Beyond the micro/macro distinction

Abstract. Discrepant findings in electoral studies, particularly in relation to the importance of class, have re-opened the issue of micro- versus macro-levels of analysis. The 'ecological' and 'individualist' fallacies are again the subject of discussion. This article considers how similar issues arise in other sciences, natural as well as social, and shows how in many cases they are not resolved but lead to the development of sub-sciences. It argues that beyond the micro/macro distinction lies another, that between 'molecular' and 'structural' approaches, which exist in parallel in most sciences. The corresponding types of data - aggregate and integral - are found at both levels of analysis. Thus we have to contend not with two distinct types of data, but with four: micro- and macro-molecular, and micro- and macro-structural. In electoral studies, as well as 'individual' and 'ecological', there are also 'personal' and 'structural' types of data to be considered, though the latter have largely been neglected in recent times. Drawing valid inferences between any two types of data is difficult. Even if fallacies are avoided, intractable problems may remain. On the experience of other sciences, electoral studies may well continue to develop along parallel but in some ways discrepant lines.