I wrote: However, this is a reactive, after-the-fact morality, based on extrapolations which have already occurred....
That should be: However, this is a reactive, after-the-fact morality, based on extrapolations from events which have already occurred... It could of course be the case, that a model has already been built and tested, allowing previous extrapolations to be fine tuned, in which case the extrapolations have already occurred. The "pro" argument concludes: "We desperately need better ways to forecast political instability, and the Policy Analysis Market had significant promise." Does anybody know of research which suggests the effect of publicized assessments of political instability on the actual level of political instability ? Interestingly, the variables mentioned, "mass demonstrations, unemployment levels, arrests, assassination attempts" could be integrated quite easily into a more comprehensive, intelligent Marxian explanation, which, unlike empiricist econometrics or psephology, offers a systematic set of theoretical concepts to analyse these phenomena. A good basic indicator of political instability would, presumably, be an increase in expenditure on small arms and military weaponry and equipment, but the question then arises, is this increase in expenditure and its result a cause, or a consequence of, political instability ? This is not a silly question, in view of the post-war "arms race" between the USA and the Soviet Bloc. This is the basic problem with empiricism, it seeks to infer determinate relationships from observed correlations, without a clear connection to theory specifying causal relationships. Another problem with arguments based on probabilistic correlations is that the variables and their relationships are relatively fixed. This may be no major problem in regard to some price data, price index data and population data, but as regards political event and political process data, cause and effect could be reversed, it would require dialectical theory to understand this. One way this problem was addressed in the past, from memory at MIT, was by tracking a very large number of indicator variables in one model or set of models. The problem with that is, that the number of possible linkages between variables becomes very great, and some theory is then required to interpret the data, but what sort of theory is this ? Reference: Taylor, C. L. and D. A. Jodice. 1982. World Handbook of Political and Social Indicators. New Haven, CT.: Yale University Press. Jurriaan