On Average: 3 Examples

A few years ago now, I put up a post called “5 Ways that Average Might Be Lying to You”. I was thinking about that post this week, as I happened to come across 3 different examples of confusing averages.

First up was this paper called “Political Advertising and Election Results” which found that (on average) political advertising didn’t impact voter turnout. However, this is only partially true. While the overall number of voters didn’t change, it appeared the advertising increased the number of Democrat voters turning out while decreasing Republican turnout. The study was done in Illinois so it’s not clear if this would generalize to other states.

The second paper was “How does food addiction influence dietary intake profile?“, which took a look at self reported eating habits and self reported score on the Yale Food Addiction Scale. Despite the fact that people with higher food addiction scores tend to be heavier, they don’t necessarily report much higher food intake than those without.  The literature here is actually kind of conflicted, which suggests that people with food addiction may have more erratic eating patterns than those without and thus may be harder to study with 24 hour dietary recall surveys. Something to keep in mind for nutrition researchers.

Finally, an article sent to me by my brother called “There is no Campus Free Speech Crisis” takes aim at the idea that we have a free speech problem on college campuses. It was written in response to an article on Heterodox Academy that claimed there was a problem. One of the graphs involved really caught my eye. When discussing what percentage of the youngest generation supported laws that banned certain types of speech, Sachs presented this graph:

From what I can tell, that’s an average score based on all the different groups they inquired about. Now here’s the same data presented by the Heterodox Academy group:

Same data, two different pictures. I would have been most interested to hear what percentage of the age ranges supported laws against NO groups. People who support laws against saying bad things about the military may not be the same people who support laws against immigrants, so I’d be interested to see how these groups overlapped (or not).

Additionally, the entire free speech on campus debate has been started by outliers that are (depending on who you side with) either indicative of a growing trend or isolated events that don’t indicate anything. Unfortunately averages give very little insight in to that sort of question.