Comet rushes in waving a new chart. "Huge news! Most members want longer evening hours!"
Her chart shows the Q3 answers from the members who usually visit in the afternoon, when the crew is there. Five of their 9 valid answers chose longer evening hours.
Wren reads the chart's title. Then he reads the whole Q3 table.
"What do you notice about who is in your chart?" he asks.
Comet blinks. "Only the afternoon members. I filtered the sheet."
Nova hovers beside the two tables. "Would you like a hint?" she asks.
"A claim about most members needs data about all the members who answered. Compare the slice with the whole."
Filtering keeps only part of a data set, such as only the afternoon members.
Filtering is a useful way to get information from data. But a claim about everyone needs data about everyone who answered.
Choosing only the slice of data that fits a claim, and leaving out the rest, is called cherry-picking.
| Q3 answer | Afternoon members only | Everyone who answered |
|---|---|---|
| Longer evening hours | 5 | 13 |
| Quiet study room | 2 | 11 |
| Repair bench | 1 | 9 |
| Bike rack | 1 | 5 |
| Total valid | 9 | 38 |
Made-up data from the Commons Count story.
Often, a single source does not hold all the data needed for a conclusion.
You may need to combine data from more than one source.
The survey heard from 40 of the Commons' 120 members. The crew's sign-in book covers all 120, so it can show who the survey missed.
| Claims checklist | Question to ask |
|---|---|
| Who | Does the claim match who is in the data? |
| Whole or slice | Does it use all the data, or a filtered slice? |
| Honest picture | Does the chart start at 0, with equal steps, a title and a source? |
| Numbers | Do the numbers really say "most", "all" or "biggest"? |
| Second source | Is there another source that agrees or disagrees? |
| Statement | True or false? |
|---|---|
| Filtering a data set can be a useful way to find information. | ? |
| A claim about all members can rest on a slice of afternoon members. | ? |
| A single source often holds all the data needed for a conclusion. | ? |
| The sign-in book could help the crew see who the survey missed. | ? |
Careful thinking, data analyst. Tomorrow is Impact Friday: who a chart can help, and who it can harm.