I'm pretty sure it's the former. I don't find it plausible that these three things wouldn't have a causal effect on income, but it's a fair point that it would

I'm pretty sure it's the former. I don't find it plausible that these three things wouldn't have a causal effect on income, but it's a fair point that it would be more useful to see those percentages broken down by starting income level.