IB Middle Years Programme · Mathematics MYP Year 3

Statistics – Averages and Analysis

Cover illustration for Statistics – Averages and Analysis (Mathematics MYP Year 3).
MYP · Mathematics MYP Year 3

Statistics – Averages and Analysis

22 min readStandardAssessed under Criterion A (Knowing & Understanding) and Criterion D (Applying Maths in Real-Life Contexts); appears in most unit tests and end-of-year eAssessment statistics questions.

Averages look simple — add up, divide, done — until an exam asks you to defend WHICH average you chose, spot what a grouped table is hiding, or reverse-engineer a missing value from a given mean. This topic is less about fresh calculation (you've done mean/median/mode since Year 1) and more about interpretation: reading a histogram's shape, catching a misleading class interval, and comparing two data sets properly using both centre AND spread together.

Overview

Every statistics question in this chapter does one of three things: it organises raw data (grouping), it works backwards from a summary number (missing data), or it asks you to judge whether a summary is fair (misuse and interpretation). The calculations themselves — mean, median, mode, range, quartiles — are tools; the marks live in the reasoning around them.

  • Grouped data trades precision for a readable shape — but every calculation from a grouped table afterwards is an estimate.
  • Box plots and IQR summarise spread using position (quartiles), so data must always be sorted first.
  • Comparing two data sets properly means quoting BOTH a measure of centre (mean/median) AND a measure of spread (range/IQR) in the same sentence.

The shape of the chapter

Command terms that dominate this topic

Command termWhat it demandsAOMark-earning move
CalculateObtain a numerical answer showing all relevant workingCriterion AWorking line AND final value both needed — the answer alone is often capped at 1 mark.
DescribeGive a detailed account in wordsCriterion A/DNeeds a full sentence linking to the data context, not just a bare number.
ExplainGive reasons or a mechanismCriterion DMust state WHY, linking cause to effect — restating the question scores 0.
AnalyseBreak down data to find patterns or relationshipsCriterion DRequires interpretation beyond the raw figures — e.g. what a distribution's shape implies about performance.
ProposePut forward a suggestion with justificationCriterion DA number alone is not enough — the justifying sentence carries roughly half the marks.
CompareIdentify similarities and differencesCriterion DMust reference both data sets explicitly in one linking sentence, not describe them separately.

Key point

Every time you group data or quote a single 'average', you're deciding what the reader is and isn't allowed to see. Question-setters test whether you noticed — always ask what got hidden.

Working with Grouped Data

Raw data — every individual score, mass, or age — is precise but unreadable in bulk. Q2's sports club records 200 ages between 6 and 18; nobody can picture 200 separate numbers at once. Grouping sorts values into class intervals so the shape of the data becomes visible at a glance — but the instant you group, the exact original values are gone forever. Every calculation from a grouped table after that point is an estimate.

  • Class width = the gap between the start of one interval and the start of the next, e.g. Table A's 0–10, 10–20, 20–30 all have width 10; Table C's 2–7, 7–12, 12–17 all have width 5.
  • All class intervals in one table should share the same width — otherwise histogram bars can't be compared fairly against each other.
  • Continuous data (height, mass, time) uses non-overlapping inequality boundaries: . Discrete whole-number data (test scores, ages in whole years) can use inclusive integer intervals like , with the next interval starting exactly at 20.
  • Choosing width is a trade-off: one giant interval (Q2's single 6–18 group) hides the shape entirely; too many narrow intervals just reproduce the raw list and lose the benefit of grouping.
Class interval
A range of values that groups individual data points together, e.g. 40–60.Written with a dash or inequality — always check whether boundaries overlap.
Class width
The size of an interval — the gap between the start of consecutive intervals.
Midpoint
The middle value of a class interval, used to represent every value inside it.
Frequency
The count of data values that fall inside a given class interval.

Midpoint of a class interval.

— smallest value the interval starts at
— largest value the interval ends at
Memorise

Estimated mean of grouped data.

— frequency of an interval
— midpoint of that interval
Memorise

Always an ESTIMATE, never the exact mean — the actual values inside each interval are unknown.

Histogram of Test Scores (30 students)

The tallest bar (40–60) shows most students scored in the middle range — but a histogram only tells you the COUNT per interval, never the exact scores inside it.

0-2020-4040-6060-8080-100Frequency axis

Worked example

Estimating total mass from a grouped frequency table

[6]
The masses (kg) of 40 Grade 8 students are grouped: 30–34 (4), 35–39 (8), 40–44 (12), 45–49 (9), 50–54 (5), 55–59 (2). (a) Find the midpoint of each interval. (b) Estimate the total mass of all 40 students. (c) A student claims the estimated total would be exactly the same as the real total mass. Comment.

Reading interval notation: discrete vs continuous

Data typeExample intervalsInequality meaningWhere a boundary value goes
Continuous (height, mass, time)140–149, 150–159A value of exactly 150 goes into the NEXT interval
Discrete whole numbers (scores, ages in years)10–19, 20–29A value of exactly 20 starts the next interval — no overlap possible

Common mistake

Using the lower bound (or upper bound) of an interval instead of the midpoint when estimating a grouped mean or total.

Consequence: Every subsequent calculation is shifted systematically — examiners award zero for the final answer even if the method (multiply by frequency, sum, divide) is otherwise correct, because the wrong representative value was substituted.

Fix: Always write the midpoint calculation as its own line — — before multiplying by frequency, so the method mark is visibly locked in even if a later arithmetic slip occurs.

Height of a bar ≠ 'the value'

When asked to explain what a histogram bar's height represents (Q1b), the mark scheme wants the word 'frequency' explicitly — 'the number of students whose score falls in that interval.' Saying 'it shows the scores' or 'it shows how many' without naming frequency loses the mark for vagueness.

Key takeaways

  • Grouping trades detail for readability — every value inside an interval is treated as if it equals the midpoint.
  • Class width must stay constant across a table for fair, comparable histogram bars.
  • Any total or mean calculated from grouped data is an ESTIMATE — say so explicitly if asked to comment.

Finding Missing Data with Given Averages

Averages compress a whole data set into one number — which means the process can run backwards. If you know the mean and every value except one, that missing value is forced: it's whatever number makes the total add up correctly.

  • The key relationship: — rearrange this before touching the missing value.
  • Always find the sum of the KNOWN values first, then subtract from the total required sum to isolate the missing one.
  • The same logic works for a missing frequency in a grouped table, or a missing raw score in a list — the engine is identical.

Total sum of all data values equals the mean multiplied by the number of values.

— sum of all data values
— the mean
— total number of data values
Memorise

Worked example

Finding a missing score from a given mean

[3]
The mean of five test scores is 14. Four of the scores are 10, 12, 15 and 18. Find the fifth score.

Common mistake

Multiplying the mean by the number of known values instead of the total number of values (including the unknown).

Consequence: The 'total sum' calculated is wrong from the first line, so every mark afterward is lost even with perfect subtraction technique.

Fix: Before calculating anything, state explicitly and count it directly from the question — 'there are 5 scores in total, so ' — as your very first line of working.

Missing frequency works the same way

If a grouped table gives a mean and asks for a missing frequency, set up exactly as above — just with frequency × midpoint terms instead of raw values. The reverse-engineering logic doesn't change.

Key takeaways

  • Rearrange mean = sum ÷ n into sum = mean × n before doing anything else.
  • Always subtract the sum of KNOWN values from the TOTAL required sum — never the reverse.
  • Double-check n counts every value, including the one you're solving for.

Box Plots and Interquartile Range

A box plot compresses an entire ordered data set into five landmark numbers — minimum, lower quartile, median, upper quartile, maximum — so spread and skew are visible in one glance without reading every value.

  • Order the data first — every quartile calculation depends on position, and position only means something once the list is sorted smallest to largest.
  • The median splits the data in half; is the median of the LOWER half, is the median of the UPPER half (excluding the overall median itself when is odd).
  • The box (from to ) contains the middle 50% of the data — a wide box means the middle half is spread out; a narrow box means it's tightly clustered.
  • Whisker length shows the range of the outer 25% on each side, but says nothing about how many points actually sit near the ends.

Interquartile range — the spread of the middle 50% of the data.

Memorise

Box plot from a five-number summary

The box spans the middle 50% of the data (18 to 30); the median line at 24 sits off-centre within the box, showing a slightly longer tail toward the lower values.

Worked example

Constructing the five-number summary and IQR

[5]
Find the five-number summary and interquartile range for: 12, 15, 18, 20, 22, 24, 25, 28, 30, 35, 38.

Frequently confused

Concept AConcept BThe distinction that matters
RangeInterquartile range (IQR)Range uses the two most extreme values (max − min), so one outlier can inflate it hugely; IQR only measures the middle 50% and ignores outliers entirely.

Key takeaways

  • Sort data before finding any quartile — position-based statistics are meaningless on unordered data.
  • For odd , exclude the overall median from both halves before splitting into lower/upper for /.
  • IQR = measures spread while ignoring extreme outliers — unlike range.

Statistical Misuse and Real-World Applications

Every grouping decision, every choice of 'average', and every axis scale is a decision someone made — and each decision can hide as much as it reveals. Q2's sports club crammed 200 students aged 6 to 18 into a single class interval: technically 'grouped', but useless, because it can't show whether the club is mostly young children or mostly teenagers.

  • A single wide interval erases the SHAPE of a distribution — no clusters, no gaps, no sense of where most data actually sits.
  • The mean is dragged by extreme values; the median is not. Reporting 'the average' without saying WHICH average can mislead deliberately or accidentally.
  • Truncated or non-zero-start axes on a bar chart exaggerate small differences — a bar chart starting at 90 instead of 0 can make a 2-point gap look enormous.
  • Small sample sizes reported as if representative ('3 out of 4 students prefer...') hide how few people were actually asked.

The 'average salary' trick

A company advertises an 'average salary of $85,000'. That figure is the MEAN, dragged upward by the CEO's $2,000,000 salary. The MEDIAN employee — the one exactly in the middle of the pay scale — actually earns $42,000. Both numbers are technically 'the average'; only one honestly represents a typical worker. Journalists and companies pick whichever average tells the story they want.

Worked example

Proposing better class intervals

[5]
200 students' ages (6–18 years) are grouped into a single interval 6–18. (a) Describe what information is hidden. (b) Propose a set of class intervals and explain how they improve understanding.

Common mistake

Assuming the mean is always the correct average to quote, regardless of the data's shape.

Consequence: In skewed data or data with outliers, quoting only the mean produces an 'analyse'/'compare' response that examiners mark as incomplete — full marks usually require acknowledging which average is misleading and why.

Fix: Before quoting any single average, check mentally for extreme values. If present, state both mean AND median, and comment on which better represents a 'typical' value.

'Analyse' wants interpretation, not just numbers

Q1c asks students to analyse what a distribution indicates about overall performance — the mark goes to a sentence like 'most students scored in the 40–60 range, suggesting the test was moderately challenging, with few students excelling or failing outright.' Restating the frequency table in words earns 0 — a conclusion about performance must be drawn.

Key takeaways

  • A single wide class interval or a cherry-picked average can be technically 'correct' while still misleading — always check what's been left out.
  • Mean is sensitive to outliers; median is not — quoting only one without checking the other risks a misleading impression.
  • Axis and interval choices distort visual comparison even when the underlying numbers are accurate.

Calculating Mean, Median, Mode and Range

These four numbers are the entry point to every statistics question — and each answers a different question about the same data set.

  • Mean: add everything, divide by how many values there are — sensitive to every single value, including extreme ones.
  • Median: sort the data, find the middle value (or average the two middle values if is even) — unaffected by how extreme the outliers are, only by their position.
  • Mode: the value that appears most often — the only measure that also works on non-numeric (categorical) data, and a data set can have no mode, one mode, or several.
  • Range: largest value minus smallest value — the crudest spread measure, entirely determined by just two data points.

Mean — sum of all values divided by the number of values.

Memorise

Range — largest value minus smallest value.

Memorise

For odd this is a whole-number position; for even , average the two values either side.

Worked example

Mean, median, mode and range from a raw data set

[4]
Find the mean, median, mode and range of: 12, 15, 18, 22, 25, 28, 31, 35.

Mnemonic

Mean = Measure it (add & divide); Median = Middle; Mode = Most common.

All three start with 'M' but the second word tells you exactly what to do — measure (calculate), find the middle, or spot the most frequent value.

Common mistake

Averaging the two middle values for the median even when is odd, or picking only one middle value when is even.

Consequence: Wrong median value — loses the answer mark even if every other calculation (mean, mode, range) is correct.

Fix: Always state and whether it's odd or even as your first line of median working — odd means ONE middle value, even means average of TWO.

Key takeaways

  • Mean uses every value; median uses position; mode uses frequency; range uses only the two extremes.
  • Sort the data before finding median — position only makes sense once ordered.
  • A data set can have zero, one, or multiple modes — don't force an answer.

Comparing and Interpreting Averages

Calculating an average is only half the question — assessment increasingly asks WHICH average best represents a data set, and requires comparing two data sets using both a measure of centre and a measure of spread together.

  • Outliers drag the mean toward them but barely move the median — always check for outliers before deciding which average to quote as 'typical'.
  • Two data sets can share an identical mean but have very different spreads — comparing means alone hides how consistent or variable each set is.
  • IQR or range should always accompany a mean/median comparison — 'Class A averaged higher AND was more consistent' is a complete comparison; 'Class A averaged higher' alone is not.
  • Skewed data (a long tail of high or low values) pulls the mean away from where most of the data actually sits — the median stays closer to the 'typical' value.

Choosing the right average

MeasureAffected by outliers?Best used when...Ignores
MeanYes — heavilyData is roughly symmetric with no extreme valuesNothing — uses every value
MedianNoData is skewed or has outliers (e.g. incomes, house prices)The actual size of extreme values
ModeNoData is categorical, or you need the single most common valueEverything except frequency

Worked example

Comparing two classes using average AND spread

[3]
Class A test scores: mean 65, range 10. Class B test scores: mean 65, range 40. Compare the performance of the two classes.

'Compare' always needs both data sets named explicitly

Writing 'the mean is higher' without saying higher than WHICH class, or discussing Class A and Class B in two disconnected sentences instead of one comparative sentence, typically loses the top mark in a 'compare' question.

Key takeaways

  • Never compare data sets using only the mean — always pair it with a spread measure (range or IQR).
  • Outliers and skew push the mean away from 'typical'; the median resists this.
  • A complete comparison names both data sets and states centre AND spread together in one interpretive sentence.

Quick Reference

Formulas

  • Represents every value inside a grouped interval
  • Estimated mean from grouped data — always an estimate
  • Rearranged mean formula — used to find missing values
  • Spread of the middle 50%, ignores outliers
  • Crude spread measure, uses only two values

Key facts

Median position for odd n
(n+1)/2-th value
IQR formula
Q3 − Q1
Grouped mean is always...
an estimate, never exact

Mnemonics

  • Mean = Measure it; Median = Middle; Mode = Most commonQuick recall of what each M-word actually calculates

Don't confuse

  • Range vs IQR: Range uses extremes (sensitive to outliers); IQR uses the middle 50% (resistant to outliers).
  • Mean vs Median: Mean uses every value (outlier-sensitive); median uses position only (outlier-resistant).

Command terms

Calculate:
Show the working line, not just the final value
Explain:
State WHY, don't just restate the fact
Propose:
Give the specific numbers AND justify them
Compare:
Name both data sets in one linking sentence, covering centre and spread

Exam checklist

  • Sorted the data before finding median/quartiles?
  • Used midpoint, not lower/upper bound, for grouped estimates?
  • Stated n correctly (including any missing value) before using sum = mean × n?
  • Paired every mean/median comparison with a spread measure?
  • Named 'frequency' explicitly when describing histogram bar height?