaluative claim that the packet has not defined or measured. A narrower comparison can report more transactions without assigning an unsupported educational benefit.

AQ2. There are sixty members, twenty-four distinct borrowers and fifty-four loan transactions across the clubs. The distinct-person totals can be added because memberships do not overlap, the membership lists are complete and every borrower belongs to the relevant club. Fifty-four is not the number of people: it counts the separate loan events. The thirty-six members who made no loan can be inferred by subtracting twenty-four borrowers from sixty members under the packet’s complete coverage. That inference would need revision if the lending record covered only selected sessions.

AQ3. Sun’s share is twelve out of twenty, or 60%. Moon’s is twelve out of forty, or 30%. Across all sixty members, twenty-four borrowed, giving 40%. The simple mean of 60% and 30% is 45%, but it gives the two clubs equal weight despite Moon having twice as many members. That average answers a different question about the two club percentages. For the share of all people, combine the compatible counts or weight each club’s share by its membership. The result is not chosen according to which headline sounds more favourable.

AQ4. Sun records 24/10 = 2.4 transactions per opening hour, while Moon records 30/15 = 2. Sun therefore has the higher average hourly rate despite the smaller total. Per distinct borrower, Sun records 24/12 = 2 transactions and Moon 30/12 = 2.5. These are different denominators. Neither rate gives the exact pattern for every hour or person. A 2.5 mean does not mean half a transaction occurred, and Sun’s mean of two does not establish that every borrower made exactly two loans.

AQ5. A suitable account is: “During the four-week period, Moon recorded thirty loan transactions compared with Sun’s twenty-four, but both had twelve distinct borrowers. Sun’s borrower share was higher because its twelve borrowers came from twenty members rather than forty. Across the sixty non-overlapping members, twenty-four borrowed, giving a combined share of 40%. Sun also had the higher average transaction rate per opening hour, while the records do not measure reading outcomes or explain the differences.” Each sentence uses a relevant measure and prevents transactions, people and educational effects from becoming one claim.

Transfer A. Add one confirmed, previously omitted transaction by a Sun member already included among its twelve borrowers. The corrected transaction total becomes twenty-five for Sun and fifty-five combined. The membership and distinct-borrower counts remain unchanged, as do the borrower shares. Sun’s hourly rate becomes 2.5 under the unchanged ten-hour total. Explain why recovering an omitted transaction changes some results but not every statistic in the paragraph. A targeted correction follows the identity and kind of record that changed.

Laboratory B: The taller bar and the unchanged numbers

B1 — the observed counts. A fictional exhibition receives eighty completed feedback cards during Week 1 and one hundred during Week 2. Both are complete counts of qualifying cards during equal-length observation periods. The card identifiers are distinct, and no card is counted in both weeks. The packet does not identify distinct people or supply daily counts within the weeks.

B2 — Display A. Two equal-width vertical bars are drawn on a linear quantity axis labelled sixty, seventy, eighty, ninety and one hundred. The visible bars begin at sixty and end at the values eighty and one hundred. Thus their visible heights above the drawn baseline represent differences of twenty and forty from sixty. The labels beside their tops correctly state eighty and one hundred.

B3 — Display B. A second version uses a zero baseline and the same number of printed centimetres per ten-card interval. Its bars end at the same values as before. The underlying records have not changed. The exercise describes both displays completely; the question concerns their stated encoding rules, not measurements from an unseen picture.

B4 — proposed captions. Kai Kai writes, “Feedback doubled in Week 2 because its bar is twice as tall.” Alicia writes, “The recorded count increased by twenty cards.” Tricia proposes adding a percentage comparison and replacing Display A with Display B. Nobody has recorded why the original designer selected the cropped baseline, so deliberate deception is not an established fact.

B5 — another representation. The group also considers a pictogram with one identical icon for every twenty completed cards. Each icon has the same size and represents a fixed count; the design does not use an enlarged icon whose area changes. A separate proposed line chart would plot the two weekly values in time order. No later week’s count is supplied.

Questions B

BQ1. Calculate the absolute increase, Week 2’s value as a multiple of Week 1’s, and the percentage increase.

BQ2. Explain why the visible heights in Display A have a two-to-one ratio while the actual counts do not. Does the presence of correct top labels eliminate that representational problem?

BQ3. Explain what Display B repairs and what it leaves unchanged. How many icons belong to each week in the proposed fixed-key pictogram?

BQ4. Would a line connecting the two weekly observations establish steadily increasing daily activity or guarantee another increase in Week 3? Explain the limits without declaring every line chart invalid.

BQ5. Write a corrected caption and one design note. Distinguish a criticism of the display from a claim about the designer’s intention.

Worked answers B

BQ1. The absolute increase is one hundred minus eighty, or twenty completed cards. Week 2 is 100/80 = 1.25 times Week 1. The increase relative to the starting eighty is 20/80 = 0.25, or 25%. Week 2 is 125% of Week 1’s value, not a 125% increase. It would need one hundred and sixty cards to double the original eighty. These calculations come from the labelled counts, not the visible distances above a nonzero baseline.

BQ2. In Display A, the first visible bar spans eighty minus sixty, or twenty axis units, while the second spans one hundred minus sixty, or forty. Forty is twice twenty, so the visible lengths double. They encode distance above sixty rather than the full card counts. Correct labels allow a careful reader to recover the actual values, but the lengths still invite an inappropriate ratio comparison. The problem is therefore not a mistyped count; it is a mismatch between an ordinary bar-length reading and the cropped representation.

BQ3. Display B makes lengths proportional to the actual counts by starting both bars at zero. It changes the display, not the observations, the twenty-card difference or the 25% increase. With one fixed-size icon representing twenty cards, Week 1 receives four icons and Week 2 five. That pictogram also preserves the five-to-four relationship when its key is read correctly. Enlarging one icon in both dimensions would be a different design and should not be treated as equivalent to adding one more fixed-count icon.

BQ4. The line can connect the two weekly observations in their time order, but it does not provide the missing daily values or a future observation. Daily activity might have risen and fallen within each week while producing the stated totals. A forecast for Week 3 would require an explicit assumption about continuation. A line chart may legitimately use a nonzero vertical range to show variation clearly when its axis is labelled; that does not justify interpreting cropped bar lengths as the ratio of whole counts.

BQ5. A corrected caption is: “Completed feedback cards increased from eighty in Week 1 to one hundred in Week 2, a rise of twenty cards or 25% relative to Week 1. The two weekly totals do not show the daily pattern or identify the reason for the increase.” A useful design note is: “Use a zero baseline for bars representing these absolute card counts so their lengths retain the value ratio.” This explains the visual problem without claiming that the designer intentionally tried to mislead readers.

Transfer B. Add a complete Week 3 count of seventy. The revised sequence is eighty, one hundred and seventy. Week 2 remains the highest recorded week, but the new third observation falls by thirty cards, or 30% relative to Week 2. Across Week 1 and Week 3, the net change is minus ten, not the sum of all three counts. Explain which earlier two-week statement remains true and which broader claim of continued growth the added observation contradicts.

Laboratory C: Nobody waited for the mean

C1 — the two complete sets. In a fictional timing exercise, five distinct visitors at Desk A have waits of two, two, three, three and ten minutes. Five other distinct visitors at Desk B each wait four minutes. For this arithmetic model, the listed times are exact, use the same start-and-stop definition and include every visitor in the specified exercise. They are not estimates of a real service’s performance.

C2 — a source check. The ten-minute entry at Desk A is checked against the original timing record and confirmed as a genuine qualifying observation. There is no transcription error and no stated exclusion rule that removes long waits. The packet supplies no explanation for why that visitor waited longer. A different experience is recorded, but its cause is not established.

C3 — the proposed summary. Alicia calculates a mean of four minutes for each desk. Kai Kai writes, “The desks gave everyone exactly the same experience.” He then proposes deleting Desk A’s ten so its average will look more typical. Tricia asks for a median, modes and range before agreeing that the two distributions are adequately described by the equal means.

C4 — a grouped version. A separate display groups Desk A’s observations into three equal-width intervals: at least zero but less than five minutes; at least five but less than ten; and at least ten but less than fifteen. Each lower boundary is included and each upper boundary excluded. The grouped display gives only the frequency in each interval, not the exact values inside it.

C5 — reporting scope. The immediate task is to describe these ten observations, not to select a universally better service or predict a new visitor’s wait. No information about visitor requirements, staffing, causes or future sessions is supplied. A response should identify a genuine difference without extending the small exercise into a general institutional judgement.

Questions C

CQ1. Calculate the mean and median for each desk. Is either mean necessarily a value experienced by an individual in its dataset?

CQ2. Identify the modes, minimum, maximum and range for each set. Explain why ten is not Desk A’s mode.

CQ3. Evaluate the proposed deletion of the ten-minute value. What can the source check establish, and what does it not explain?

CQ4. Give Desk A’s frequencies in the three stated intervals. Could those grouped counts alone recover the exact original mean and median?

CQ5. Write a short comparison that reports the equal means and the different distributions. Do not claim that either dataset predicts every future visitor’s experience.

Worked answers C

CQ1. Each set totals twenty minutes across five visitors, giving a mean of four minutes. Desk A’s ordered middle value is three, so its median is three; Desk B’s median is four. Nobody at Desk A waited exactly its four-minute mean, while everybody at Desk B did. The arithmetic mean summarises the values and need not be one of them. A median also may not be an observed value in an even-sized set when the two middle observations are averaged under the stated elementary convention.

CQ2. Desk A has two modes, two and three, because each occurs twice. Its minimum is two, maximum ten and range eight minutes. Desk B has mode four, minimum four, maximum four and range zero. Ten is the largest Desk A value but occurs only once, so it is not a mode. The two most frequent Desk A values each occur for two of five visitors and therefore do not individually form a majority. Largest, most frequent and more than half name different relationships.

CQ3. Deleting ten solely to make the mean look more typical is not justified. C2 confirms that it is a valid observation inside the exercise’s coverage. The writer can report its influence on the mean and use the median alongside the full distribution. The check establishes that the entry was genuinely recorded under the rule; it does not explain the longer wait’s cause. Calling the value unusual should prompt inspection, not automatic erasure or an invented story about a careless visitor or an inefficient worker.

CQ4. The interval frequencies are four, zero and one. The exact ten enters the interval beginning at ten, not the interval ending just before it. From these grouped counts alone, the exact original mean and median cannot generally be recovered because many different sets of values would fit the same bins. A reader can still say that four of five observations are below five minutes and one is at least ten but below fifteen. Grouping preserves some useful information while discarding the precise values needed for other questions.

CQ5. One comparison is: “Both desks have a mean recorded wait of four minutes across five visitors. Desk A’s median is three minutes, with four short waits of two or three and one confirmed ten-minute wait; its range is eight minutes. Desk B’s five recorded waits are all four minutes, so its range is zero. The equal means therefore do not establish identical individual experiences, and these exercise records do not determine future waits or their causes.” The account combines centre and variation without forcing one summary measure to describe everything.

Transfer C. Add a sixth distinct Desk A visitor with a confirmed wait of two minutes. The ordered values become two, two, two, three, three and ten. Their sum is twenty-two, mean eleven thirds of a minute, approximately 3.67 minutes, median 2.5 minutes and mode two. The range remains eight. The median now lies between two observed values rather than matching one visitor’s wait. Explain why the extra observation changes some summaries while leaving the extremes unchanged.

A writing bridge. Before choosing a phrase such as typical wait, ask which feature the reader needs. A person interested in the most common observations may need the modes; a comparison of aggregate waiting across equal group sizes may use means; a concern about the longest observed wait needs the maximum. A well-written answer states the selected measure and can acknowledge another when it changes the interpretation. The purpose is not to find one favourite average and apply it everywhere.

Laboratory D: More participants, but a smaller participation share

D1 — two separate cohorts. A fictional workshop offers two introductory events during equal-length invitation windows. Event One has forty eligible invitees and twenty participants. Event Two has sixty eligible invitees and twenty-four participants. No person belongs to both cohorts. Every participant belongs to the eligible cohort for that event, and the registers completely and accurately record participation under the same stated rule.

D2 — what participation means. A participant is an eligible person whose attendance at the introductory session was recorded. Each person is counted once, even if they also attend a later activity outside this packet. An expression of interest without attendance does not qualify. The eligible counts include people who did not attend. No participant’s reason for attending or staying away was collected.

D3 — a change before the second event. The organisers display a new poster before Event Two. The packet does not establish who saw it, whether other publicity changed, or whether the two cohorts had the same availability or interests. The poster’s appearance is a documented fictional event, but this is not a controlled comparison designed to isolate its effect.

D4 — competing headlines. Alicia writes, “Participation increased by 20%.” Tricia writes, “The participation share fell from 50% to 40%.” Kai Kai insists that one calculation must be wrong and proposes, “The poster caused a twenty-per-cent improvement in everybody’s interest.” He also calculates the simple mean of 50% and 40% as 45% and treats it as the proportion attending across both cohorts combined.

D5 — the writing task. The editor wants to know what changed in the observed counts and shares, not which headline makes the event look most successful. There is no future attendance record, no survey of interest and no repeated measurement of the same individuals across events. Explain the supplied comparison without inventing missing responses or motives.

Questions D

DQ1. Calculate the absolute and relative change in the participant count. Revise Alicia’s headline so its measured quantity is explicit.

DQ2. Calculate the attendance share for each cohort. State the change in percentage points and the relative percentage change in that share.

DQ3. Calculate the attendance proportion among all eligible people across the two events. Explain why the simple mean of the two percentages gives a different result.

DQ4. Assess the claim that the poster caused an improvement in everybody’s interest. Distinguish a recorded sequence, a measured outcome and an unsupported causal or individual claim.

DQ5. Write a four-sentence comparison that preserves both the higher count and the lower share, with one relevant limit.

Worked answers D

DQ1. The count increases from twenty to twenty-four, an absolute increase of four participants. Relative to the starting twenty, four is 20%, so the participant count increases by 20%. A clearer headline is, “The recorded participant count increased by 20%, from twenty to twenty-four.” This identifies the quantity Alicia calculated. The original word participation could also be read as a proportion or a broader judgement, so the repair makes the valid arithmetic’s meaning explicit rather than changing its numerical result.

DQ2. Event One has twenty participants among forty eligible invitees, giving 50%. Event Two has twenty-four among sixty, giving 40%. The share decreases by ten percentage points. Relative to the original 50% share, that ten-point decrease is 20% of the starting share. A 20% increase in the participant count and a 20% relative decrease in the attendance share are therefore both correct here. They compare different quantities. The eligible population expands from forty to sixty, and that changed denominator explains the apparently conflicting directions.

DQ3. Across the two non-overlapping cohorts, forty-four of one hundred eligible people attend, giving a combined attendance share of 44%. The simple average of 50% and 40% is 45%, but it gives each cohort equal weight despite their unequal sizes. The combined-person proportion instead gives each eligible person one place in the denominator. Weighting the two shares by forty and sixty produces the same 44% as combining the counts. Neither averaging method should be selected merely because it produces a preferred number; the intended question determines the appropriate weighting.

DQ4. The packet records that the poster appeared before Event Two and that the two events had different attendance counts and shares. It does not isolate the poster’s causal effect, establish that every participant saw it or measure interest directly. Attendance and interest are different variables. Because the cohorts contain different people, the comparison also does not show that any particular individual became more interested between events. The poster may have contributed, but the supplied information does not establish that explanation. Lack of demonstrated causation is not proof that the poster had no effect.

DQ5. One suitable account is: “Event Two recorded twenty-four participants compared with twenty at Event One, a 20% increase in the participant count. Its attendance share nevertheless fell from 50% to 40% because the eligible cohort increased from forty to sixty. Across the one hundred non-overlapping eligible people, forty-four attended. Although a new poster appeared before Event Two, these records do not establish its effect or measure changes in individual interest.” The paragraph preserves useful results while separating the interpretation that the packet cannot support.

Transfer D. Start a separate altered scenario in which Event Two has forty eligible people rather than sixty, while its participant count remains twenty-four. Its share is now 60%, ten percentage points above Event One’s 50% and a 20% relative increase in the share. The participant-count increase remains four, or 20%. This is not a correction to the original record: it is a deliberately changed denominator. Explain why changing one condition reverses the share comparison without changing the two participant counts.

The sentence-level lesson. The phrase “more participants” reports a count comparison, whereas “a greater proportion participated” compares shares. Neither is automatically a complete judgement of success. A workshop could have several aims, such as reaching more people, engaging a larger share of invitees or serving a previously excluded group. The task must define the criterion before a writer turns one numerical movement into the general word improvement.

Laboratory E: The missing Wednesday and the rising running total

E1 — two independent panels. This laboratory contains a four-week record and a separate five-day trial. They describe different fictional activities and must not be added together. Panel One records complete counts of eligible transactions in four equal weeks: Week 1 has eighteen, Week 2 twenty-four, Week 3 twelve and Week 4 thirty. Each transaction belongs to one week only, and there are no later removals or reclassifications in the panel.

E2 — the running-total convention. Panel One’s cumulative count begins at zero immediately before Week 1. Each week’s contribution is added to all previous weeks. Its cumulative values at the four week endings are eighteen, forty-two, fifty-four and eighty-four. A weekly count and the cumulative count through that week are two descriptions of the same transactions at different scopes, not independent sets of transactions.

E3 — the separate daily trial. Panel Two records Monday: two; Tuesday: four; Wednesday: missing; Thursday: zero; Friday: six. All five days had an open session, and the four supplied values are complete, accurate counts under the same definition. Wednesday’s register was lost. Thursday’s zero is a confirmed observation of no qualifying transactions, not a missing entry or a closed-session code.

E4 — a confirmed bound. Independent checks establish that Wednesday could have contained no fewer than zero and no more than ten qualifying transactions. Its unknown value is therefore a whole number from zero to ten inclusive. These checks do not reveal which value occurred. They supply a boundary, not a replacement observation. No other information about Wednesday’s count is available in the original packet.

E5 — the proposed repairs. Alicia adds the four known daily counts and obtains twelve. Kai Kai divides twelve by five and reports an exact five-day mean of 2.4, silently treating the missing day as zero. Tricia proposes distinguishing the mean of the four known observations from a bounded five-day result. A separate graphing suggestion estimates Wednesday as two by drawing a straight line between Tuesday’s four and Thursday’s zero.

Questions E

EQ1. Explain how Panel One’s weekly counts fluctuate while its cumulative counts rise at every week ending. Give its four-week transaction total and the net difference between the first and last weekly counts.

EQ2. Calculate the mean of Panel Two’s four known daily observations. Explain why it is not automatically the exact mean for all five days.

EQ3. Using the confirmed Wednesday bound, state the smallest and largest possible five-day totals and means. Is 2.4 impossible, established or one possible endpoint?

EQ4. Evaluate the interpolated Wednesday value of two. Explain how a display should distinguish that estimate from a recorded value and from Thursday’s zero.

EQ5. Write a concise description of Panel Two that reports the known observations and the remaining uncertainty without making every day sound unknown.

Worked answers E

EQ1. The separate weekly counts rise from eighteen to twenty-four, fall to twelve and rise to thirty. They fluctuate, with Week 3 contributing fewer transactions than Week 2. Every contribution is positive, so the cumulative count still increases through eighteen, forty-two, fifty-four and eighty-four. The four-week total is eighty-four transactions. The difference between the last and first weekly counts is thirty minus eighteen, or twelve transactions. Eighty-four measures all transactions across the period; twelve compares two weekly values. Summing the cumulative values would repeatedly count earlier transactions.

EQ2. The four known daily counts total twelve, and their mean is 12/4 = 3 transactions per observed day. That result describes Monday, Tuesday, Thursday and Friday. The five-day mean depends on the unknown Wednesday count and uses five in its denominator. Reporting three as the five-day mean would assume Wednesday’s value equals three, while reporting 2.4 would assume it equals zero. Neither assumption follows from the lost register. A valid summary of known observations should keep its four-day coverage explicit.

EQ3. Let Wednesday’s count be w. The total is 12 + w, where w is an integer from zero to ten. The smallest possible total is twelve and the largest twenty-two. Dividing these endpoint totals by five gives possible mean bounds of 2.4 and 4.4 transactions per day. A mean of 2.4 is possible if Wednesday actually had zero, but it is not established. Not every decimal between the bounds is possible under the whole-number count rule; the means occur in steps of one fifth. The bounds remain useful even without an exact total.

EQ4. Two is the midpoint produced by a linear interpolation between Tuesday’s four and Thursday’s zero across equally spaced days. It is consistent with the allowed range, but it is not a recovered observation. Wednesday might instead have had eight or another allowed count without contradicting those neighbouring values. A display can leave a clearly labelled gap, or show a separately marked estimate with its method when that serves the task. It should not quietly draw an ordinary observed point at two. Thursday’s zero remains a directly recorded result under the packet’s convention.

EQ5. A suitable account is: “Four of the five daily counts are known: two on Monday, four on Tuesday, zero on Thursday and six on Friday. Their total is twelve and their mean is three transactions per observed day. Wednesday’s lost record leaves the full five-day total unresolved, but the independently confirmed zero-to-ten bound places it between twelve and twenty-two transactions inclusive. The corresponding five-day mean is bounded by 2.4 and 4.4.” This description is confident about the known values and precise about the specific information gap.

Transfer E. A recovered, authenticated record in the fictional case establishes that Wednesday had eight qualifying transactions. The complete daily sequence is now 2, 4, 8, 0 and 6. Its total is twenty and its five-day mean is four. The recovery changes what the readers know about the past; it does not cause eight transactions to happen retrospectively. The old four-known-day mean of three remains an accurate calculation of that subset, but it should not be retained as the best summary of the now-complete five-day record.

A limit can be a result. The original packet does not force a choice between pretending to know the exact count and saying nothing useful. Its confirmed bound excludes totals below twelve and above twenty-two. A learner who explains that interval has made progress. The answer should not acquire an invented midpoint merely because one exact number would fit more neatly into a worksheet box.

Laboratory F: Eighty per cent of whom?

F1 — the scheduling question. A fictional club invites all sixty eligible members to answer whether they support moving a workshop to an earlier time. By the stated deadline, ten distinct members have submitted valid answers: eight support the move and two oppose it. Each of those ten answers the question explicitly. The other fifty members’ views are unknown. Responses are voluntary, and the packet does not establish a representative sampling procedure.

F2 — the duplicate row. The original form export contains eleven rows. One positive response appears twice with the same confirmed submission identifier and unchanged content. The data checker establishes that the second row is an accidental copy of the same submission, not another member’s answer or a later change of view. The counts in F1 already remove that extra copy. The remaining ten submissions belong to ten distinct eligible people.

F3 — an unrelated activity question. A separate five-person exercise asks which of art and music each person enjoys, allowing both. Person A chooses art and music; B chooses art only; C chooses music only; D chooses art and music; E chooses art only. All five submit complete answers. These five people are a separate fictional group; their responses must not be combined with the sixty-member scheduling question.

F4 — the intended counts. For the activity exercise, count each person once in each category they select. A person who selects both therefore contributes to both category counts. For a distinct-person total, count each person only once. The packet provides the complete identities and selections, so the overlap is known rather than estimated. No one was asked to choose a single favourite.

F5 — the draft report. Kai Kai writes, “Eighty per cent of all sixty members want the earlier time, so forty-eight support it. Art and music account for 140% of five people, proving that the second survey is wrong.” Alicia checks the first denominator. Tricia separates selections from people in the second question and asks whether a pie chart could use mutually exclusive response-pattern categories instead.

Questions F

FQ1. Explain why the eleven exported rows represent ten respondents. Calculate the share supporting the schedule change among valid respondents and assess the claim of forty-eight known supporters.

FQ2. Without inventing the fifty missing answers, state the smallest and largest possible numbers of supporters among all sixty members, treating the recorded answers as fixed for this exercise. Does the packet establish a whole-club majority?

FQ3. Calculate the art and music category counts and percentages in the five-person exercise. Explain why their percentages total more than one hundred without an error.

FQ4. Calculate the number choosing at least one activity and divide the five people into art only, music only and both. Explain which grouping could form parts of one pie chart.

FQ5. Write one corrected sentence for each survey. Preserve their different populations and question formats.

Worked answers F

FQ1. One of the eleven rows is a confirmed duplicate of the same positive submission, so it must not count as another person. Removing the extra copy leaves eight positive and two opposing answers from ten distinct respondents. The support share among them is 8/10 = 80%. Multiplying that share by sixty to claim forty-eight known supporters incorrectly applies the respondents’ share to the entire membership. The record establishes eight known supporters, not forty-eight. Identical values alone would not justify deletion; the confirmed shared submission identity is what establishes this duplicate.

FQ2. At least eight members support the move because their support is recorded. At most fifty-eight could support it: the eight known supporters plus all fifty members whose views are unknown, while the two known opponents remain opposed under the exercise’s fixed-answer assumption. The corresponding possible shares run from eight sixtieths to fifty-eight sixtieths. Both a whole-club minority and a whole-club majority remain compatible with the packet, so neither is established. These are logical bounds from missing answers, not confidence intervals or estimates from a representative sample.

FQ3. Art is selected by A, B, D and E: four of five people, or 80%. Music is selected by A, C and D: three of five, or 60%. The percentages sum to 140% because A and D belong to both categories. There are seven category selections made by five distinct people. This is a legitimate consequence of a multiple-response question. The figures would be inconsistent only if the report falsely described art and music as mutually exclusive parts that each person could enter just once.

FQ4. Four choose art, three choose music and two choose both, so the distinct number choosing at least one is 4 + 3 − 2 = 5. The mutually exclusive pattern groups are art only: B and E, two people or 40%; music only: C, one person or 20%; both: A and D, two people or 40%. These three groups partition the five people and could form one pie chart of response patterns. Separate art and music sectors using 80% and 60% would not form a valid partition of that same whole.

FQ5. For the first question: “Eight of the ten valid respondents supported the earlier workshop time, but the other fifty eligible members’ views were not recorded by the deadline.” For the second: “In the separate five-person multiple-response exercise, four selected art and three selected music, with two selecting both.” The sentences preserve the useful findings without extending one sample to all members or treating overlapping selections as different people. They also avoid calling either preference record a measurement of what the individuals actually did during a later workshop.

Transfer F. Change the activity question so that each person must now choose one favourite. The previous multiple-response answers do not reveal which single option A and D would choose. Collecting that different answer requires another question or additional supplied information; assigning one of them to each category for a tidy result would invent preferences. The old art-only and music-only answers indicate what was selected under the earlier question, but the exercise should not assume that an altered question has already received five new responses.

Bring the six laboratories back to the words

The recurring errors are changes in meaning: a transaction becomes a person, a visible bar length becomes an absolute count, a group mean becomes an individual experience, a count becomes a share, a missing value becomes zero, and a category selection becomes a distinct respondent. The calculations expose those changes, but the sentences carry them. Accurate vocabulary is what keeps the quantity, its reference and its evidential limit together when the result leaves the working page and enters a reader’s understanding.

Original story: The Missing Word in the Headline

The following story is an invented editorial scene using the scheduling figures from Laboratory F. Its conversations and actions are fiction, not a report about real students or an actual club. Read first for what the characters do and want. Then examine how the statistical distinction affects their choices. The language of data becomes part of the story’s conflict rather than a set of definitions interrupted by a decorative plot.

Kai Kai had left a space exactly the width of his ruler across the top of the newsletter. By lunch, that space contained six large words and a number: “80% OF MEMBERS BACK AN EARLIER START”. Beneath it, a carefully drawn clock pointed to the proposed workshop time. He lifted the page to see whether the headline could be read from the doorway. It could. The letters looked decisive from almost anywhere in the room.

Alicia was still at the smaller table with the response register. She had drawn a thin line beside two identical submission numbers, then copied the checker’s note below them: same submission, accidental duplicate. Eleven rows had become ten valid responses. Eight supported the change and two opposed it. She had worked through the fraction twice. Eight tenths was four fifths, and four fifths was eighty per cent. That part was settled.

“Does it fit?” Kai Kai asked.

Alicia looked up. Her first reaction was relief. The page no longer resembled a collection of separate tasks. The headline, the short report and the small notice about the next exhibition finally looked as though they belonged together. She nodded before reading the heading again. Then her finger stopped on the word members.

“Where did respondents go?”

Kai Kai turned the page towards himself. “It was too long. Members means the same people.”

“Some of the same people.” Alicia pushed the register across the table. “Ten people answered. There are sixty members.”

He studied the fraction she had written. “But the eighty is right.”

That was what made the mistake difficult to point at. No digit needed crossing out. The denominator had not been changed on the working page. It had disappeared on the newsletter, taking one word with it. Alicia could still see the original sentence on her draft: Eight of the ten respondents supported the earlier time. She had thought that giving Kai Kai the checked figure meant giving him a checked headline as well.

Tricia returned carrying the teacher’s folder. She read the heading without seeing Alicia’s notes. “Forty-eight people, then?” she asked.

Kai Kai opened his mouth and closed it. Forty-eight was not written anywhere in the response register. Yet it was exactly what eighty per cent of sixty would mean. The page had allowed Tricia to calculate a number that none of them had observed.

“Eight,” Alicia said. “Eight known supporters. Fifty people haven’t answered.”

Tricia lowered the folder. “Then perhaps we cannot use the survey.”

“We can use the answers we have.” Alicia tapped the ten checked entries. “We just cannot make them answer for the other fifty.”

Outside, a chair scraped along the corridor. The teacher would collect the proof after the next activity. Kai Kai measured the headline space again, as though its dimensions might have become more generous during the conversation. Respondents had eleven letters. Members had seven. He had made the change to keep the line straight, not to make the group larger. On the page, however, the reason for the edit did nothing to limit its effect.

“We could put the explanation in small writing below,” he said.

Tricia held the newsletter at arm’s length. From there, she could read the headline but not the proposed note. “Someone might stop here,” she said, covering the body with the folder. “They would still think forty-eight.”

Alicia took a clean strip of paper and wrote: “EIGHT OF TEN RESPONDENTS SUPPORT EARLIER TIME”. It ran past the ruler. Kai Kai did not object this time. He drew a faint mark where the line could break. The second half dropped neatly beneath the first. The clock could move a little lower.

“Now it looks like a smaller story,” he said.

“It is a story about ten answers,” Tricia replied.

She sat beside Alicia and began checking the report under the heading. Its first sentence said that the club had discovered what its members wanted. She underlined discovered. Its last sentence said the schedule would change next month. No decision notice appeared in the teacher’s folder. There was only a request for an accurate summary of the responses received by the deadline.

Kai Kai watched the pencil move. “Are we losing the whole article?”

Tricia shook her head. She wrote above the first sentence: The club received ten valid responses. Beside the last, she wrote: No decision about the timetable is recorded in this packet. Between them, the useful part remained: eight responses in favour, two opposed, and fifty members without a recorded answer.

Alicia checked the final figure against the membership list. Sixty minus ten was fifty. She almost wrote that those fifty had ignored the question, then stopped. Some might not have seen it. Some might have intended to reply. Some might have chosen not to. The register contained no explanation. She wrote had not responded by the deadline instead. It was less lively than ignored, but it did not assign a motive to people whose answers they did not have.

The teacher arrived while Kai Kai was lifting the old headline. One corner resisted, leaving a pale rectangle where the paper had been attached. He held the discarded strip without folding it. The teacher looked from the strip to the new proof.

“What changed?”

“Not the percentage,” Kai Kai said. “The people it seemed to describe.”

He explained the duplicate first, because it had changed the number of valid submissions. Then he pointed to respondents, because that word had changed the headline’s scope. Finally, he showed the sentence about the timetable, because a preference result had been made to sound like an announced decision. The three repairs concerned different things. He could no longer fix them all by saying that the figures had been checked.

The teacher read the proof and the record together. “Read it to someone who has not seen your calculations.”

Alicia covered her working. Tricia read only the revised newsletter. When she finished, she said, “Eight of the ten people who replied support the proposal. We do not know the other fifty members’ views from these records. And the timetable has not been confirmed here.”

Kai Kai looked at Alicia. She checked each statement against the packet and nodded. This time her nod came after the reading, not before it.

There was still space under the clock. Kai Kai filled it with a short source note identifying the response deadline and the ten valid answers. He did not put the missing word there. It stayed in the headline, large enough to read from the doorway. When he set the final proof beside the register, he placed the rejected strip behind it instead of in the bin. Across the back he wrote what the first version had lost: not four letters, but fifty unanswered people.

Reading workshop: interpret the story without inventing evidence

Reading question 1. Why does Alicia initially approve the page and then question it? Support the change with two moments from the story. Worked answer: Her first response is to the layout: the separate pieces finally appear to belong together, so she nods before reading the heading carefully. Her finger then stops at members because that word extends the percentage beyond the ten respondents. The contrast shows that an attractive finished appearance can encourage premature approval. It does not establish that Alicia knowingly approved a false claim.

Reading question 2. How does Tricia’s calculation of forty-eight help reveal the problem? Worked answer: She interprets the heading as referring to all sixty members and calculates 80% of that whole. The arithmetic follows naturally from the misleading heading, but forty-eight supporters are not present in the response record. Her reading demonstrates a consequence of the wording that the designers had not intended. The problem is therefore more than a theoretical concern about an omitted word: another reader actually reconstructs a different claim.

Reading question 3. Why is moving the qualification into small writing not an adequate repair in this scene? Worked answer: The large heading would still independently state the unsupported whole-membership claim, and Tricia shows that a reader could stop at it or be unable to read the small note from a distance. The necessary limit belongs with the main claim. A source note can add useful context, but it should not contradict or quietly undo what the most prominent sentence says.

Reading question 4. Does the story portray Kai Kai as deliberately dishonest? Distinguish evidence of motive from the effect of his action. Worked answer: The narration states that he shortened the wording to fit the layout, not to enlarge the group. Nevertheless, the resulting headline does enlarge the apparent scope. His later willingness to break the heading across two lines and explain the corrections supports a reading of someone reconsidering an error. The harmful effect of an edit need not establish a dishonest intention; both the effect and the supported motive can be described accurately.

Reading question 5. Explain the final phrase about losing fifty unanswered people. Worked answer: The edit from respondents to members removes the distinction between ten known answers and fifty unknown views. The phrase gives a human consequence to an apparently minor wording change. It does not mean those fifty people physically disappeared or that their opinions were known to be different. It means the first headline represented the whole group as though the missing answers had already been supplied.

Writing transfer. Retell the scene from Tricia’s viewpoint, beginning with her entry into the room. Keep the numerical facts unchanged. Because this version follows her perspective, do not state Kai Kai’s private intention as something she already knows before he explains it. She can observe the ruler, hear his explanation and revise her judgement. Viewpoint control and statistical scope have a common writing demand: distinguish what is available to the speaker from what the author or reader knows elsewhere.

Alternative opening to study. “Forty-eight, Tricia thought, before she had reached the table. The number was not printed on the newsletter. It arrived from the large eighty-per-cent heading and the sixty names she had counted yesterday.” This opening foregrounds the reader’s inference rather than the designer’s effort. It uses the same fictional facts to create a different route into the conflict. A change of narrative emphasis need not change the evidence that the scene eventually resolves.

Writing clinic: explain the number without changing its meaning

A useful data paragraph answers a question. It does not simply repeat a display’s title, list every visible value and finish with an unsupported explanation. Start by deciding whether the reader needs a count, share, rate, distribution or trend. Locate the information required for that question, then choose words that preserve the relationship. Technical vocabulary is helpful when it names a distinction the reader needs, not when it makes an already clear sentence harder to follow.

Build from a complete factual sentence

“Moon was higher” leaves the measure unstated. “Moon recorded thirty loan transactions during the four-week period” supplies a subject, quantity, value and period. To make the comparison, add Sun’s corresponding figure: “Moon recorded thirty loan transactions compared with Sun’s twenty-four during the same four-week period.” Now ask whether the question requires anything further. A question about distinct borrowers needs the separate twelve-borrower counts; a question about transactions per hour needs the opening-hour denominators.

Do not force every relevant fact into one sentence. A clear first sentence can establish the comparison, a second explain the denominator and a third identify a necessary limit. The paragraph’s length follows the question. When the question asks only which club recorded more loans, an answer does not need an essay on every possible sampling limitation. When it asks whether more children benefited, the distinction between transactions, people and an unmeasured benefit becomes central.

Choose a connector that matches the relationship

Use whereas to hold different comparisons together: “Moon recorded more transactions, whereas Sun had a higher transaction rate per opening hour.” Use because only when the explanation is established at the relevant level. “The pooled percentage differs from the simple average because the groups have unequal sizes and the calculations weight them differently” explains a mathematical relationship. “Moon borrowed more because its members enjoyed reading more” adds an unmeasured psychological explanation. The same connective can introduce either sound reasoning or an unsupported claim.

Although can preserve a limitation: “Although both desks had a mean of four minutes, their individual waits differed.” Therefore should follow adequate premises: “The source confirms ten valid respondents and eight favourable answers; therefore, the favourable share among respondents is 80%.” The final phrase matters. Without among respondents, the conclusion may become broader than the premises. A transition does not relieve the writer of checking the subject to which the conclusion applies.

Protect by, to, from and of

“The count rose to twenty-four” names its endpoint. “The count rose by four” names the increase. “The count rose from twenty” names its starting value. “The new count was 120% of the original” describes a ratio of full values. These expressions can refer to the same twenty-to-twenty-four change, but they are not interchangeable. A careful writer can check them by reconstructing the original arithmetic from each sentence.

The same attention helps with lists written in words. In Laboratory B’s transfer task, the three separate weekly values are 80, 100 and 70. Week 3 is seventy, not part of a single value of one hundred and seventy. Using numerals or repeating the week labels removes that possible reading: “Week 1 recorded eighty cards, Week 2 one hundred, and Week 3 seventy.” Clarity sometimes comes from adding a label rather than replacing a familiar word with a technical one.

Keep a reported result separate from a forecast

“The recorded count increased” describes the supplied observations. “The count is expected to increase” reports an expectation, which needs its source or method. “The count would reach sixteen if the two-per-week increase continued” gives a conditional calculation. These sentences do different jobs. A forecast should not become a past observation merely because it sits beside measured values in a table. Its status must survive the move from the working to the final paragraph.

When revising, underline verbs that imply evidence status: recorded, estimated, predicted, confirmed and suggested. Ask what supports each one. A confirmed transcription means the copied value matches the checked source; it does not necessarily establish the original measurement’s accuracy. Suggested can mark an interpretation, but it should not be used to smuggle in an explanation with no supporting connection. State the actual basis for the sentence instead of relying on a cautious-sounding verb alone.

Select evidence without hiding the awkward part

A paragraph describing the waiting times should not delete the ten-minute observation merely because most waits were short. It can say that four visitors waited two or three minutes and one waited ten. That sentence identifies the concentration and the exception. A comparison focused on the arithmetic means may then add that both desks averaged four minutes. The order can vary with the question, but the relevant distribution should not disappear behind the chosen summary.

Selection is not the same as distortion. A short answer need not mention every possible statistic. It does need to preserve evidence that materially changes its central claim. “Both groups had the same mean” can be sufficient for a mean question. “Both groups had the same experience” cannot be substantiated by that equality. The correction should target the actual overstatement, not require every answer to become a complete statistical report.

Write a specific limit, not a general disclaimer

“The data might not be perfect” tells the reader little. “Wednesday’s register is missing, so the exact five-day total is unknown” identifies the affected claim and the reason. The four other daily counts remain available. Similarly, “The other fifty members did not respond” limits a whole-club preference statement without casting doubt on the arithmetic for the ten known answers. A precise qualification gives the reader a boundary they can use.

Place that boundary near the claim. A large heading saying all members support the change cannot be repaired adequately by a distant note saying only ten answered. First make the main sentence accurate. Then use a source note for details that help readers understand its method or period. A qualification should complete the claim rather than contradict it after the reader has already formed a different understanding.

Word families: change the grammar without losing the evidence

Compare and comparison. “Alicia compares the borrower shares” describes an action. “The comparison concerns borrower shares” makes the analytical task the subject. Both should preserve the same quantities and denominators. “The clubs are comparison” does not work because the noun has been placed where a different construction is required. Practise the whole sentence pattern rather than memorising an ending and inserting it wherever a word looks longer.

Vary, variation and variability. “The recorded waits vary” states that they differ. “There is variation among the recorded waits” expresses the same broad finding in a noun construction. “The variability increased” makes an additional comparison between sets or periods; it cannot be inferred from one list alone. Changing the grammatical form can introduce a new claim, so check meaning as well as sentence correctness.

Infer and inference. “Tricia infers that thirty-six members did not borrow” relies on the complete membership and borrower counts. “Her inference is supported by the complete record” evaluates that reasoning. Neither should become “The record proves nobody wanted to borrow”, which changes behaviour into desire. The related word forms keep a connection to reasoning, not permission to enlarge the conclusion.

Estimate and estimated. “Kai Kai estimates Wednesday’s count” identifies an action that needs a method. “Wednesday’s estimated count is two under a linear interpolation” marks the result’s status. Removing estimated makes it look like an ordinary observation. A grammatical editing exercise should therefore ask whether the evidence label survived the rewrite, not merely whether the new sentence has a subject and verb.

Accurate and accuracy. “The copied value is accurate against the source” describes a verified transcription. “The accuracy of the original measurement remains unassessed” names a separate issue. The two statements can be true together. More decimal places would alter displayed precision, not complete the missing measurement check. Use the specific noun phrase that identifies the layer being evaluated.

Qualify and qualification. “Alicia qualifies the headline by restoring respondents” names the repair. “The qualification limits the claim to the ten answers received” explains its effect. The second sentence shows why the word matters. A learner who can name that effect has more control than one who says a qualification simply makes writing sound less confident.

Twelve precision repairs with explanations

Each first sentence below is an original practice error, not a quotation from a real pupil. Cover the repair, write your own correction and explain which relationship changed. Several revisions can be acceptable. The test is whether the result preserves the evidence, not whether it matches a model sentence word for word.

1. “The club recorded thirty loans, so thirty different people borrowed.” Repair: “The club recorded thirty loan transactions; the distinct-borrower count must be checked separately.” The original turns events into people. One borrower can contribute several valid transactions. The correction does not claim that thirty borrowers are impossible; it says that the transaction total alone does not establish them. Check identities when the question concerns different people.

2. “Both clubs had twelve borrowers, so their borrower percentages were equal.” Repair: “Their borrower counts were equal, but their percentages depend on the respective membership totals.” Twelve out of twenty is 60%, while twelve out of forty is 30%. Equal numerators do not imply equal ratios when the denominators differ. Keep the counts as a genuine similarity and describe the share comparison separately.

3. “The combined member share is the simple average of the two club percentages.” Repair: “For the share among all members, combine compatible borrower and membership counts or weight each club’s share by its membership.” A simple average gives each club equal weight. That is not the same as giving each person equal weight when the groups have unequal sizes. Name the unit receiving the weight before choosing the calculation.

4. “A share falling from 50% to 40% decreased by ten per cent.” Repair: “The share fell by ten percentage points, which is a 20% decrease relative to the original 50% share.” The original confuses an absolute difference between percentages with a relative change. Reporting both endpoints helps readers check the relationship. Do not shorten percentage points to per cent merely to make the sentence look more familiar.

5. “The mean wait was four minutes, so everybody waited four minutes.” Repair: “The arithmetic mean was four minutes; individual waits must be read from the distribution.” The five values 2, 2, 3, 3 and 10 have that mean without containing a four. A mean is a summary, not a fictional person whose experience can be assigned to every member of the group. Use another statistic or the actual values when the question requires them.

6. “Ten is the mode because it is the largest value.” Repair: “Ten is the maximum; the mode is determined by frequency.” In the supplied list, two and three each occur twice and therefore share the greatest frequency. The maximum occurs only once. Sorting can help locate the extremes and middle, but finding the mode also requires counting how often each value occurs.

7. “Wednesday is blank, so there were no transactions.” Repair: “Wednesday’s count is missing; a blank cell does not establish a recorded zero.” The source note says the register was lost. Treating the blank as zero creates an observation and changes the five-day mean. Preserve the gap or use a separately labelled estimate under a stated method. Neither option should conceal what was and was not actually recorded.

8. “Two people gave the same answer, so one row should be removed as a duplicate.” Repair: “Matching answers can come from distinct respondents; duplicate status requires evidence that the same eligible unit and submission were entered twice.” In Laboratory F, the shared confirmed submission identifier establishes the duplicate. Two different people selecting art do not form a duplicate merely because the values match. A correct cleaning rule follows identity, not a desire to make every answer unique.

9. “The activity percentages total 140%, so the survey must be wrong.” Repair: “The multiple-response categories overlap, so their separate shares can sum above 100%.” Four of five choose art and three of five choose music, with two choosing both. The percentages count category membership, not mutually exclusive parts. A different display can show art only, music only and both as a partition, but the original choices should not be changed to force a pie chart.

10. “The cumulative total rose, proving that every week’s count increased.” Repair: “The cumulative total rose because each recorded week added transactions; the separate weekly counts must be compared to describe weekly change.” The sequence 18, 24, 12 and 30 has a falling third-week contribution despite rising cumulative totals. A running sum contains earlier activity. Its direction does not automatically describe the direction of the newest contribution.

11. “The line passes through two on Wednesday, so two was observed.” Repair: “Two is a linear interpolation between neighbouring readings, not an observed Wednesday count.” Drawing or calculating the midpoint does not recover the lost register. A forecast outside the observed period requires an equally visible distinction between predicted and recorded. The arithmetic can be exact under an assumption while the assumed relationship remains unverified for the missing event.

12. “The report was printed this year, so its survey values describe this year.” Repair: “The values describe the observation period stated in the source note; publication does not automatically update them.” An older survey may still answer a historical question accurately. The missing evidence concerns current conditions, not the usefulness of every earlier observation. Read collection date, publication date and later access date as different pieces of information.

Independent transfer: one record, two legitimate summaries

New fictional packet. A workshop records the number of completed paper models made by each of six participants during one specified session: A makes zero, B one, C one, D two, E two and F six. The count concerns completed models only; all six eligible participants are included, and the records are complete and accurate within the exercise. No record explains the differences, the time each person spent actively working or the number of unfinished models.

Writing task. Produce one sentence for someone asking about total output, then a separate short paragraph for someone asking whether all participants produced similar amounts. Select the evidence each reader needs. Do not label a participant lazy, talented or distracted from the count alone. Finally, explain why combining the observations into one total does not answer every question about the participants.

Worked response. “The six participants completed twelve models during the session.” For the distribution question: “The completed-model counts varied from zero to six, with most participants producing no more than two. The mean was two models per participant, but the ordered middle pair of one and two gives a median of 1.5, and one and two are tied modes. The single six-model count differs markedly from the other recorded amounts. These figures describe completed output, not effort, ability or the cause of the differences.” The total and distribution summaries answer different questions using the same valid observations.

Changed condition. A source correction establishes that F completed four models, not six, because two unfinished models were mistakenly included. The corrected total is ten, mean five thirds, approximately 1.67, median 1.5, modes one and two, and range four. Explain which sentence changes and why the median and modes remain unchanged. This correction follows a confirmed counting error; it is not permission to reduce a genuine large value because it looks inconvenient.

Reader test. Give the revised paragraph to a partner without the full packet and ask them to state the total, the unit and one thing the paragraph does not establish. When their reconstruction differs from your intended claim, locate the wording that allowed the difference. A missing unit, vague subject or unsupported adjective may be the real problem. Repair that feature and test again instead of automatically adding more technical vocabulary.

A six-session teaching sequence for data vocabulary

This is a suggested sequence, not a promise that every learner will master all 100 entries in six meetings. A session can be divided into shorter lessons or revisited with a different packet. Choose a small active group of words that addresses the learner’s present difficulty. Some specialist terms can remain recognition vocabulary while the learner explains their meaning accurately in ordinary language. Saying “these are the people who answered, not everybody invited” can demonstrate the essential distinction before respondent becomes an independently retrieved word.

Keep an initial response before showing the worked answer. Ask the learner to identify the question’s quantity, find its supporting record and explain one choice. Note any assistance: an adult reading the packet aloud, pointing to the correct column or supplying the denominator changes what the attempt demonstrates. Supported success is useful information, but it should not be confused with an independent reading performance. The next task can then address the actual remaining difficulty rather than repeat what the learner already controls.

Session 1: establish what receives one count

Use Laboratory A with count, total, respondent, proportion and rate. Begin without calculations. Ask what one transaction represents, what one distinct borrower represents and why the same person can contribute differently to those measures. Then calculate the club comparisons. Require the noun and period beside each result. A correct figure described as people when it counts loans needs a meaning repair, not another multiplication exercise.

End with the omitted-transaction variation. The borrower is already included, so recovering another loan changes the transaction count but not the number of distinct borrowers. Ask which sentences must change and which should stay. This checks whether the learner follows the identity of the new information instead of assuming that every corrected record alters every statistic. The explanation of the selective revision is the central outcome of this session.

Session 2: distinguish a display change from a data change

Use Laboratory B with baseline, scale, bar chart, percentage change and annotation. Have the learner state the two observed values before considering the bars. Calculate their difference and ratio from the numbers, then compare those results with the visible lengths above sixty. The learner should explain what the cropped lengths measure. A sketch can support the discussion, but the task’s stated values and conventions control the answer.

Compare the zero-based repair with the fixed-key pictogram. Both retain the original counts. Then discuss why a line chart can use a different axis range without making every such display invalid. The lesson is about how a particular representation encodes information, not a memorised accusation that every nonzero axis is dishonest. Finish by adding the third week’s count and revising only the time-window claims affected by that new observation.

Session 3: describe centre and variation together

Use Laboratory C with mean, median, mode, range and distribution. Let the learner sort and count the observations before selecting a summary. Ask whether anyone actually experienced the mean and whether the largest value is the most frequent. Then compare the two equal means alongside the different individual waits. The goal is to understand what each summary keeps and what it leaves unexpressed.

Inspect the proposed removal of ten. The record confirms it as a valid case, so the writer should discuss it rather than delete it for convenience. Follow with the grouped-interval display and ask what exact information is lost. This makes a useful distinction between a legitimate change of representation and an unjustified alteration of the evidence. A student can prefer the median for one descriptive purpose without declaring the mean mathematically wrong.

Session 4: keep the denominator beside the headline

Use Laboratory D with numerator, denominator, percentage point, percentage change and qualify. Write the two counts and the two eligible totals separately before forming shares. Ask how a count can rise while its share falls. Next distinguish the ten-percentage-point difference from the relative decrease in the share. Use the actual starting value each time instead of treating the word percentage as one universal operation.

Read the poster claim after the numerical work is secure. The packet documents a poster and different attendance results, but it does not isolate the poster’s effect or measure everybody’s interest. Ask the learner to retain the valid comparison while removing that unsupported extension. Finish with the changed denominator of forty. The resulting reversal tests whether the learner understands the relationship rather than remembers that this particular story always has a falling share.

Session 5: give missing information a precise boundary

Use Laboratory E with missing, zero, cumulative, estimate and net change. Keep the two independent panels separate. First compare the weekly values with their running totals. Then calculate the known-day mean in the daily panel without silently supplying Wednesday. Ask which observations are secure, which value is missing and what the confirmed zero-to-ten bound permits the writer to conclude.

Introduce interpolation only after the information gap is understood. A midpoint estimate can be calculated under a rule without becoming an observation. Compare it with the later recovered value and explain the difference between modelling an unknown and finding the original record. The learner should revise the whole-period summary when the missing day becomes known while recognising that the earlier calculation of the four-day subset was valid for its stated scope.

Session 6: edit a claim for a real reader

Use Laboratory F with population, sample, nonresponse, duplicate and overlap. Ask the learner to explain why eleven rows can represent ten respondents and why seven selections can belong to five people. These are different situations: one contains an accidental copy, while the other deliberately allows several category choices. Do not let a general instruction to remove repetition erase valid overlapping responses.

Read the editorial story and revise its headline before consulting the reading answers. Then use the independent paper-model packet to write for a new question. Ask a partner to paraphrase the finished finding without additional spoken hints. Where their reading changes the quantity or its scope, identify the missing word or reference and repair it. The final response should make the evidence easier to understand, not merely contain more entries from the vocabulary list.

Assess the particular skill, not a general label for the learner

Use a simple informal record: demonstrated independently, demonstrated with support, or not yet demonstrated in this task. Attach a specific example to each judgement. This is a teaching record, not an examination-board scale or a conversion to a school grade. A pupil who selects the correct denominator but makes one division error needs different feedback from a pupil who divides accurately using the wrong population.

Meaning control: Can the learner paraphrase the term and explain a contrasting case? For majority, test a largest category of four among ten as well as a six-person group. For duplicate, compare a confirmed copied submission with two genuine people giving the same answer. The contrasting case reveals whether the learner controls the boundary rather than only recognising the word beside one familiar example.

Reference and calculation: Does the response retain the row, column, unit, period and denominator that make the arithmetic relevant? Ask the learner to reconstruct the calculation from their own sentence. A mismatch may reveal that the written noun changed after the working was correct. Check the quantity’s identity before assigning more arithmetic practice. Repetition of an operation will not necessarily correct a misread heading.

Writing and evidence: Can another reader distinguish what was recorded, calculated, estimated and inferred? Inspect words such as all, caused, exact and current when they affect the claim. Do not demand a warning after every secure fact. The goal is calibrated writing: confidence about what the packet establishes and a specific limit on what it leaves open. An accurate short answer can demonstrate more control than a longer answer that repeatedly changes the measure.

Independent transfer: Change one consequential condition and ask which result changes. Add a transaction from an existing borrower, alter a cohort’s size or recover a missing observation. The learner should preserve unaffected facts. Their explanation of why some sentences stay the same shows whether the distinction is available for use, rather than merely remembered as a line from the previous worked answer.

Questions about learning charts and tables vocabulary

Must a Primary 6 learner use all 100 terms independently? No. The collection includes optional stretch vocabulary and several phrases whose underlying ideas can first be explained in everyday words. Select terms according to the learner’s curriculum and present task. A correct explanation of an unknown count’s lower and upper limits is worthwhile even when the learner does not yet use the term bounded uncertainty.

Does qualification make an answer weak? Not when it states the actual boundary of a strong finding. Eight of ten respondents is an exact result in the supplied packet. The qualification among respondents protects it from becoming a different claim about all sixty members. A vague perhaps would be less useful because it fails to distinguish secure arithmetic from unsupported population extension.

Will a calculator detect the main mistakes? A calculator can check an operation, but it cannot establish that the selected number counts the people named in the sentence. Dividing twelve by twenty accurately does not prove that twenty is the appropriate whole. Read the definitions and identify the required quantity before calculating. Use the result to answer that same question rather than switch meanings at the writing stage.

Does success with text packets establish every graph-reading skill? No. These complete written cases isolate meanings, numerical relationships and evidential limits. Reading an actual visual chart also requires locating plotted marks, following axes and interpreting its presentation. Use the two forms as complementary practice. A correct explanation of a described bar chart is not proof that a learner independently read an unfamiliar visual display.

How can multilingual learners use the material? Discuss the underlying relationship in a familiar language when that helps, then compare the English sentence pattern and scope. A translated word may cover more than one English distinction. Work with complete examples, such as count versus share or respondent versus eligible member, rather than assume that one memorised equivalent supplies every use.

What should a parent correct first? Begin with the error that changes the answer: an incorrect unit, denominator, comparison period, missing-value interpretation or evidence claim. Ask the learner to explain the repair. Then address awkward grammar or unnecessary complexity. Replacing the entire response with an adult paragraph may hide the specific distinction the child still needs to practise independently.

Reference notes and further reading

The Common Core Grade 6 statistics standards provide one curricular reference for statistical questions, centre, variability and data summaries. They do not prescribe this vocabulary selection or establish equivalence among international school levels. Statistics Canada’s introduction to data types provides another starting point for thinking about what different recorded values represent.

The Australian Bureau of Statistics provides guides to variables, censuses and samples, metadata, frequencies, measures of central tendency and measures of spread. Its discussions of types of error and correlation and causation support further checking of the named distinctions.

For display choices, consult the Office for National Statistics guidance on axes and gridlines, bar charts and line charts. Its percentage-writing guidance and the House of Commons Library percentage guide offer further clarification of relative changes and percentage-point differences. The NIST histogram explanation supplies a reference for grouped numerical displays.

These sources support the specified concepts; they do not endorse this collection or report its classroom results. The fictional datasets, story, questions, worked responses and transfer variations were written for this manual. Their numerical conclusions follow from their stated values and assumptions. They should not be presented as survey findings about actual children, evidence of tuition effectiveness or predictions about a real service.

Continue through the Primary 6 Advanced Vocabulary Collection

Return to the Primary 6 Advanced Vocabulary Collection within English Vocabulary Lists or the Vocabulary Article Directory to choose the next useful application. The Grade 6 Vocabulary List supplies broader word foundations, while the eduKate Vocabulary Learning System connects meaning, retrieval, application and transfer. Reading every preceding volume is not a requirement for using one relevant laboratory here.

Volume 5: Map Descriptions, Scale and Spatial Relationships examines positions, distances and spatial patterns. Use it when the difficult relationship lies in a map’s reference system or legend. This volume concentrates on charts, tables and statistical descriptions: a count, share, rate or average must remain attached to the question it answers.

Volume 4: Historical Timelines, Continuity and Change develops date and source distinctions, while Volume 2: Argument, Evidence and Critical Reading examines what claims and sources establish. For the literary choices in the editorial story, use Volume 1: Literary Analysis, Tone and Subtext. For explaining how a sequence operates, use Volume 3: Explanatory Writing, Processes and Systems.

The final principle: let the sentence keep the evidence intact

A number needs a quantity. A share needs its whole. An average needs its observations. A trend needs its period. A conclusion needs evidence suited to its scope. These relationships are carried by ordinary words as well as specialist vocabulary: among, per, recorded, estimated, by, to and from can determine what the entire answer means.

Start with the record, choose the comparison and write the strongest statement the evidence supports. Keep a real uncertainty visible without weakening facts that are secure. Then test the sentence with a reader and revise where its meaning changes in the handover. Advanced vocabulary is doing useful work when the reader understands not only the number, but also what was counted and where the conclusion must stop.

Continue to Volume 7: Experimental Design, Fair Tests and Scientific Conclusions

For the next application, open Primary 6 Advanced Vocabulary Collection | Experimental Design, Fair Tests and Scientific Conclusions for Grade 6. Volume 7 develops 100 entries for research questions, variables, controls, measurement, reliability, uncertainty, causation and evidence-based conclusions. Six original fictional investigations provide 30 questions with worked answers, supported by an original story, twelve precision repairs and a six-session teaching sequence.

Use Volume 6 when the main difficulty is reading charts, tables, percentages and averages. Use Volume 7 when the learner needs to decide whether an investigation is a fair test, distinguish controlled variables from control conditions, identify confounds or write a conclusion whose strength matches the experimental design.