The authors indicate that quantitative and qualitative research methods in social science only differ stylistically – they share the same underlying logic of inference, which in a inquiry takes the form of statistical methods. Whilst the qualitative is apt to discover differences in kind, quantitative research can judge phenomena on their degree.
The goal of inquiry is always inference. Descriptive inference uses observations to learn about unobserved facts, while causal inference enables to learn something about the causal effects from the observed data. The procedures by which one conducts his research are public, which is why a researcher should always state explicitly his methodology. If not, it is impossible to reproduce or check a research design. Since inference per definition leads to uncertain results, one should always include an estimate of uncertainty. Since ‘the unity of science consists alone in its method’, science only distinguishes itself from another text by its method. Science is a social enterprise, implying that one builds on the foundations laid by others, adding just a bit.
In order to say something meaningful, we have to abstract from the infinitely complex reality and conceptualize each studied ‘case as a member of a class of events about which meaningful generalizations can be made’ (p. 10).
All research is analytically divided into four parts:
1. Research question – should always be important in the real world, make a contribution to an identifiable scholarly literature by increasing our ability to construct verified explanations of some aspect of this world, which means nothing more than that one should situate himself in existing literature. Of course, a question that cannot be translated into a project permitting descriptive or causal inference is useless.
2. Theory – is a speculation about the answer to a research question, implying several descriptive or causal inferences. It should generate as much observable implications as possible, in order to be able to falsify it afterwards. Formulate the theory as concrete as possible, whether you adhere to the concept of parsimony or not. Do not adapt your theory to data afterwards – unless it is to widen the range of phenomena to which it applies. ‘Human beings are very good at recognizing patterns but not very good at recognizing nonpatterns’ (p. 21), meaning, do not add just a restrictive condition without new data to test the new theory.
3. Data – record and report the process by which data is generated. Collect data on as many observable implications of a theory as possible, trying at all times to maximize the validity of our measurements, meaning that we have to have data that actually says as much as possible about our theory. Again, data should be reliable, meaning that following the same procedure produces the same data. Everything should be as replicable as possible. In qualitative research this traditionally expresses itself in footnotes and bibliographical references and essays.
4. Improving use of data – this is the main goal of inferential statistics in social science. Data should be unbiased, meaning that a ‘procedure will be correct when taken as an average across many applications – even if no single application is correct’ (p. 28). Maximize efficiency – use all the relevant information and aggregations as much as possible to support implications.
In this volume, we will furthermore study the following themes:
· Using observable implications to connect theory and data;
· Maximizing leverage – meaning explaining as much with as little as possible, either by increasing the number of observable implications of a hypothesis, or by improving the data as to indeed observe more of the implications. Always state all possible observable implications, as to enable other researchers to continue or judge your work – quantitative and qualitative implications alike.
· Reporting uncertainty – always report an estimate of the degree of certainty we have in an inference.
No comments:
Post a Comment