Module 4/4 26 min
Being honest with data, and the final test
Step 1 / 2·Four ways to mislead without lying
0/2 steps done
Learning objectives — after this module you will:
- Recognise presentation choices that mislead
- State the limits of your data properly
- Complete the final test
Learn ~12 min
Four ways to mislead without lying
Never do it accidentally, and notice when others do.
Four techniques and their effect
| The choice | Effect on the viewer |
|---|---|
| A bar chart whose vertical axis does not start at zero | A 2% difference looks like a doubling |
| Choosing a convenient time window | Cutting off an unfavourable period reverses the trend entirely |
| Using an average on a skewed distribution | An average of 8 million hides that half the group is under 5 million |
| Presenting correlation as causation | The room concludes A causes B when the data only shows they move together |
The fourth is the commonest and the hardest to spot
'Departments with high engagement scores have low attrition' — engagement may retain people, or people about to leave may score low, or both may stem from a third cause, management quality. Say 'these two move together' rather than 'this one reduces that one'.
State the limits up front, do not wait to be asked
'These figures exclude seasonal workers' at the start of a presentation increases credibility. The same sentence, when someone else spots it, destroys the credibility of the whole thing.