« Lovely short VR video: "Pearl" | Main | State of the vacation »

Misleading graphs

| No Comments

I've seen several potentially misleading graphs in articles lately, so here's a reminder of a couple of common y-axis issues:

  1. If the y axis doesn't have numbers on it (for example, if it's just labeled as going from “less” to “more”), it's hard to tell how serious the issue that the graph describes really is.
  2. If the y axis doesn't start at zero, the low point on the graph may look lower than it really is.
  3. Special case of item 2: If the y axis runs from one number to a barely larger number (for example, if it runs from 0.5 to 0.525), it may look like the change shown is much larger than it really is.

I saw an article this morning that did all of those things (in different graphs), in an attempt to convince readers that (problem x) was far worse than it's ever been. The article's thesis may well have been true, but the graphs didn't support the thesis nearly as well as the article said they did.

(Note: There are legitimate uses for all of those y-axis-labeling techniques. But if you're using those techniques, you should be very clear about it in the accompanying text.)

Post a comment