Comment by rconti
2 years ago
I'm not sure if it was just me, but I struggled with the visual style. In some groups there were more rows than others, but then the rows would be of different lengths, making it difficult to intuitively compare the population sizes, especially when trying to break them down by color coding.
It felt like the "some adverse experiences" group was worse off than the "many adverse experiences" group, which I'm guessing is incorrect.
I was a bit sceptical to start with about correlation versus causation. Causes are what we are looking for here. If we see someone get shot, does that mean we decide not to go to university[1]?
I watched the video, and the semantic meaning of pink people kept changing, and I couldn't follow the story because too many moving parts.
There's a study looking at people from a "bad" neighbourhood, that used data on immigrants to and emmigrants from the neighbourhood to try and track causation.
If I was feeling obnoxious I would grab the data, and massage it until the conclusion is that we should blind children so they don't see someone get shot so that they go to university.
[1] actually I can think of plenty of friends where that would be plausible (disclaimer: gun violence isn't so common in New Zealand). I'm trying to pick an example where causation and correlation are more disjoint but I think I've failed here.