I don't think it's implying causation where there's only correlation. It seems to be implying non-causation due to non-correlation. I'd say the bigger flaw is

I don't think it's implying causation where there's only correlation. It seems to be implying non-causation due to non-correlation. I'd say the bigger flaw is that it only looks at the very top tax rate without looking at the point at which the top bracket cut in. In the 1950s the top tax rate was enormous but it only affected a tiny number of people. Now it's a lot lower but affects a larger slice of the population.