IB Middle Years Programme · Mathematics MYP Year 3

Data Handling and Representation

Cover illustration for Data Handling and Representation (Mathematics MYP Year 3).
MYP · Mathematics MYP Year 3

Data Handling and Representation

MYP Mathematics — Statistics & Probability strand (Year 3)

22 min readStandardCore statistics unit — typically assessed under Criterion A (Knowing & Understanding) and Criterion D (Applying Mathematics in Real-Life Contexts); expect 20–30% of a data-handling unit test to be frequency-table and representation-choice questions.

This is the unit where marks get lost for silly reasons, not hard ones — a miscounted tally, a pie chart that doesn't add to , a chart chosen because it 'looks nice' instead of because it fits the data. The maths here is genuinely simple; the discipline of doing it carefully and in the right order is what examiners are actually testing.

Overview — The Data Handling Journey

Every data-handling question follows the same pipeline: collect raw data (often via a survey), organise it (tally and frequency tables), classify it (discrete, continuous, categorical), represent it (bar chart, line graph, pie chart, stem-and-leaf, dot plot), then interpret it (mode, range, trend). Examiners test each stage separately — and they test whether you can justify why you picked a particular representation, not just whether you can draw one.

  • Raw data is the unsorted list you're handed — it's almost never usable on its own.
  • A frequency table is the first transformation: it answers 'how many of each?' and nothing more.
  • Choosing the right chart depends on the type of data, not on which chart you find easiest to draw.
  • Command terms like 'construct', 'complete', 'explain' and 'analyse' each demand a different depth of answer — mixing them up is the single biggest source of dropped marks in this unit.

The shape of the chapter

Command terms that control the mark scheme in this unit

Command termWhat it demandsAOMark-earning move
ConstructBuild a table or chart from raw data, following all conventions (labels, scale, title).Criterion AMarks split between correct data AND correct presentation — a perfectly plotted chart with no title or axis labels still drops a mark.
CompleteFill in missing tally/frequency values in a table that's already been started for you.Criterion AEvery value must match the raw data exactly — one miscounted row forfeits that row's mark even if the rest is flawless.
IdentifyName one specific feature — a category, the mode, the modal class.Criterion AA single word/phrase is enough — but it must be the actual name (e.g. 'Blue'), not a number or a description.
ExplainGive a reason with a stated cause-and-effect link.Criterion DA bare assertion earns 0 — you must connect a feature of the data to a consequence, e.g. 'because... which means...'
InterpretDraw a conclusion about the real-world data using the completed table or chart.Criterion DMust reference actual figures from the table — a vague comment with no numbers loses the mark.
AnalyseBreak down a method or dataset and discuss a limitation or its reliability.Criterion DMust name the limitation AND state its effect on the validity of the conclusion — naming it alone gets half marks at best.
ApplyUse the frequency table to make a real decision (e.g. how much stock to order).Criterion DMust scale the real quantity proportionally to the frequencies, not just restate the table.
JustifySupport a choice (e.g. of representation) with evidence.Criterion DMust name a specific property of the data (type, number of categories, spread) and tie it directly to the choice made.

Key point

A frequency table throws away the original order of the data. The moment a question needs individual values back (exact range, ranking, outliers) you need a stem-and-leaf plot or dot plot, not a grouped table — grouped tables only give you classes, never exact figures.

Frequency Tables and Tally Charts

A tally chart is just a fast, error-resistant way of counting as you go through raw data once. Each tally mark represents one occurrence; marks are bundled in groups of five (four verticals, the fifth stroke crossing them diagonally) purely so the final count is instant addition, not one-by-one counting.

  • Ungrouped frequency table: one row per distinct value or category — used when there aren't too many different values (colours, dice outcomes, shoe sizes).
  • Grouped frequency table: raw numerical data is sorted into class intervals (e.g. 40–49, 50–59) — used when there are too many distinct values for a one-row-per-value table to be readable.
  • Class width is the size of each interval; it must be constant across the table unless the question specifically asks for unequal classes (rare at this level).
  • The frequency column must always sum to the total number of data items — this is the built-in check every student should run before submitting.
Frequency
The number of times a particular value or category occurs in a dataset."How many" — never a percentage or proportion unless converted.
Tally mark
A single stroke recorded each time a data value is encountered, grouped in fives for fast counting.
Class interval
A range of numerical values grouped together as one row in a grouped frequency table, e.g. 40–49.
Class width
The size of a class interval; for 40–49 the width is 10 (49 − 40 + 1, or upper boundary − lower boundary if using continuous boundaries).
Raw data
Data exactly as collected, before any sorting, grouping or tallying.

Grouped table quantities you're expected to calculate

Fixed once chosen — every row uses the same width unless told otherwise.

Memorise

Used to estimate an average from grouped data (exact individual values are lost once grouped).

Read straight off the raw data before grouping — grouping can only give you an estimated range.

Anatomy of a tally chart

Red = 7 shown as one group of five (four verticals + diagonal strike) plus two singles; Blue = 6; Green = 4; Yellow = 3.

ColourTallyFrequencyRedBlueGreenYellow

Worked example

Building a simple frequency table (favourite colour)

[2]
A class of 20 students chose their favourite colour: red, blue, green, red, yellow, blue, red, green, blue, red, yellow, red, blue, green, red, yellow, blue, red, green, blue. Calculate the frequency of each colour.

Worked example

Grouped frequency table with justification of class width

[4]
Test scores (out of 100) of 25 students: 45, 67, 72, 88, 55, 63, 71, 79, 84, 52, 68, 74, 91, 48, 60, 77, 85, 59, 66, 73, 80, 62, 70, 86, 93. (a) Create a grouped frequency table using class intervals of width 10, starting at 40. (b) Explain why a class width of 10 is suitable.

Worked example

Frequency table as evidence for fairness (die rolls)

[4]
A fair six-sided die is rolled 30 times: 3,5,2,6,1,4,3,2,5,1,6,2,4,3,1,5,2,6,4,3,1,5,2,6,4,3,1,5,2,6. (a) Tally the results. (b) The expected frequency for each outcome is 5. Explain whether your table supports this.

Common mistake

Reporting the frequency number when asked to 'identify' the most/least frequent category (e.g. writing '7' instead of 'Red').

Consequence: The command term 'identify' demands the category name — a numerical answer alone is marked wrong even though the reasoning behind it was correct.

Fix: Always re-read what's being asked for: a colour/category name, not the count that helped you find it.

The self-check that saves marks

Before moving to part (b) or (c) of any tally question, add the frequency column and compare it to the stated sample size. A mismatch means a tally was miscounted or double-counted somewhere — fix it before your interpretation is built on wrong numbers.

Key takeaways

  • Tally in one continuous pass through the data, grouping in fives — never count categories separately from memory.
  • Grouped frequency tables need a fixed, examiner-stated class width; boundaries must not overlap or leave gaps.
  • Always total the frequency column and compare it to the original sample size before answering later parts.
  • 'Identify' answers must be the category name, not the frequency number that helped you find it.

Bar Charts, Line Graphs and Pie Charts

These three charts each answer a different question about the same data. A bar chart compares sizes of separate categories. A line graph shows how one quantity changes over a continuous variable, usually time. A pie chart shows how a total splits into proportions — it only makes sense when the categories genuinely add up to a meaningful whole.

  • Bar chart: bars have gaps between them because categories are separate and unordered — height/length represents frequency.
  • Line graph: points are plotted and joined by straight line segments; the x-axis is usually a continuous variable like time.
  • Pie chart: each category becomes a sector whose angle is proportional to its frequency out of the total — angles must sum to .
  • Too many categories ruins a pie chart (slivers become unreadable) and ruins a bar chart less badly, but still makes comparison harder.

Pie chart angles and line graph rate of change

Angle for one sector.

— sector angle (degrees)
— frequency of that category
— total frequency (sample size)
Memorise

Used to read a trend between any two points on a line graph, not just adjacent ones.

Bar chart of favourite colour (20 students)

Bars have gaps between them because colour is a categorical variable with no natural order.

ColourFrequencyRedBlueGreenYellow

Line graph — monthly temperature trend

The segment between any two points has its own average rate of change — the graph is NOT a straight line overall.

MonthTemperature (°C)

Worked example

Reading an average rate of change from a non-adjacent segment

[2]
Using the temperature line graph above, calculate the average rate of temperature increase between March and June.

Worked example

Converting frequencies to pie chart angles

[2]
Convert the colour frequencies (Red 7, Blue 6, Green 4, Yellow 3, total 20) into pie chart sector angles.

Pie chart of favourite colour

Sector angles are proportional to frequency; total must equal 360°.

Red 126°Blue 108°Green 72°Yellow 54°

Bar chart vs line graph vs pie chart

FeatureBar ChartLine GraphPie Chart
Best forComparing sizes across separate categoriesShowing change over a continuous variable (usually time)Showing proportion of a whole
X-axis typeCategorical, unorderedContinuous, orderedNo axis — sectors of a circle
Gaps between bars/pointsYes — categories are distinctNo gap; points joined by linesNot applicable
What it can't showTrend/change over timeExact frequency comparisons at a glanceMore than a handful of categories clearly

Bar chart vs histogram — the one thing that matters

Concept AConcept BThe distinction that matters
Bar chartHistogramBar charts have gaps between bars (categorical/discrete data); histograms have no gaps because the x-axis is a continuous numerical scale — bars touch because the classes are adjacent.

Common mistake

Drawing a pie chart for data with 8+ categories (e.g. 'favourite subject' across a whole year group with 10 subjects).

Consequence: Sectors become too thin to label or compare — the 'construct' mark is lost for choosing an unsuitable representation even if the angles are calculated correctly.

Fix: Group small categories into an 'Other' sector, or switch to a bar chart, when there are more than about 6 categories.

Pie charts in election-night graphics

News broadcasts use pie charts for vote share because there's a genuine fixed total (100% of votes) split into a handful of parties. The same data over the last 5 elections is shown as a line graph instead — because now the story is the trend, not a single snapshot's proportions.

Key takeaways

  • Bar chart = compare separate categories; line graph = show change over a continuous variable; pie chart = show proportions of one whole.
  • Pie chart angles always use and must sum to exactly .
  • A line graph's average rate of change between two points uses — it is NOT the same as the gradient of any one visible segment unless those are your exact two points.
  • Too many categories breaks a pie chart before it breaks a bar chart.

Stem-and-Leaf Diagrams and Dot Plots

Both of these keep every individual data value visible — unlike a grouped frequency table, nothing is thrown away. A stem-and-leaf diagram splits each number into a 'stem' (all digits except the last) and a 'leaf' (the last digit); a dot plot stacks one dot per data value directly above its position on a number line.

  • Stem-and-leaf leaves must be written in ascending order within each stem row — an unordered stem-and-leaf loses presentation marks even with all the right digits present.
  • A key/legend is compulsory: e.g. '4 | 5 = 45' tells the reader how to decode the diagram.
  • Dot plots suit discrete numerical data with a manageable range of distinct values (shoe sizes 4–8, not test scores 0–100).
  • Both diagrams let you read off the mode, range and shape of the distribution directly — no calculation needed.
Stem
The leading digit(s) of a data value, shared by a row — e.g. the '7' in 72 when values are two-digit.
Leaf
The final digit of a data value, written next to its stem in ascending order.
Key (legend)
A short statement decoding the diagram, e.g. '4|5 = 45', without which the stem-and-leaf is ambiguous.
Dot plot
A diagram where each data value is represented by one dot stacked above its position on a number line, with taller stacks showing higher frequency.

Stem-and-leaf plot of 25 test scores

Key: 4 | 5 = 45. Leaves are written in ascending order within each row.

StemLeaf

Worked example

Constructing a stem-and-leaf plot in order

[3]
Test scores of 25 students: 45, 67, 72, 88, 55, 63, 71, 79, 84, 52, 68, 74, 91, 48, 60, 77, 85, 59, 66, 73, 80, 62, 70, 86, 93. Construct a stem-and-leaf diagram.

Dot plot of shoe sizes

Shoe size 6 is the mode, with 8 students — the tallest stack.

Shoe SizeFrequency

Reading the mode straight off

In a dot plot or stem-and-leaf, the mode is the tallest stack of dots or the row with the most leaves — you should never need to rebuild a frequency table to answer a 'state the mode' question if one of these diagrams is already given.

Key takeaways

  • Stem-and-leaf and dot plots preserve every individual data value — grouped frequency tables don't.
  • Leaves must be reordered ascending within each stem row after all data is placed — this is a common lost mark even when every digit is correct.
  • A key/legend is compulsory for stem-and-leaf plots — no key means the diagram can't be marked.
  • Dot plots only work well for discrete data with a small, manageable range of distinct values.

Choosing the Best Data Representation

This is the skill examiners specifically probe with 'justify' questions — because it's the one place a student can't just apply a memorised procedure. The choice depends on three things: what type the data is, how many categories/values exist, and what the reader actually needs to see (a single snapshot's proportions, a comparison, or a trend over time).

  • Categorical data with few categories → bar chart or pie chart.
  • Categorical data with many categories → bar chart only (pie chart becomes unreadable).
  • Discrete numerical data with a small range of distinct values → dot plot or bar chart.
  • A large, spread-out numerical dataset (e.g. 25+ test scores) → grouped frequency table, or stem-and-leaf if individual values still matter.
  • Data changing over time → line graph, always — no other chart shows a trend as clearly.

Matching data type to representation

Data descriptionBest representationWhy
Favourite colour of a class (4 categories)Bar chart or pie chartFew categories; pie chart works if proportions are the focus
Favourite subject of a year group (10+ subjects)Bar chartToo many categories for a readable pie chart
Shoe sizes of 20 students (range 4–8)Dot plotSmall range of distinct discrete values — every value stays visible
Test scores of 30 students (range 45–93)Grouped frequency table or stem-and-leafToo many distinct values for a dot plot; stem-and-leaf keeps individual scores if needed
Monthly temperature over a yearLine graphContinuous variable (time) with a trend to show

Worked example

Justifying a choice of representation

[2]
A school records the number of pets owned by each of 18 students (values range from 0 to 4). Identify the most suitable representation and justify your choice.

What 'justify' actually needs

A justify answer for choice of representation must name (1) the data type and (2) how many distinct values/categories exist, then link both to the chosen chart. Two sentences of vague praise for the chart ('it's clear and simple') score zero of the two available marks.

Common mistake

Choosing a pie chart because the data happens to have percentages already, without checking how many categories there are.

Consequence: If there are more than about 6 slices, sectors become too thin to compare — the construct/justify mark is lost even though the angle arithmetic is correct.

Fix: Count categories first. Above ~6, switch to a bar chart or merge minor categories into 'Other'.

Key takeaways

  • Choice of representation depends on data type, number of distinct values, and the question being asked of the data.
  • Justify answers need two things: the specific data property AND how it connects to the chosen chart.
  • When a range is small enough to list every value (like 5 pet counts), a dot plot beats a grouped table — nothing needs to be thrown away.

Types of Data: Discrete and Continuous

Every dataset splits first into qualitative (categorical) or quantitative (numerical). Quantitative data then splits again into discrete (countable, only certain values possible) or continuous (measured, any value within a range is possible). Getting this classification right is the gate that decides which representation is even valid — a line graph for eye colour makes no sense, and neither does a dot plot for height measured to the nearest millimetre across 1000 people.

  • Discrete data: counted in whole, separate steps — number of siblings, shoe size, number of cars in a car park.
  • Continuous data: measured, can take any value (including decimals) within a range — height, mass, temperature, time.
  • Qualitative/categorical data: describes a quality or category, not a number — eye colour, favourite subject, gender.
  • A quick test: if halfway between two consecutive values makes physical sense (e.g. 5.5 people does NOT, but 5.5°C does), that tells you discrete vs continuous instantly.
Discrete data
Numerical data that can only take specific, separate values, usually whole numbers, found by counting.
Continuous data
Numerical data that can take any value within a range, found by measuring, and limited only by the precision of the measuring instrument.
Qualitative (categorical) data
Data describing a quality, label or category rather than a numerical amount.
Quantitative data
Numerical data — the parent category that splits into discrete and continuous.

Classification tree for data types

Classify top-down: category or number first, then countable or measurable.

DataQualitative (Categorical)Quantitative (Numerical)DiscreteContinuous

Discrete vs continuous

If you COUNT it, it's Discrete. If you MEASURE it, it's Continuous.

Counting only ever gives whole, separate results (you can't count '3.5 students'); measuring can always in principle be made more precise (height can always be given to more decimal places).

Worked example

Classifying five real datasets

Classify each as qualitative, discrete numerical, or continuous numerical: (i) shoe size, (ii) height in cm, (iii) number of siblings, (iv) eye colour, (v) time to run 100m.

The classification traps

Concept AConcept BThe distinction that matters
Discrete dataContinuous dataDiscrete = counted, only specific values possible (whole numbers or fixed steps); continuous = measured, any value within a range is possible.
Qualitative dataQuantitative dataQualitative describes a category/label with no numerical size; quantitative is a genuine number that can be added, averaged or ordered.

Sensors vs counters

A traffic sensor counting cars passing a junction each hour produces discrete data — you can't have 47.5 cars. The same junction's temperature sensor logging degrees Celsius every minute produces continuous data — the reading could genuinely fall anywhere between 17°C and 18°C.

Key takeaways

  • Classify top-down: category or number first (qualitative vs quantitative), then countable or measurable (discrete vs continuous).
  • Discrete = counted in fixed steps; continuous = measured, can take any value within a range.
  • The data type you identify directly restricts which chart is valid — get this step wrong and every later choice is built on sand.

Designing and Conducting Surveys

A survey is only as good as its design. Two things can quietly wreck an otherwise well-analysed dataset: an unrepresentative sample, or questions that push respondents toward a particular answer. Both are frequently tested through 'analyse the limitation' questions, and both need a stated consequence, not just a name.

  • Population is everyone you're actually interested in (e.g. all students in the school); a sample is the smaller group you actually ask.
  • A representative sample should reflect the population's mix (age, year group, gender etc.) — surveying only your friend group is not representative even if 100 people answer.
  • A leading (loaded) question nudges respondents toward a particular answer by embedding an assumption or emotive language.
  • Closed questions (with fixed response options) are easier to turn into a frequency table; open questions capture more detail but are harder to tally.
Population
The entire group a survey is ultimately interested in describing.
Sample
The subset of the population actually surveyed, used to draw conclusions about the whole population.
Bias
A systematic distortion in the results caused by how the sample was chosen or how questions were worded, rather than by genuine variation in opinion.
Leading (loaded) question
A question worded to push respondents toward a particular answer, e.g. through an embedded assumption or emotive phrasing.

The survey design process

Skipping straight from 'define aim' to 'design questions' without deciding the sample is where representativeness problems start.

Define aimChoose sampleDesign questionsCollect dataAnalyse results

Worked example

Spotting and fixing a leading question

[3]
A student surveys the canteen menu and asks: 'Don't you think the canteen should serve more pizza, since everyone loves it?' Identify what is wrong with this question and rewrite it.

Common mistake

Surveying only a friend group (e.g. 15 people from one class) and generalising the conclusion to 'all students in the school'.

Consequence: The sample isn't representative of the full population (different year groups, interests etc. aren't included), so any 'analyse the limitation' question expects this exact point, with the consequence that conclusions can't be generalised.

Fix: State explicitly how the sample will be chosen to reflect the population — e.g. a fixed number of students randomly selected from each year group.

Naming your sampling method explicitly

In a 'design a survey' task, always state who you'll ask, how many, and how they're chosen (e.g. 'random sample of 10 students from each of the 6 year groups, selected using a random number generator'). A list of questions with no stated sampling method loses the methodology marks even if the questions themselves are perfectly worded.

Key takeaways

  • Population = who you ultimately want to know about; sample = who you actually ask — the sample must represent the population's mix.
  • A leading question embeds an assumption or emotive language that pushes respondents toward one answer — always name the mechanism, not just the label 'biased'.
  • 'Analyse a limitation' answers need the flaw AND its consequence for the validity of the conclusion.
  • Closed, neutrally worded questions with fixed response options convert directly into frequency tables — that's why they're preferred for numerical analysis.

Quick Reference

Data Handling and Representation — at a glance

Formulas

  • Pie chart sector angle — all sectors must sum to 360°.
  • Kept constant across all rows of a grouped frequency table.
  • Used to estimate averages from grouped data.
  • Read from raw data, not from grouped classes, whenever possible.
  • Between any two points on a line graph, not just adjacent ones.

Key facts

Pie chart total angle
360°
Tally grouping size
5 (4 verticals + 1 diagonal strike)
Bar chart bars
always have gaps (categorical data)
Histogram bars
never have gaps (continuous data)

Mnemonics

  • COUNT it, it's Discrete. MEASURE it, it's Continuous.The fastest way to classify numerical data under time pressure.

Don't confuse

  • Bar chart vs Histogram: Bar chart = gaps between bars, categorical/discrete data; histogram = no gaps, continuous numerical scale.
  • Discrete data vs Continuous data: Discrete = counted, fixed possible values; continuous = measured, any value in a range.
  • Population vs Sample: Population = everyone you want to know about; sample = the smaller group you actually survey.

Command terms

Construct:
Build the full chart/table with correct data AND presentation (title, labels, scale).
Complete:
Fill in missing values exactly matching the raw data — check totals.
Identify:
Name the category/value only — no explanation, but must be the exact name not a number.
Explain:
State a reason with a clear cause-and-effect link — never a bare assertion.
Interpret:
Draw a conclusion using actual figures read from the table/chart.
Analyse:
Name a limitation AND state its effect on the validity of the conclusion.
Justify:
Name a specific data property and connect it directly to the choice made.

Exam checklist

  • Frequency column totals match the original sample size.
  • Tally marks grouped in fives, counted in one continuous pass through the data.
  • Pie chart angles sum to exactly 360° (adjust rounding on the largest sector if needed).
  • Stem-and-leaf leaves reordered ascending within each row, key/legend included.
  • Data type identified (qualitative / discrete / continuous) before choosing a representation.
  • Sample and sampling method explicitly stated in any survey design answer.
  • Justify answers name a data property, not just 'it looks clearer'.