Statistics and the Inference to the Best Explanation: Living Without Complexity?
Statistics are everywhere. In their descriptive mode they often indicate the level of satisfaction of clients/customers of post offices and hospitals, of one’s use of an email system (e.g. Eudora), of the number of downloads (provided by publishers of journals), and are gathered for various purposes (including to evaluate scholarly conferences such as AERA). This may be innocent as far as it goes, but it carries with it a number of presuppositions that should not necessarily be taken for granted. One of these is that it is possible to represent reality (and have a grip on it), shifting often to the simple equation that this is the reality we belong to. This either implies that (1) there is nothing else to know about a subject or (2) if there is anything else to know, this is still the best way to deal with reality, the one that is the least harmful and most objective. Another presupposition (or should one say promise) that descriptive statistics carry with them is the possibility of control, particularly when an explanation is offered (by this we mean the control of the evolution of dependent variables in terms of the independent variables which have led to the distributions that are offered or the correlations that may be observed). Thus a paradigm of (quasi-)causality enters that goes together with estimation and management of risk (this applies in the case of experiments as it does for randomized field trials). At the same time, the attraction of statistics lies in its simplicity (the reduction of the variability of the data to as few ‘factors’ as possible, as few homogeneous classes as the data permit, even for N = 1 studies where a simple formula is strived for or if more variables are dealt with, where the interaction between them is estimated/calculated), as well as in the accompanying belief that it is possible to characterize reality and the way it shapes our lives. Though what has been outlined so far may well be recognized (or deplored) by many, there is a more complex story to be told. Moreover, though we may be familiar with the possible objections and take these ‘warnings’ into account, this does not really change the fact that we often unthinkingly rely on what is presented to us. Why is that the case, or in other words, what do statistics do for us that make them so irresistible?
KeywordsRecognition Memory Reference Class Bayesian Hierarchical Model Physical Evidence Current Mood
- Hempel, C. (1965). Aspects of scientific explanation. New York: Free Press.Google Scholar
- Lipman, P. (2004). Inference to the best explanation (2nd ed.). London: Routledge.Google Scholar
- Salmon, W. C. (1989). Four decades of scientific explanation. Pittsburgh, PA: University of Pittsburgh Press.Google Scholar
- Winch, P. (1958). The idea of a social science and its relation to philosophy. London: Routledge.Google Scholar
- Wittgenstein, L. (1953). Philosophical investigations/Philosophische Untersuchungen (G. E. M. Anscombe, Trans.). Oxford: Basil Blackwell.Google Scholar
- Wittgenstein, L. (1966). In C. Barrett (Ed.), Lectures and conversations on aesthetics, psychology and religious belief. Oxford: Basil Blackwell.Google Scholar