Understanding Research (37 page)

Bruce Berg puts it well when he notes that a more productive way of proceeding is working with a ‘model that encompasses both the research-before-theory and theory-before-research models’ (Berg 2009: 26). This is, not unlike Holmes in fiction and many scientists and scholars in everyday life, a mixture of inductive and deductive modes of reasoning with preferences and their consequences emerging along the way.

In the above section we looked at the differences between inductive and deductive modes of reasoning; how these approaches diverge in theory yet also overlap in practice along the course of research projects in which the gathering of primary data is a core activity more or less, but not exclusively, prior to its analysis.

Alongside these two paths runs another set of debates about the nature, the ‘special status’ claimed by scholarly research.
Chapter 3
addressed some of these issues in terms of distinctive worldviews (
Chapter 3
, 000 section).
Chapter 6
and the first part of this chapter also looked at the practical implications of not only divisions but also overlaps between how quantitative and qualitative sorts of evidence relate to particular rules and procedures for gathering and then analysing this evidence, including the sorts of conclusions, generalizations or specific details, that a researcher can expect to make from their findings.

There is a larger issue, one that could be seen as the quintessential marker of the quantitative–qualitative divide if we were to step back and take a bird’s eye view: the difference between research based on
behaviouralist
understandings of human nature and behaviour, and research based on
constructivist
ones. How these differences become evident within and across disciplines, even departments, differs from place
to place and generation to generation, their more starkly drawn
incommensurability
often emerging during public debates, peer-assessment procedures, or in research seminars.

What are the key differences and why do they matter?

1. Constructivist approaches to understanding research place the explanatory emphasis on a combination of forces, with the claim that ideas (thought), actions (behaviour) and societal-level phenomena are interconnected.
   
2. Behaviouralist approaches understand research as primarily the pinpointing of a cause-and-effect relationship. They aim to isolate a clear ‘variable’ that can be observed, measured, and then analysed accordingly based on hypothesis testing.
   

Working example
: if we take
sex-gender roles
, indeed the whole way in which the term gender is conceptualized within these two schools as an example, the implications of these differences are that:

1. Constructivist research examines and then interprets – draws conclusions – about a phenomenon as multidimensional; for example, that social attitudes about sex-gender roles and which tasks around the home are done by women or men, are mutually reinforcing rather than biologically predetermined. At the one end, gender is performed and lived (Butler 1990) and thereby not reducible to chromosomes, genitalia, or brain-size. At the other end, biological bodies, minds – neurological and consciousness – develop in a symbiotic relationship (see Haraway 1990, Swaab 2010).
   
2. Behaviouralist research concerns itself less with these issues. Rather it looks to isolate specific behaviours and attitudes in controlled ways (see
   Chapter 6
   on surveys). In the case of sex-gender roles, or gender-based behaviour differences that are manifested in actions, or attitudes between men and/or women around the home we can see research to ascertain if, in fact, attitudes about sex-gender roles and who actually does which household chores correlate in observable practice or popular opinion.
   

Both approaches could rely on the accumulation and use of quantitative or qualitative sorts of evidence, although constructivist research tends to favour the latter. The key difference boils down to diverging ways of drawing conclusions about first how people, individuals or groups, and societies operate and, second, explanations for why they do so.
¹⁶

Let’s zoom in on the behavioural approach. For the purposes of this discussion the focus is on the historical link between behaviouralist psychological models and behaviouralist approaches to social research. Both are based on a core question: ‘Why do people behave the way they do?’ (Sanders, cited in Marsh and Stoker 2002: 58). This is different from asking another core question: Why do people think the way they do? Or: Why do people think, or say one thing and behave in quite another way? Moreover, answers to the first question will differ according to the underlying, if not explicit psychological approaches to why human beings behave the way they do.
¹⁷

Two operating principles distinguish behaviouralist from non-behaviouralist takes:

1. Any analysis has to be based on observable behaviour, individual or group-based. For instance, in political science behaviouralists examine why people participate in politics, what influences voters’ political preferences. They also look at political leaders and interest groups, and how they behave at election time or during a crisis. These observations are usually gathered through survey instruments, experiments, or a combination of the two.
   
2. As this approach is based on a view that empirical data is defined by it being observed, within the larger debates touched on in earlier chapters, this means that behaviouralists position themselves as researchers who take a ‘scientific approach’ to the way they study social and political phenomena – a source of past and ongoing friction between advocates of this view of what constitutes science and their critics.
   

Any theory purporting to explain the political (or social) behaviour under study must be empirically, in this case meaning
quantitatively
, verifiable. In behaviouralist psychology, political science, media and communications studies we see then how this interest in why people behave they way they do (based on what they say and observations of what they do) have become equated with quantitative methods and large scale data-gathering techniques.

However, the behavioural movement’s embrace of quantitative techniques and with that its claiming of the latter as the only way to understand the notion of empirical (and so ‘objective’) knowledge has not gone unchallenged from within those disciplines where it has pride of place. For example in political research, theorists and political philosophers were quick to suggest that those working in this tradition had lost touch with the real world of politics and the best way to study it, because of their increased reliance on refining data-gathering techniques (see Barber 2006, Berns 1962, Lippmann 1998 [1922]). As far back as the 1960s the tendency to create sophisticated survey instruments was to some commentators tantamount to ‘the sacrifice of political relevance on the altar of methodology. The questions asked and pursued are determined by the limits of the scientific method rather than by the subject matter’(Berns 1962: 55).

This view has been echoed in other parts of academe; critiques have been developed and rich literature and research traditions have evolved out of this fundamental difference about the aim of scholarly research and its claims to produce a particular sort of superior knowledge; from feminists and postcolonial scholars coming of age in the 1960s to generations today who stake their scientific reputation, and thereby claim to engage in scholarship as a socially and politically worthwhile enterprise, on treating the ‘subject matter’ as the primary objective; method follows these matters.

Whether quantitative or qualitative approaches to investigating these matters do, or should be confined to the behaviouralist side of the divide is a moot point. As I note in the opening chapters of this book and all the way through, endeavouring to quantify something is not in itself a pointless exercise for a non-behaviouralist research project. Nor should the incorporation of survey data or quantitative content analysis necessarily relegate your project to the ‘enemy camp’; conversely, neither should
undertaking an ethnographic study that includes questionnaires or other uses of statistics turn you into a ‘woolly postmodern’ thinker. In many respects, research in the humanities and social sciences has moved past these polarizations as the emergence of mixed method, multi-sited analysis, or hybrid research designs and guides show.

The polemics surrounding the rights and wrongs of particular methods reduces these procedures to their (mis)application in certain instances, thus caricaturing the subtleties and corresponding debates occurring within any mode of research. Such standoffs actually obfuscate the practical realities of research design decisions, given that no data-gathering as process or product is above reproach. Stronger still, popular imaginaries and media discussions of ‘good’ or ‘bad’ science perpetuate caricatures of academic research rather than facilitate productive debate, reproducing some longstanding stereotypes along the way.

In the meantime, academic research continues as debates proliferate. Let’s return to some working examples of where practice belies political differences.

For research projects pursued as part of university degrees, the narrow notion of empirical as synonymous with quantitative modes of data-gathering, quantitative
data as facts pure and simple, and then their statistical analysis as ipso facto ‘scientific’ can result in either an overly reverent or dismissive views of their usefulness before a research project has even got off the ground. Many research students hobble themselves all too early in the day by rushing to characterize their project as either quantitative or qualitative per se, and then launch into the corresponding ways of data-gathering without considering whether these are actually best suited to their research question or even match the purpose of the inquiry. This is effectively research design by prejudice rather than consideration.

Putting it another way: whether you are interested in quantifying in order to make predictions and generalizations about the social world, or interested in symbolic, representational and experiential dimensions to the world, the way you go about gathering and making sense of material pertinent to the research question at hand may require more pragmatism than dogmatism, more use of available knowledge and research findings than overly ambitious claims of originality.

How do these conundrums, and the dead-ends they can create during a project, unpack in the real world of academic research?

1. One of the first questions you may ask is whether there are data already out there somewhere, collected and present, that may already answer your research question or allow you to further refine it.
   
   • (a)   There are large data archives that keep, document and distribute both qualitative and quantitative data for secondary use. But if you are going to collect your own – primary – data, compile your own ‘data sets’ or raw material based on your own labour and collection method, there are some general rules of thumb to bear in mind.
     
   • (b)   Recall that ‘data’ can be understood as either qualitative or quantitative in terms of the nature of the material gathered (see
     Chapter 1
     ); for example, in-depth personal interviews and focus groups provide data in large chunks of text (recorded or written) whilst large-scale surveys, using questionnaires for instance, provide answers that are designed to be counted up, collated and then analysed statistically.
     
2. Within the quantitative tradition and to some extent within qualitative research discussions, there is a common misconception that qualitative data only ever refers to that which is immeasurable. First, this overstates the case for research based on statistical – or numerical forms of measurement. As noted in
   Chapter 2
   a certain degree of imprecision is accepted in quantitative data sets. There is plenty of qualitative data that also measures attributes and qualities. It simply does not use numerical categories to do so; recall the discussions about the Body Mass Index (BMI) as an indicator of weight/degrees of thinness or fatness that is both qualitative and quantitative.
   
3. Not only between but also within both traditions, quantitative are distinct from qualitative data. Whereas the first designates numerical quantities such as measurements or counts, for example, height, weight, price, the second can be used to describe other sorts of attributes, for example, sex, nationality or commodity. In the idiom of quantitative research, these sorts of attributes are referred to as ‘soft’ data, approximating rather than precisely measuring the characteristics of a thing or phenomenon.
   
4. What may be referred to as quantitative data can be simply an assignment of numbers to qualitative categories, for example,
   gender
   – as represented by designations of masculine or feminine/male or female – is often represented as two quantitative categories (1 assigned to women and 0 assigned to men, for example) for the purposes of statistical analysis.
   

Second, what do such conventions mean for the design and successful execution of the research part of your project?

• More important than privileging one sort of data (and its gathering) over the other is that the theoretical premises – including the research question, or hypothesis if you prefer – underpinning any sort of investigation (measurement, fieldwork) make sense. There needs to be some sort of inner logic to the relationship between the conceptual framework, or
  substantive theory
  , being put forward, the techniques used to gather data, and the analysis carried out.
  
• This cuts both ways; if the theoretical side of things, research question and research plan are well thought-out and articulated then so should the data-gathering tools and forms of analysis (measurement or otherwise) be also. This interplay, inner coherence, is what gives researcher, and audiences, confidence in the eventual findings and integrity of the means by which the knowledge has been produced. Others may disagree with your findings or challenge you on these mechanics but that is a different matter (see
  Chapter 8
  ).
  
• Once again, with all other choices in the research process, these decisions rest on the question you are asking; do not assume too quickly there is only one best way of doing things. Moreover, not all data and data-gathering are transferable.
  

Third, what do these principles mean at the analysis stage and how is the handling and eventual presentation of any data gathered, however defined, to make sense in the final report, whether or not your conclusions are negative, positive, categorical or qualified? By way of illustration, the three cases below, drawing on studies of voters and political elites from quantitative research literature, show the ways in which notions of qualitative data are applied within quantitative research traditions. They also highlight how not all sorts of qualitative data lend themselves to quantification.

(a)
Assigning qualities
: A commonly used question in large-scale surveys of citizens and attempts to measure how ‘warmly’ a voter feels about a political leader is a ‘feeling thermometer’. Respondents to the survey are shown a picture of a thermometer marked from 0 to 100 degrees and given the name of a party leader. The respondent then indicates the warmth of feeling he or she has toward the party leader with 100 being the warmest. The thermometer here is a heuristic device to elicit how favourable a voter is toward a party leader, to capture emotion or affection for the party leader rather than how respondents evaluate a party leader based on the policy positions of the party.

This approach recognizes a truism for qualitative research: in practice it is difficult to separate affective judgments from policy-based judgments. The feeling thermometer attempts to measure the former along a quantitative scale so as to make
comparisons across a larger sample in order to conduct a statistical analysis (by calculating the mean thermometer ranking for a party leader or candidates across all respondents).

(b)
Gender as an empirical category
: Gender is considered a qualitative category for quantitative researchers as well. It is impossible to talk about an ‘average gender’. However, it is possible to talk about the ‘quantity’ or percentage of women in the population. Researching gender as an empirical category (part of a demographic) is also where research drawing on a range of feminist approaches parts company (see Hughes and Cohen 2010); gender is construed as a relational and analytical category in which quantifying is not the main objective (see Butler 1990, Peterson and Runyan 1999, True 2001).

(c)
Qualitative databases
: The
Economic and Social Data Service
(ESDS) has been running a project to capture qualitative data and make it available to secondary users.
¹⁸
While the ESDS has been known for providing access to quantitative data, this newer initiative provides access to a range of social science qualitative data sets. Electoral Reform and British Members of the European Parliament, 1999–2004 (designated in the archive as SN 5372) comprises qualitative interviews with sixty British MEPs of the 1999–2004 European Parliament. The interviews focus on the representative role of the MEP and the impact upon this of the change in the electoral system (from single-member district representation to multi-member regions) in Britain for the 1999 European elections. The available data if downloaded from the data archive consist of the transcribed interviews with the sixty MEPs.
¹⁹

The latter example underscores how transcribed interview material is not amenable to the same mathematical transformations as the quantitative indicators used in the thermometer example. Moreover, analysis that understands
gender
in other ways than a countable empirical category would not be content with numerical assignations such as ‘0’ or ‘1’ particularly for research projects looking at sexuality, social relations based on people living with or being assigned multiple (rather than two) genders, and feminist research approaches that regard gender as a lens, or as a facet of power relations (Peterson and Runyan 1999, Shepherd 2009).