I love statistics, but one of its recurring problems in practice is the loss of fidelity in high abstraction. After chi-squaring a meta-analysis of questionnaires and applying this or that test for “significance” or whatever, it’s easy to forget what exactly we were talking about in the first place. Sure, SES correlates with IQ, both of which being approximations of indicators ad nauseum. Remind me, what use is the correlation of the approximations of two or more indicators of categorical variables? Ah, right, description and prediction. Seems in practice we can’t describe how well we’re describing and most of us loony tunes can’t even predict how well we’re predicting.
It is an egregious reality that after a few such abstractions, we often find ourselves talking in terms of real numbers and dimensionless ratios. But these are the tools of the physical sciences, and psychology is not yet a physical science and- as I understand it- never will be. Lamentations of the Dawkins-Harris crowd notwithstanding, naturally. The fact remains that no matter how many times you add, subtract, multiply, and divide integers, you’ll never come up with a real number. And you might not want to look too closely at the method by which those integers were divined in the first place.
Well, something I figured early on is that if you aren’t sure what you’re talking about, it’s a good idea to look at the units of measure. And since I’ve defined my first approximations for conscientiousness and associative horizon as relics of intelligence, it might be a good idea to figure out what intelligence is, exactly, and maybe write it down somewhere. So I guess that’s my goal for this weekend.