Data Handling · Cambridge syllabus 9.2 · Lesson 1 of 1
Statistical Reasoning
Choose displays to suit data: category counts use bar charts, continuous grouped data use bars that touch (histograms, at Extended level), paired values use scatter diagrams. Compare centre and spread, consider what the data can and cannot show, and check whether truncated axes exaggerate differences.
By the end, you can
- Select appropriate statistical diagrams.
- Compare distributions using suitable statistics.
- Recognise misleading representations.
- Explain conclusions using evidence from data.
Data interpretation
Interpreting data means saying what the figures mean for the situation, not restating them.
A full comparison has two parts, an average and a measure of spread, and both stated in context: "the second group has a higher mean, so they scored better on average, and a smaller range, so they were more consistent". At Extended level the median and interquartile range are often used instead.
Choose the measures to suit the data. With outliers, the median describes it better than the mean, and the interquartile range (Extended) better than the range, and saying why is often the question.
Choose the right diagram for the data: bar charts, pictograms and pie charts for categories, stem-and-leaf diagrams for small sets of numbers, scatter diagrams for pairs of values, and, at Extended level, histograms and cumulative frequency curves for grouped continuous data. Be alert to misleading presentation: a vertical scale that does not start at zero exaggerates differences, and a pie chart shows proportions but not amounts, so two charts can look identical for groups of very different sizes.
Be careful about what data can show. A correlation is not a cause. A small or unrepresentative set of data does not support a general conclusion. And extending a line of best fit beyond the data assumes a pattern that may not continue.
Worth remembering
- Compare an average and a spread, both in context.
- Correlation is not causation, and a small or unrepresentative data set proves nothing.
- Check the scale starts at zero before trusting a comparison.
Worked example
Two groups can share a median but differ greatly in range, or at Extended level in interquartile range; report both an average and a spread to describe them.
Practice
Warm-up · not marked
Check your understanding
Answer in your own words, then compare with the worked answer. Nothing here is saved or scored.
Worked answer
The visible bars are 10 and 20 units high, making a 10% increase look like a doubling.
Check your method, units and explanation against it.
Marked practice
Set 1 of 5 · easiest first · Checked as you go, with the worked answer. Sign in or create a free account to save your progress.