Approaches to data analysis are important in that they offer a theoretical orientation to practice. Three particular types of approach are often highlighted in the literature:
Induction: this is the process by which we draw a general conclusion from individual instances or observations. The benefits of an inductive approach, as seen for example in grounded theory, are that it allows flexibility, attends closely to context and supports the generation of new theory [see the paper on social loss as example]. To its critics, however, inductive research painstakingly works from first principles when there is no overriding need to do so given there is already a huge amount of existing literature.
Deduction: The deductive method seeks to draw valid conclusions from initial premises. It follows the logic of syllogy expressed in classical form as:
Socrates is a man (major premise )
All men are mortal (minor premise )
Therefore Socrates is mortal (conclusion)
Deduction as an approach to social research has had considerable appeal, and has been most clearly associated with a kind of classical and logical positivism. In its purest form deductive logic has been associated with the hypothetico-deductive approach. This involves generating formulating quite specific hypotheses about phenomena generally on the basis of existing practical and theoretical knowledge. The hypothesis is then tested under experimental conditions. If the data support the hypothesis then the hypothesis can be said to hold in this context; if not, then assuming that the research was well designed and carried out rigorously, the hypothesis, and the theory underlying the hypothesis, is challenged or at least the limits of the theory may have been reached.
The hypothetico-deductive approach is most associated with 'the scientific method' but also underpins desk based research such as large N studies, meta analysis and systematic reviews. While powerful, deductive logic is criticised as misrepresenting the methods of natural science and, for that matter, makes an assumption that all disciplines in natural science work the same way when they do not. Second, it carries an inbuilt logic of confirmability; if you go looking for an association you are likely to find it. Third, it focuses on association between events but fails to provide the detailed analytical explanation which is a necessary part of establishing causality - this is why counter examples (such as the Conger and Donnellan, 2007 paper) are so important.
Abduction: The claims made for an inductive or deductive approach are contested fiercely but there is increasing recognition that this might not be a choice between one or the other. Instead research can, and often does proceed, by taking an alternating inductive and deductive perspective – with observation leading to hypotheses which are then explored in relation to the data. In practice whatever their explicit particular stances on research and social theory, our interviewees show the flexibility associated with an abductive approach.
For you to discuss
- how can you best describe the approach to analysis in your work?
- what are the weaknesses and strengths of this approach?
If you would like a basic introduction to deductive and inductive logic you might try this You Tube video: