The nature of quantitative research

‘Split-half’ in research means grouping indicators so that the degree of co-relation between the answers can be examined. Typically, ten indicators would be divided into two groups of five each. Now we can see if respondents who scored high on one group also scored high on the other. We have, literally, split the group of indicators in half. Why? To show that the indicators we have used, actually relate to the concept, and thereby guarantee internal reliability.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 157

An experimental design allows us to test for causal connections between variables, because one of the variables (the ‘independent’ variable) is manipulated to track changes in the other (the ‘dependent’ variable). However, most social survey research uses cross-sectional designs, where such manipulation is not possible. Consequently, degrees of correlation between variables can be determined but causality remains inferential. If you gave answer (b), you should recognize that very few researchers are interested in mere descriptions of things. They usually want to find out why things are the way they are so that they can be remedied or replicated. Causality is an appropriate goal, simply difficult to achieve.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 163,164

Some critics of quantitative research see it as pretending that a photograph is a good representation of life, rather than being a ‘frozen’ instant of it. As a consequence, quantitative research is accused of assuming that social life is static, clearly not the case. Furthermore, the ontological basis of this kind of research obliges the social-science researcher to regard people in the same way that physical-science researchers regard nature and again, clearly there is a ‘world’ of difference. However, the measurement process, largely because of the need for all those tests of validity and reliability, does tend to leave quantitative researchers with a deep sense of accuracy of their research results. In the view of some critics, this confidence is misplaced, because among other things, it is unlikely that respondents will share a precise interpretation of the terms used with the researcher. Most of the criticism comes from proponents of qualitative research.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 166, 167

Devising measures of concepts is shown as step 4 in the process of quantitative research (fig. 7.1, p150). Bryman points out that this step is often referred to as operationalization, in other words, the series of separate steps we will take to make our research work for us. This is very important when we think about tests of validity of the research. The operational definition is therefore, the very opposite of abstract; attempting to phrase the concept so precisely as to make it capable of being tested in the research context.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 150

The issue here is with the application of the research findings to people who were not part of the research focus. If we select our sample of respondents randomly from the population as a whole, we can be quite sure that the findings can be applied to the whole population. But if we interviewed people casually, we could not generalize our findings beyond the actual people interviewed. This is the essence of external validation of research: how universally can the research findings be applied? It must be said that even with random sampling, we have no right to apply our findings to other populations, no matter how strong the temptation.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 163, 164

Measurement validity is concerned with whether the measure used actually measures what it says it will. Bryman uses examples of formal exams and the Consumer Price Index. Do these measures really, truly, measure academic ability (as opposed to rote learning, say) or inflation (as opposed to some government norm)? The various types of validity include ‘face validity’: does the measure strike us intuitively as being capable of measuring the concept?; ‘concurrent validity’: if some people say they like cream in their coffee, do they also say they dislike coffee without cream, for example, on the basis that we might expect such opinions to be held concurrently; ‘convergent validity’: does the measure we use tend to produce the same kind of results as another measure to track the same concept? This final test can be ‘passed’ by using two research instruments, with one used as a check on the other. ‘Conductive validity’ is a concept that applies to logical argumentation and is not a form of measurement validity.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 158-161

Measures include things like demographics (of age, income and so on), which can be counted. In fact, usually we think of measures as raw numbers. Often though, what we want to research does not lend itself immediately to straightforward calculation on numbers of things and how they vary but on slightly vaguer concepts. Like job satisfaction, for example. In this case we need a number of attitude statements, which, taken together, can be argued to represent the concept. These separate statements are indicators and often represent our ‘common sense’ understanding of a concept. Later, these can be coded, to turn them into numbers for statistical analysis.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 152, key concept 7.1

It should be obvious by now that developing measures that are valid and reliable is an extremely rigorous process. This can explain why researchers are often tempted into short-cuts, since they really are concerned with discovering things and reporting on them as urgently as possible. Although this means that a lot of fascinating research remains at the indicative level only, the underlying impulse can be understood. This does not provide an excuse for haphazard research methodology. On the contrary, it means that for your research to be taken seriously, you must pay great attention to the research tools you use. The more attention you give to development of your methodology, the less the criticism can be of your findings.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 167, 168

Under the heading, “Why measure?” on page 152, the author offers three reasons for our concern with measurement in research. Firstly, it “allows us to delineate fine differences between” cases or people. General observation might be enough to detect extremes of opinion but measurement is needed for the more subtle variations that actually exist. Establishing a measure once allows us (or others) to use it again, later on, with the same people or with others, providing a consistent benchmark. Finally, by studying co-relationships, we have a basis for studying how closely concepts relate to each other. So, answer (d) is correct: “all of the above”!
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 152

Bryman defines ‘reverse operationism’ (cited as Bryman 1988a:28 on p167) as an eventuality in research whereby concepts are generated by measures, or indicators, rather than the other way around. We might think of this as ‘reverse operationalism’ to stay consistent with the terms used in this chapter. Obviously this is not an intended procedure but rather something which emerges from extensive analysis of indicators, typically through factor analysis. As with any other statistical analysis technique, this certainly has a place in quantitative social research. Factor analysis is a sort of ‘trial-and-error’ analysis, attempting to discover which indicators are more likely to belong to a particular group of indicators than another. It is a useful tool in re-thinking social segments, leading to the formulation of new concepts for testing.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 167