This is a selective summary of parts of the above text (part 2).
3. Articulating the Theoretical and Empirical Fields
The Problem
The conceptual area in which any study is located must be made explicit. Key insights from prior work, as they relate to the study, create a working space within the theoretical field in which the problem is to be developed. This the authors call the problematic.
Within the problematic the problem is specified. Though this is logically the next phase it may only be within the process of investigation that the problem emerges fully. The problem may be expressed in the form of a proposition, a hypothesis or a research question, and ultimately in a conclusion or as a particular finding.
Levels of measurement
Propositions relate variables. Variables take different values/states within a range. The full range constitutes the variable's scale. Scales have different levels of measurement:
- Nominal scale: one without logical order (e.g. gender)
- Ordinal scale: one with logical order (e.g. developmental stages)
- Interval scale: numerical, where the intervals between adjacent points are equal (e.g. time in years). This can have a conventional zero (e.g. 1 AD) or an absolute zero (time since the beginning of an experiment).
- Ratio: a numerical scale with an absolute zero.
Interval scales permit arithmetical operations. More sophisticated operations may be performed on ratio scales. The frequency of a category occurs on a ratio scale since logically, even if not actually, the minimum frequency is zero.
Operationalisation: The Empirical Measurement of Theoretical Propositions
Operationalisation is the process of connecting theory and practice. The research problem, however articulated, must be framed in such a way that it is empirically accessible. Concept variables, appropriately scaled, are defined and applied within the empirical setting to cast light on the research question.
Validity and reliability
Validity is a measure of how accurately theoretical, concept variables are shown forth in empirical, indicator variables. As a concept variable is applied the units of analysis (the objects to be described) are coded on its defined scale. Face validity is exercised when there is understood to be a direct correlation between empirical observations and their associated concept values. Of course there may be ambiguities and these weaken the validity of a particular coding: 'Validity... is a measure of the extent to which you are measuring what you think you are measuring.'
Reliability is a measure of how consistently coding is carried out (e.g. by different people and/or at different times).
Demonstrating valid coding on a concept variable scale can be achieved in several ways. Elaborated description is a way of apprenticing the reader in the method of coding, serving to convince him/her of the reliability of the judgements made: repeated reference to anecdotal evidence, commentaries, illustrative examples and counter examples together strengthen validity. Alternatively, declaring explicit rules that link indicators with concepts provides validity, where this is possible. Precoding (for example when offering multiple choice responses in a questionnaire) is a third approach, though the validity of the precoding itself must be demonstrated.
Operationalization: Sampling Procedures
All empirical research necessarily involves sampling. There are a number of categories of sampling procedures:
- Opportunity sampling: taking an opportunity that presents itself
- Theoretical sampling: e.g. critical cases, a representative sample, quota sampling. Drawing samples in these ways depends on an underlying, theoretical reason for selection
- Random sampling: one way to attempt to draw a representative sample. The method itself is not random.
Brown, A. & Dowling, P. (1998), Doing Research/Reading Research: A Mode of Interrogation for Education, London, Falmer Press
