Hacker News new | ask | show | jobs
by Syrup-tan 3725 days ago
How to mislead with bar graphs; the tutorial

https://denpa.moe/~indy/e75d18.png

2 comments

The bar graphs were not designed to compare total amount of lines for each gender/age-range combination. Rather, they're meant to indicate how the percentage of lines for each gender is skewed older for male actors vs younger for female actors.

If they were to determine the length of each bar using amount of lines instead of percentage of lines, all that would be immediately clear is that females have less lines, a fact well established in the rest of the article, and the point would be lost.

author here. YESSSSS
What do you find misleading? The absence of 100%? Or the age ranges they chose?

The bar graphs might visually amplify the gap, but 1.1 million lines versus more than 5 million is a real gap, for example.

If anything, they gave us 2 different views on the data, and that double ;-) what we usually get.