This article from 2014, with the provocative title “Poor kids who do everything right don’t do better than rich kids who do everything wrong” is terrible. It makes a shitty and misleading graph, and makes a variety of nonsensical judgements based on it.

The graph used makes it really hard to see what the actual distributions of overall income are. If you make, say, a bar graph of the income distributions, you’ll see that poor college grads do much better than rich high school dropouts. I made such a graph (here’s the spreadsheet):

Bar chart

I also estimated the average incomes of both groups based on the given statistics, and found an average income of $78k for the poor college grads and $60k for the rich high school dropouts.

The main piece of evidence that this blog post uses to support its thesis is: “Specifically, rich high school dropouts remain in the top about as much as poor college grads stay stuck in the bottom — 14 versus 16 percent, respectively. Not only that, but these low-income strivers are just as likely to end up in the bottom as these wealthy ne’er-do-wells. Some meritocracy.”

The first of these statistics is not clearly related to meritocracy. If you want high income mobility, you want both these numbers to be low. The sentence is phrased as if it’s bad that the numbers are similar; that doesn’t make sense.

The second of these statistics is correctly interpreted, but seems cherry-picked: I could just as easily point out that 41% of poor college grads end up in the top 40% of income, while only 19% of rich high school dropouts do. To prevent such cherry-picking, we should probably use the Schelling point summary statistics like mean income or median income, both of which indicate that the poor college grads are doing significantly better.

As far as I can tell, the original paper didn’t do anything wrong, this “reporter” just decided to make up some bullshit conclusions from the statistics. This is even worse than most reporting–the mistake isn’t something you have to read the original source to find, the mistake is right there in the graph that Facebook is suggesting as the image preview. Alas!