Toolkit

Descriptive statistics

Summary measures such as the mean, median, mode and range, and charts, used to describe and compare business data.

SL and HL Verification: Verified In the IB toolkit Updated 3 November 2026

Section 01: What it is for

To summarize business data so that it can be understood, compared and used in decisions: the centre of the data with the mean, median and mode, its spread with the quartiles and the standard deviation, and its shape with a well-chosen chart.

Class
Situational (the platform's reading; the TSM does not classify this tool)
Suggested units
Unit 2, Unit 3, Unit 4, Unit 5 (guide, p. 14)
The guide includes
Mean; Mode; Median; Bar charts; Pie charts; Infographics; Quartiles; Standard deviation

When to use it

  • When describing data such as sales, survey results, absence, labour turnover or waiting times.
  • When comparing groups, such as branches, products or periods.
  • When choosing a figure to plan with, such as the demand in a typical week.

How to build it

  1. Decide the question the data must answer, and check their source, sample and reliability.
  2. Put the values in order, then find the mean (the total divided by the number of values), the median (the middle value) and the mode (the most frequent value).
  3. Find the range, the quartiles and the interquartile range, and the standard deviation where the variation matters.
  4. Choose a chart that fits the data: a bar chart to compare categories, a pie chart for the parts of one whole, an infographic to communicate a few key figures.
  5. Interpret the figures in context: which average represents the data, what the spread means, and which values are outliers.

The evidence it needs

  • A data set relevant to the question, with its source and how it was collected.
  • Enough values for the measures to mean something.

Section 02: What it shows, and what it hides

It reveals
  • What is typical, and how much the data vary around it.
  • Outliers and differences between groups that deserve investigation.
  • Patterns that a chart makes visible at a glance.
It does not reveal
  • Why the data look as they do: causes need further evidence.
  • Whether a sample represents the whole population, unless it was chosen soundly.

Limitations

  • A single summary can hide important detail, and the wrong average can mislead.
  • Small samples and biased collection make the measures unreliable.
  • Charts can exaggerate or hide differences through their scales and their design.

Section 03: Applied: Labour turnover across Silvertree's clinics

From the case Silvertree Clinics.

Source: Simulated data

Exhibit 3 gives labour turnover in 2025 for Silvertree's six clinics: Observatory 16.7%, Bellville 43.8%, Mitchells Plain 35.3%, Wynberg 33.3%, Durbanville 28.6% and Sea Point 31.3%. The mean of the six rates is 31.5% and their median is 32.3%. The range is 27.1 percentage points, the interquartile range 6.7 points (from 28.6% to 35.3%) and the standard deviation about 8.1 points, treating the six clinics as the whole set. Across the whole network, 30 of 96 nurses left, a rate of 31.3%.

What it showed

Five clinics lie between 28.6% and 43.8%, and one, Observatory, stands well below them at 16.7%. Without it the mean of the rates would be 34.5%, so the problem is general and Observatory is the exception worth studying, which is why the case looks at how it manages its nurses (Exhibit 8). The staff survey adds a second lesson (Exhibit 4): after the pay rise, the share of nurses who think their pay is fair rose from 31% to 52%, while the share intending to stay fell from 58% to 55%, so pay was not what was driving them away.

Read the case file: Silvertree Clinics

Section 04: In the exam

Descriptive statistics are a toolkit tool for units 2 to 5 and can support a question on any data in a case, such as survey results, sales or labour turnover. Calculate the measures accurately, choose the average that fits the data, and interpret the result in context; a figure without an interpretation earns little.

Common errors

  • Using the mean when an outlier distorts it, where the median would represent the data better.
  • Forgetting to put the data in order before finding the median or the quartiles.
  • Treating a sample, such as a survey of the people willing to answer, as if it described everyone.