Statistics and probability is the largest topic in IB Maths AI SL: the current guide suggests 36 of the course's 150 hours for it. Two parts of that topic carry a lot of exam weight. Descriptive statistics asks you to summarise a data set (centre, spread, outliers). The normal distribution asks you to turn a mean and a standard deviation into probabilities and back again. This guide treats them together, because exam questions often do.
You will find four worked examples, the GDC habits that protect marks, and a self-check set at the end.
Where this sits in the course
Both topics belong to Topic 4, Statistics and probability, and both AI SL papers require a GDC. That shapes the questions: you are rarely asked to grind numbers, and often asked to choose the right statistic, read the calculator output correctly and say what the result means. If your child wants weekly practice with a tutor who knows these question styles, see our AI SL maths tuition.
The descriptive side covers mean, median, mode, quartiles, range, interquartile range (IQR), standard deviation and variance, plus box plots, histograms and cumulative frequency. The normal side covers the properties of the normal curve, normal probabilities and inverse normal calculations. Our broader probability distributions guide compares the normal with the binomial if you need to decide between them.
Descriptive statistics: what each measure tells you
| Measure | Describes | Affected by an outlier? |
|---|---|---|
| Mean | Centre, using every value | Yes, strongly |
| Median | Middle value | Hardly |
| Range | Total spread | Yes, strongly |
| IQR = Q3 − Q1 | Spread of the middle half | Hardly |
| Standard deviation | Typical distance from the mean | Yes |
An outlier is a value more than 1.5 × IQR below Q1 or above Q3. Use that test; "it looks far away" does not earn the mark.
Worked example 1: commute times with an outlier
Twelve students record their commute, in minutes:
12, 15, 15, 18, 20, 21, 22, 24, 25, 27, 30, 52
(a) Find the median, quartiles and IQR.
With 12 values, the median is the mean of the 6th and 7th: (21 + 22)/2 = 21.5.
Lower half 12, 15, 15, 18, 20, 21 gives Q1 = (15 + 18)/2 = 16.5.
Upper half 22, 24, 25, 27, 30, 52 gives Q3 = (25 + 27)/2 = 26.
IQR = 26 − 16.5 = 9.5.
(b) Is 52 an outlier?
Upper fence = 26 + 1.5 × 9.5 = 40.25. Since 52 > 40.25, yes. Lower fence = 16.5 − 14.25 = 2.25, so there are no low outliers.
(c) Mean and standard deviation, with and without 52.
With it: mean = 23.4, standard deviation = 9.99 (GDC, σx).
Without it: mean = 20.8, standard deviation = 5.27, median = 21.
One value moved the mean by more than 2 minutes and almost halved the spread when removed. The median barely changed. That is the comment examiners want when they ask which average is "more appropriate".
Worked example 2: grouped data
Fifty students time a puzzle.
| Time t (minutes) | 0 ≤ t < 10 | 10 ≤ t < 20 | 20 ≤ t < 30 | 30 ≤ t < 40 | 40 ≤ t < 50 |
|---|---|---|---|---|---|
| Frequency | 6 | 14 | 19 | 8 | 3 |
Use the mid-interval values 5, 15, 25, 35 and 45 as the data, with the frequencies as weights (enter both lists in the GDC).
- Estimated mean = 22.6 minutes; estimated standard deviation = 10.5 minutes.
- Modal class: 20 ≤ t < 30.
- The median is the 25.5th value. Cumulative frequencies are 6, 20, 39…, so it lies in 20 ≤ t < 30.
Write "estimate". The raw times are lost once data is grouped, so the mean cannot be exact.
Changing every value by a constant
Room temperatures have mean 22 °C and standard deviation 3 °C. Converting with F = 1.8C + 32:
- the mean becomes 1.8 × 22 + 32 = 71.6 °F;
- the standard deviation becomes 1.8 × 3 = 5.4 °F.
Adding a constant shifts the centre but not the spread. Multiplying scales both. The +32 never touches the standard deviation.
The normal distribution
A normal variable X ~ N(μ, σ²) has a symmetric bell curve centred on μ. About 68% of values lie within 1 standard deviation of the mean, about 95% within 2 and about 99.7% within 3. Note that the second number in N(μ, σ²) is the variance. The GDC asks for σ.
Worked example 3: battery life
The battery life of a phone model is X ~ N(42, 3.5²) hours.
(a) Find P(X < 36). Normal cdf with lower bound a very large negative number (for example −10^99), upper 36, μ = 42, σ = 3.5: 0.0432.
(b) Find P(40 < X < 48). Lower 40, upper 48: 0.673.
(c) A shop sells 500 of these phones. How many batteries would you expect to last between 40 and 48 hours? 500 × 0.673 = about 336.
Sketch a curve and shade the region each time. It takes five seconds and catches answers that are obviously the wrong side of the mean.
Worked example 4: inverse normal
Same battery, X ~ N(42, 3.5²).
(a) The manufacturer wants to state a life that 90% of batteries exceed. Find it.
P(X > k) = 0.9 means P(X < k) = 0.1. Inverse normal with area 0.1 to the left: k = 37.5 hours. On newer calculators you can instead choose the right tail and enter 0.9 directly; either way, check that k is below the mean, as it must be.
(b) Find the battery life that only the longest-lasting 5% reach.
P(X < k) = 0.95, giving k = 47.8 hours.
Is the normal model sensible?
Check the context before trusting it. Normal is reasonable for measurements that cluster symmetrically around a typical value: heights, masses, filling machines, timings. It is a poor model for data that is clearly skewed, such as income, or bounded at zero with lots of values near zero. If a question asks whether a normal model is appropriate, compare the mean and median, look for symmetry in a histogram or box plot, and use the 68–95–99.7 pattern as a rough check.
Mistakes our tutors see
- Entering the variance instead of σ. In N(42, 3.5²), σ = 3.5, not 12.25.
- Using the wrong standard deviation output. For a data set in AI SL, use the GDC's σx, not sx.
- Forgetting to flip the tail. P(X > k) = 0.9 means 0.1 to the left.
- "Outlier" without the 1.5 × IQR test. Show the fence calculation.
- Exact language for estimates. Grouped data gives an estimated mean.
- No context. "The median is more appropriate because 52 is an outlier that raises the mean" beats "the median".
Exam technique
- Write the distribution and the probability statement before the number: "X ~ N(42, 3.5²), P(40 < X < 48) = 0.673".
- Answers to 3 significant figures, but keep full values for later parts.
- For expected numbers, multiply and round sensibly; you cannot have 336.45 batteries.
- Box plot questions often ask you to compare two groups. Comment on a centre (median) and a spread (IQR) in context.
Self-check
- For the data 3, 7, 8, 8, 10, 12, 15, 30, find Q1, Q3 and decide whether 30 is an outlier. Answer: Q1 = 7.5, Q3 = 13.5, IQR = 6; 30 > 22.5, so yes.
- X ~ N(65, 8²). Find P(X > 75). Answer: 0.106.
- For the same X, find k such that P(X < k) = 0.2. Answer: k = 58.3.
- A data set has mean 40 and standard deviation 6. Every value is doubled and then 5 is added. Find the new mean and standard deviation. Answer: 85 and 12.
Where this leads next
Descriptive statistics feeds straight into bivariate data, covered in our sibling post on regression and Spearman's rank, and into the hypothesis tests in AI SL chi-squared and t-tests. Probability from Venn and tree diagrams is in our AI SL probability guide. At the next level, AI HL adds further distributions, including the Poisson distribution.
How Maths Bodhi helps
Our tutors begin with a short diagnostic in the free demo: one data set to summarise and one normal question to set up, so we can see whether the gap is concepts, calculator use or written interpretation. Each student is matched to a tutor who teaches AI SL, and practice moves from single skills to full past-paper style questions marked step by step. Families who prefer a statistics specialist can also look at our statistics tutors.
Home tuition is available in Gurgaon (Gurugram). Students elsewhere in India and abroad learn with us online. You can book a free demo class.
Written and checked by the Maths Bodhi team. Facts checked against official IB sources, November 2026.
Frequently asked questions
- Is the normal distribution in IB Maths AI SL?
- Yes. AI SL covers the properties of the normal curve, normal probability calculations and inverse normal calculations, all done on the GDC.
- How do you find an outlier in IB maths?
- Find Q1, Q3 and the IQR. A value more than 1.5 × IQR below Q1 or above Q3 is an outlier. Show the fence calculation in your answer.
- Which standard deviation should I use on my calculator?
- For describing a data set in AI SL, use the population standard deviation, which most GDCs label σx.
- How do I do an inverse normal question on the GDC?
- Write the probability as an area to the left, for example P(X > k) = 0.9 becomes P(X < k) = 0.1, then use inverse normal with that area, the mean and σ.
- What happens to the mean and standard deviation if every value is multiplied by a constant?
- Both are multiplied by that constant. Adding a constant changes the mean but leaves the standard deviation unchanged.
- What are typical IB Math AI SL descriptive statistics questions?
- Finding quartiles and testing for outliers, estimating a mean from grouped data, comparing two groups with box plots, and choosing between mean and median in context.
