Seems easy to prove wrong by trivially looking at high school dropout vs graduate incomes.
For example, they might be risk averse and a risk averse personality might lead to other decisions that make more money. (or the opposite)
It would be easy for a fad to arise calling education unnecessary and a waste of time, and for people to consider the option to drop out less risky just because it gets talked about more. So even if the correlation held statistically now, in any year it could change just because the subgroup of risk-averse people are more likely to consider it. So you still couldn't use it to make permanent decisions(like creating a school system for hundreds of years).
Homeschooling vs. high school graduates would remove "choice to drop out" from the student, but then "choice to homeschool" on the parents might carry down genetically or culturally and maybe "aversion to government authority" improves earnings on average.
I try to reject all statistics on "easily understandable" social topics. Leave that for scientists only. It's too easy to make a mistake and not get any warning that you made a mistake, and believe wrong things your whole life.
All observable evidence shows those who complete their schooling in any form make more money
I don't care whether schooling causes or doesn't cause a benefit: You called it "trivial" by looking at statistics, and that supreme confidence that there can be no other explanation seems unjustified.
> Seems easy to prove wrong by trivially looking at high school dropout vs graduate incomes.
There is an implied statistical argument here that by plotting "did X drop out" on one axis and their future income on another, that a correlation would refute a statement that "Schools cause students to earn less".
Specifically, you would be:
1. Gathering many examples of individuals(a "population") with boolean "did they drop out?" V1, some income metric V2, maybe some others.
2. Comparing and aggregating V1 and V2 across the population
3. Using it to refute a statement S
That is a statistical argument for ~S because it involves comparing population aggregates. You may not have explicitly used the word "statistics" or realized you did it, but you made a statistical argument.