Quantitative and qualitative methods

QUANTITATIVE AND QUALITATIVE METHODS

Within the social sciences, a never-ending discussion and debate is centred on this issue of method. Should the social world be explored and explained quantitatively (e. g. using numbers and statistics), or qualitatively (e. g. using words and narratives, in the wide sense), or both?

This question has a partly philosophical basis concerning the fundamental features of the world (what the world ‘is’, or what exists) and how we can study those fea- tures, and what the researcher’s task consists of. Quantitative methodology is historically tied to the philosophical branch of positivism and post-positivism, where the (somewhat simplified) stance is that the world exists independently of us and our understanding of it. The essential features can be counted in numbers (e. g. how high is the risk that a given individual re-offends after x years of abstinence? ) and thus be studied with the help of statistical methods.

Qualitative methodology, on the other hand, is commonly associated with the traditions of constructivism and symbolic interactionism. This tradition stresses the basically constructed nature of the social world where nothing (not very much, at least) is natural or static, but the result of a specific social and historical context. How people perceive, interpret and understand the world is a big part of what actu- ally shapes that world.

This relatively stiff dichotomy of quantitative and qualitative methodology (also discussed in Chapters 1 and 3 of this volume), and their associated philosophical underpinnings, is becoming less and less distinct, however – most prominently within life-course research. The dichotomy is being replaced by a kind of methodological pragmatism where the researcher simply chooses the method(s) that suit(s) the research question and that it is possible to use, given the various practical boundaries of a given project. Maruna (2010: 127f. ) expresses this point clearly:

Qualitative methods involve ‘deep’ immersion into a social scene that allows for aware- ness of situational and contextual factors that are often missed in [quantitative] research. They produce ‘rich’, ‘holistic’ data, as opposed to the focus on ‘variables’ … In its published form, qualitative analysis provides vivid illustration of phenomena, bringing social processes ‘to life’ for readers. Quantitative research does little of this,

but has considerable strengths precisely where qualitative research is weak. Quantitative methods are transparent and do not rely on a ‘take my word for it’ approach. This work is therefore more replicable, precise (some would say ‘objective’), and generalizable than qualitative research. Additionally, statistical techniques allow for the eliminating of confounding influences and better assess cause and effect rela- tionships among variables. In published form, they produce findings that are notable for their clarity, succinctness, exactitude, and parsimony.

Criminal career research and life-course criminology in the 1970s and 1980s was dominated by influential, quantitatively driven studies. This is evident from several of the traditional key concepts: prevalence, offending frequency, duration, intensity, etc. (see Blumstein et al., 1986). Much of the most prominent work within the field has been done with quantitative data, including the pioneering studies of Wolfgang, Figlio and Sellin (1972), Blumstein et al. (1986) and later studies such as that by Piquero, Farrington and Blumstein (2007), to mention just a few.

Two forms of quantitative data

There are, you could say, two forms of quantitative data: official records (i. e. those that are collected and kept by the various authorities of a society) and specific quan- titative data collected by the researcher within the frame of a given project – for example, your official employment history, health history and whether you at some point in time have lived in a single household or not, which exists in many countries’ official records.

Register data is therefore an amazing research resource. In most countries of the world, but particularly in welfare states such as Sweden and some other European countries, governmental and municipal agencies collect huge amounts of high-quality data on their citizens. Collecting and using these records is made simple by the fact that every person has a unique personal identification number. After getting ethical approval, researchers can be allowed access to these records.

Now, importantly, this data is collected for administrative purposes, which can entail a problem for the researcher: the purpose of the agency is not always compat- ible with the specific problem the researcher is interested in. Here, then, is a prime example of those compromises the researcher has to make when it comes to the research design and the content of the data. The researcher is often forced to use variables and scales which constitute far from optimal operationalizations of the various constructs the researcher is interested in. However, often the only option to access data is to use data with certain types of restrictions. If the researcher is aware of these limitations and takes them into account in the interpretation of the results, they can still be used.

At the same time, the agencies’ administrative records have one huge advantage: data are dated and usually entered into the register in a chronological order.

This means that official register data has an inherently longitudinal character, even if it is collected by the researcher at a single point in time.

We noted above that whether or not you have lived in a single household often exists in official register data. But, for life-course criminology, it is a crucial question whether you experience that you had a ‘good upbringing’ or not. You can get indi- cators of ‘good upbringing’ by combining a large number of registry variables regarding the individual’s early years and possibly even parents, but, in general, that kind of data is better quality if it is gathered through more or less structured inter- views, surveys and/or tests specifically constructed for the study, and the validity of the measurement will be better. The great value in doing this is that it can usually provide a much richer description of the people you study and in a much better way covers the research questions and theoretical points of departure than mere official records can.

One possible issue with this, which is very relevant in studies with a prospective design, is that what is considered a ‘good’ measurement of something can change over time. So, for example, in the research project we worked on (the Stockholm Life-Course Project; SLCP) 287 Stockholm boys born in the 1940s and 1950s were enrolled in a study and underwent a large number of tests, including IQ tests (Terman-Merrill) and Rorschach tests. Today, the first of these measurements is con- sidered an acceptable but not very good indicator of a person’s intelligence (and who knows how it will be considered in the future? ). The second form of test, Rorschach, is far too unreliable and imprecise to be used. This dilemma – what time does to a data set – is unsolvable but important to keep in mind when it comes to longitudinal studies with long follow-up periods.

Similarly, when researchers construct indicators of important theoretical con- structs, they do so from a set of raw variables. The variables we need to use to construct something change over time, as society changes. Stop and think about this: What did it mean to have ‘good economic standards’ economically in the 1980s? One such study tried to capture this by asking a whole bunch of specific questions, and then adding the answers to those questions together, forming an ‘index’. Among other things, that battery of questions included a question about whether or not the respondent owned a VHS player (Sarnecki, 1985). Today, such a question is highly outdated, and if we were to conduct the same measurement of having ‘good eco- nomic standards’ today, we would need to include other questions. The researcher, in other words, must continuously be conscious of the temporal dimension of their study, and how the relevance of certain questions, measurements and variables may change as society changes.