Robert: > The problem is that concepts like heteroskedasticity > refer to samples and how well they reflect the total > population. Here we have the total population of US > presidential elections, so we do not need > statistical inference
I don't want to turn this into a pissing contest but we are talking about time-series data here. Of course we need statistical inference. Each cross-sectional population, even when we have the total population at a given time, is a random draw from a stochastically evolving process. By the way, sticking with concepts like wage-labor, capital, return on investment, profit, overproduction, etc, is not in contradiction with also sticking with concepts such as Hilbert Spaces, Measure Theory, Probability Theory and the like. Also, you may complain, as I always do, about the jargon, or the code, that only those who know the code can follow but this is no reason to think that it is obscurantism. I find it difficult to agree with claims such as that we should disregard the statistical obscurantism in favor of an historical analysis, whatever that means. Statistical and historical analysis can go hand in hand, as long as we are aware of the limitations of both. Best, Sabri
