3 ms·
A nitpick with the article. He argues that education -> democracy by using a test case to rule out democracy -> education. But he makes no attempt to rule out
by yummyfajitas 17y ago
A nitpick with the article. He argues that education -> democracy by using a test case to rule out democracy -> education.
But he makes no attempt to rule out the possibility that XXX -> {education, democracy}, where XXX might be { wealth, free markets, socialism, various other relevant factors }.
The real problem here is that he seems to want to do 2-variable correlations rather than multivariate regerssion.
- jacobolus 17y agoA major nitpick with the whole line of reasoning (and basically much of the work in comparative politics). We've (in political science) been trying to study democratization in a formal/quantitative way for at least the last 50 years, and every single explanation we come up with, however many variables we put into our regression, fails to informatively explain the data. We have too poor a grasp on the independent variables (what they are, how to measure them, etc.), too poor a collective definition of democracy ("I know it when I see it", essentially), and way way too little data to make definitive causal models. Running experiments to test our hypotheses would take decades, at least, so instead we mine through historical data looking for trends. As Feynman said, “To report a significant result and reject the null in favor of an alternative hypothesis is meaningless unless the alternative hypothesis has been stated before the data was obtained.”