The main task of the social sciences is to explain social phenomena. It is not the only task, but it is the most important one, to which others are subordinated or on which they depend.
A phenomenon in the technical sense used in this book is an effect or process that:
- Is (usually unobservable but) measurable
- Is of scientific interest of noteworthy
- Must be inferred from the data using more or less sophisticated techniques
- Is to some extent idealised
- Comes in two kinds: types (repeatable) and tokens (unique)
A scientific explanation falls into two parts:
- An explanandum (the thing to be explained), a description of the phenomenon of interest.
- An explanans (the explaining thing), the sentences that are thought to give an account of the phenomenon.
The logical positivists thought that the right kind of connection between the explanans and the explanandum was that of logical consequence: the explanandum must follow from the explanans as a matter of logic. The logical positivists imposed two main conditions. First, the sentences in the explanand must be true or at least verifiable and verified. The argument that is to serve as a scientific explanation should not only be valid but also sound. Second, the permisses must contain at least one scientific law non-redundantly. This condition in fact comprises three requirements: at least one of the premises must be a law; the law must be ‘scientific'; and removing the law(s) from the set of premises renders the argument invalid.
According to this model, explanations are deductive arguments that contain at least one scientific law among their premises. The DN-model is now generally regarded as inadequate. Nevertheless, its discussion is useful as a starting point because of the terminology that was first introduced in the context of the model and that is still being used by philosophers, and because many of its deficiencies have triggered searches for alternatives that are more defensible.
A scientific explanation is a logically valid and sound argument (‘deductive') that contains at least one scientific law essentially (‘nomological'). Schematically:
Laws L1, L2, ... Ln
Initial conditions C1, C2, ... Cn_____________________________
Therefore, E (the explanandum or phenomenon of interest)
Criticism on the DN-model
First, some philosophers have argued that human behaviour does not fall under laws. CN- explanation of human behaviour would require laws that connect mental events such as beliefs and desires with physical events such as actions but there cannot be any psycho-physical laws. And yet, it seems as though many explanations of human behaviour are successful.
Second, even generalisations that do not cite human motivations or beliefs are seldom strictly universal. All generalisations are rough, subject to qualifications and exceptions. Law statements do not (usually) express universal generalisations but rather causal tendencies or ceteris paribus laws. A causal tendency obtains when some factor pushes another in a certain direction, even when its operation is interfered with by a disturbing factor. Ceteris paribus laws are similar, they hold under certain conditions.
Third, it is not clear whether laws qua regularities (whether accidental or genuine) are explanatory at all. The reason these laws hold is that few regularities in the social science are brute, not subject to further scrutiny. Rather, empirical generalisations hold on account of a deeper socio-economic structure, and only if the underlying structure operates in a certain way, an empirical regularity will ensure. A good explanation in economics will therefore ask for the detailed causal process or mechanism that is responsible for the phenomenon of interest.