Discussion about this post

User's avatar
A disgruntled reader's avatar

Also….proper argumentation: Claim -> Evidence -> Source -> Citation.

Expand full comment
A disgruntled reader's avatar

your argumentation is missing some critically important information. Sure, most people don’t understand how multiple regression works and are subject to all kinds of fallacious thinking. here you’ve also faulted at making the point. You can control for whatever variables you want and the math won’t be upset at your choices. Which is to say, multiple regression is just correlation analysis. As a form of correlation analysis, it has no mechanism for determining causal forces. The theory of causality comes from outside of the math. SO, you’d want to explain how each of the factors in the wage gap analysis has a causal influence on wages. If they don’t have a causal influence then there’s no reason why they can’t be replaced or left out entirely. Moreover, you’d want to explain how gender discrimination DOES NOT account for differences between men and women along those factors. If they’re completely unrelated to gender, then you’re in an even worse analytical position because multiple regression would be the wrong tool for the job. You’d need probably structural equation modeling. While we’re on that point, the data need to be defined at the appropriate level to use multiple regression. If the data are provided in a aggregated form (I.e., at a city, state, industry, country level, etc.) then the statements about causality have to stay at that level because of Simpsons paradox and the ecological fallacy.

Expand full comment
4 more comments...

No posts