OrientationObservationIn-depth interviewsDocument analysis and semiologyConversation and discourse analysisSecondary Data
SurveysExperimentsEthicsResearch outcomes
Conclusion
8.2.3 Falsificationism and middle-range theorising The positivist explanatory approach implies causal relationships (see Section 2.2.1.1). However, establishing cause and effect in the social world is extremely difficult. The quantitative survey approach implies cause but rarely proves a causal relationship. Social phenomena are generally too complex for it to be possible to establish causes.
The approach, then, tends to be to identify 'likely factors' that lead to a particular phenomenon, rather than their specific causes. The aim of most positivist survey research is to identify the independent factors that have an effect upon a given dependent factor.
The process by which this is done is based upon Robert Merton's notion of middle range theory (see Section 2.2.3). Middle range theory lies between grand, sweeping, but untestable speculation about the social world, on the one hand, and microscopic 'proven' statements about an insignificant aspect of the social world on the other. Middle range theories are statements about aspects of the social world that can be tested by reference to data. Epistemologically, this is directly linked to a form of positivism called falsificationism (see Section 2.2.1.6).
Falsificationism asserts that it is impossible to prove scientific statements on the basis of observable data because there may always be a future instance that will reverse any proof. What it is possible to do, they claim, is to disprove statements on the basis of empirical evidence. Proper scientific theories, then, are ones that involve statements that are testable and are capable of being refuted. These are what Karl Popper (1980), the best known falsificationist, calls 'bold conjectures'. At any point in time scientific theories, consisting of conjectures that have been tested and not refuted, are the best theories we have. Science progresses, he claimed, through a constant process of testing and refuting bold conjectures.
The conventional positivist approach to surveying is, thus, to adopt a falsificationist methodology.
For example, Travis Hirschi and Rodney Stark (1969) drew up three hypotheses that suggested a relationship between delinquency and religion and then collected data to test them out (See CASE STUDY Religion and Delinquency) Their third hypothesis, for example, suggested that religious training prevents delinquency because it promotes belief in the existence of supernatural sanctions. As the case study revealed, there is no relationship between delinquency and belief in supernatural sanctions. This leads to the rejection of the hypothesis. Having tested all three hypotheses the study showed that, contrary to the original theory, religion does not have an effect on delinquency.
In this way positivist survey researchers attempt to uncover causal-type factors. By testing hypotheses derived from theory, they are able either to confirm that a theory appears to be adequate or to suggest that the theory is inadequate and needs to be modified.
The middle range theory approach is cumulative. It suggests that testing one set of hypotheses derived from theory will lead to modifications in the theory. A new set of hypotheses will then emerge by which the new theory can be further tested, and so on round the circle. Each circuit, it is assumed, will lead to an ever better theory.
In practice, few sociologists are likely to go round the circuit more than once. The usual approach, when writing a report, is to put forward how suggested changes in theory can be further tested and to leave this up to other people to pursue.
