At least when I was in school, one thing that kept getting hammered home was that people had difficulty understanding graphs. That I didn’t see as being an issue, not unless you’re talking about something pretty exotic, but if accurate, it seems like it’d be pretty limiting. And I’ve seen a lot of later articles also saying that a lot of people have trouble with graph comprehension.
Last month, Alessandro Romano, Chiara Sotis, Goran Dominioni, and Sebastián Guidi surveyed 2,000 people to demonstrate that “the public do not understand logarithmic graphs used to portray COVID-19.”
They found that only 41% of participants could correctly answer basic questions about log-scaled graphs (v.s. 84% accuracy for linear-scale).
But the problem is harder than log scales. As you’ll see below, much of “the public” struggle with even the most basic charts and graphs, let alone complex visualizations.
I wish they’d had more statistics. In my high school, they had half a semester as an elective. In my university curriculum, it wasn’t a core class (social science majors would study it, I would guess).
I have explained basic sampling to a ton of people on Reddit when polls came up, because they didn’t believe that a sample of 1000 people out of a much larger population could result in a representative outcome. We see poll data all the time. Even if you never perform a poll, understanding the mechanism at least enough to trust it and understand when a poll might not be representative (e.g. self-selecting Internet polls). Confidence levels. I think I covered regressions in high school in an (elective) physics class, not even in a statistics class. That’s a useful skill – get a bunch of numbers, be able to produce a formula to predict more of them and get an idea of how accurate your model is. I assume that if you didn’t take it or it wasn’t offered, then you just wouldn’t ever touch on them.
At least when I was in school, one thing that kept getting hammered home was that people had difficulty understanding graphs. That I didn’t see as being an issue, not unless you’re talking about something pretty exotic, but if accurate, it seems like it’d be pretty limiting. And I’ve seen a lot of later articles also saying that a lot of people have trouble with graph comprehension.
googles
https://towardsdatascience.com/numeracy-and-graph-literacy-in-the-united-states-ea2a11251739
I wish they’d had more statistics. In my high school, they had half a semester as an elective. In my university curriculum, it wasn’t a core class (social science majors would study it, I would guess).
I have explained basic sampling to a ton of people on Reddit when polls came up, because they didn’t believe that a sample of 1000 people out of a much larger population could result in a representative outcome. We see poll data all the time. Even if you never perform a poll, understanding the mechanism at least enough to trust it and understand when a poll might not be representative (e.g. self-selecting Internet polls). Confidence levels. I think I covered regressions in high school in an (elective) physics class, not even in a statistics class. That’s a useful skill – get a bunch of numbers, be able to produce a formula to predict more of them and get an idea of how accurate your model is. I assume that if you didn’t take it or it wasn’t offered, then you just wouldn’t ever touch on them.