At the end of the day, all that the investigative psychologist can do is to measure, record, or obtain records of, behaviour: whether it be the behaviour of neurons in the striate cortex, the behaviour of rats in a maze, or the behaviour of people brawling in a pub. What someone says is behaviour. What someone fills-in on a form is a record of behaviour.
From these behavioural measurements/records, the investigative psychologist extracts/imposes patterns and deduces/infers meaning - often of cause and effect. Basically, this is all she/he can do. And it is all constrained, constructed and driven (knowingly or not) by a particular conceptual framework; an ethos; a set of assumptions; particular ways of seeing, modelling and symbolising; ways of understanding which are the result of an active, co-operative enterprise of persons in relationship (Gergen, 1985).
It is within the particular conceptual framework just outlined that this essay
sits. It concentrates on three methods used by social psychologists as alternatives
to experiments. My choice as to which methods to explore is based on their popularity
as indicated by their use in recent articles (January 1992 to December 1994)
in issues of the Journal of Social Psychology (70 articles surveyed) and the
British Journal of Social Psychology (40 articles surveyed). From this mini-survey,
I found that experiments were used in 37% of studies. The three leading alternative
methods were:
It would be nice to explore these three methods by saying where they are conceptually "coming from", where they are practically "going to" and pointing out their strengths and weaknesses along the way. However, due to the general theoretical upheaval in social psychology (legitimised by Kuhn's notion of "paradigm shifts"), and given the constructivist framework outlined above, the basis on which to build such a critique is neither solid nor obvious. Therefore, I believe a better strategy is to outline a number of conceptual strands and see how the three methodologies fit into them (as well as various other methods, when applicable). Five meta-theoretical strands will be considered: (1) language and introspection; (2) theoretical frameworks; (3) causality, correlation and meaning; (4) structures and processes; (5) practicality, ethics and usefulness.
Depending on where one stands, questionnaires are either far more revealing than observation and field studies, or far more open to bias and misinterpretation. This is primarily due to two key aspects of questionnaires: they must involve language and they almost always require introspection. From a Behaviourist point of view, the use of language and, especially, introspection in exploring the mind and behaviour was an anathema, and some still strongly believe that what people say when answering questions may indicate little about how they actually reason, feel and behave (e.g. Nisbett & Wilson, 1977).
However, the opposing view is becoming increasingly popular. For instance, introspective
verbal reports have been described as "a valuable and thoroughly reliable
source of information about cognitive processes" (Ericsson & Simon,
1980). As far as language itself is concerned, it is not too far from the truth
to say that it is the leading edge topic in social psychology today. For example,
Harre's (1977) ethogenic approach relies on the idea that what people say ("accounts")
and what they do (actions) are produced from the same stock of knowledge ("resource");
also the concept on which the autobiographical methodology is based (De Waele
& Harre, 1979). In addition, discourse analysis and conversation analysis
are both founded in the revealing nature of speech/text. This move can also
be seen in developmental psychology, which has moved from the generally non-linguistic
theories of Piaget to the sociocultural ideas of Vygotsky, where thought is
seen as developing from social interaction (primarily, language). It is not
surprising then that questionnaire surveys are so popular (especially as they
usually use quantitative analysis techniques - still generally preferred to
less rigorous techniques).
As alluded to earlier, all investigative methods must categorise and compare various phenomena. A major difference between methods is when this process is made. In questionnaires and field studies, the framework is usually well in place before the data is gathered. In observation studies, however, there is more variation: studies run from those which rigidly prescribe observational parameters and rules, and rigorously train observers (e.g. Bales's IPA coding scheme, 1970) to those which collect a large amount of (relatively) unstructured data then proceed to analyse it. This temporal relationship between theory and research process has been seen as being the essence of the distinction between quantitative and qualitative methods (Henwood & Pidgeon, 1992). Arguments concerning the pros and cons of when to impose (or extract?) conceptual categories are epistemologically complex. In brief, defenders of the "a priori" approach would point to the fact that all observation is driven by a conceptual framework, and it is better to have this well defined before data gathering since this will afford more rigour, reliability and validity. Defenders of the "data towards theory" approach would argue that it is free of theoretical strait-jackets, thus can generate new theories by tapping into the richness and complexity of social behaviour. Furthermore, it challenges the dualistic distinction between knower and known; i.e. the personal is always present in research, objectivity is impossible, and observer bias cannot be eradicated, so must be confronted.
Causality is often seen as the preserve of the experiment. This is not the case. For instance, analyses of recorded data, with no intervention from the investigator can yield causal relations: e.g. Jonas's (1992) use of time series regression analysis to show some suicides are due to imitation of famous cases. Hence, the name "quasi-experiments" for such techniques. Field studies are usually quasi-experiments, trading the control of true experiments for real-life situations. Thus, they often involve non-randomly assigned subjects, and may rely on naturally occurring changes as the manipulation of the independent variable.
Questionnaire surveys can offer causal explanations based on stratification
variables (c.f. "intrinsic" independent variables), such as age and
sex, but there are usually so many "confounding variables", that convincing
cases are rare (though the use of longitudinal surveys can improve matters).
Usually, then, questionnaires reveal correlational data, as do most observation
studies. Is this a bad thing? In some cases, yes. For instance, a correlation
between stress at work and absence rates, does not necessarily mean taking measures
to reduce stress will actually improve absenteeism (Manning & Oswald, 1989).
However, the question of whether causality is of much use in elucidating the
complex and dynamic web of social behaviour has become increasingly raised,
and forms a major part of the debate between experimental/positivist and naturalistic/interpretative
methodologies. According to the latter approach, it is as important, if not
more so, to investigate the meaning by which people interpret their and other's
behaviour, something which experiments rarely achieve due to their decontextualised
and artificial nature.
Questionnaires are a snapshot of behaviour. Hence their ubiquitous use in attitude research - attitudes, by their very nature, involving relatively stable knowledge structures built through experience (c.f. the schema of social cognition theory). Observation studies, on the other hand, reveal behaviour as a dynamic process. Observation studies, can inspect the sequential nature of behaviour using various techniques such as slow motion, time lapse and still-frame film/video. Hence, the predominant use of observation in studies of interpersonal behaviour - revealing the subtle and intricate nature of utterances, NVC and extralinguistic cues. The classic studies of Bales (1950, 1970), mentioned earlier, were concerned with the dynamics of group interaction and can seen as a forerunner of sequence analysis (Argyle, 1979). Indeed, some have argued that the basis of social psychology is best studied by finding the 'grammatical' rules, structures and sequential patterns of social encounters as key indicators of an actor's belief-systems (Harre, Clarke and De Carlo, 1985).
Viewing events over extended time periods can be very revealing, something which
the majority of experiments miss due to their short time frames. Hence, longitudinal
surveys, observational studies and field studies, which often encompass lengthy
time periods, have the advantage of seeing events unfold in time, thus can investigate
the processes whereby structures are formed, changed and used.
The undoubted popularity of questionnaire surveys is partly due to the fact that they can provide large amounts of easily analysable data relatively cheaply and quickly. They also have the flexibility of allowing for open-ended answers which can be studied using various post hoc methods (Jones, 1985). They are ethically sound since respondents choose to reply, although this can lead to self-selected samples. The practical usefulness of questionnaires would seem to be unquestionable given their proliferation of use by all manner of professional bodies - political parties, consumer businesses, advertising agencies, newspapers and university lecturers. However, as research on attitudes and beyond has shown, people often have highly complex, context specific, ill-formed and contradictory views which questionnaires cannot hope to access. Newer techniques, especially discourse analysis, have been developed to try to overcome these problems (Potter & Wetherall, 1987).
In contrast to surveys, field studies and observations can be time consuming,
expensive and raise serious doubts over ethical issues (particularly invasion
of privacy for subjects not volunteering for the study). The observation studies
featuring most in the recent literature, however, generally used volunteers
who were aware that their behaviour was being recorded on tape or video. Though
this monitoring could be seen to alter behaviour, most studies point out that
the subjects soon forgot they were being recorded and behaved normally (though
I have my doubts).
By delving into the meta-theoretical debates that have consumed much social
psychological thinking over the past 15 years, it is easy to lose sight of the
fact that concepts such as "truth" and "methodological rigour"
are only means to an end; the end being that psychology should be useful. Indeed,
if more attention were paid to the philosophical views of C. S. Pierce (who,
90 years ago, argued that there is no objective knowledge, hence its truth-value
can only be judged by its practical usefulness) then there may be more consensus
and coherence. The usefulness (or uselessness!) of social psychology in the
jury room has been brilliantly examined by McCloskey & Egeth (1983). All
psychologists, especially proponents of purely qualitative research, would do
well to consider the sobering ideas in this review, which includes a mock prosecutor-psychologist
cross-examination showing that at best qualitative testimony is useless, and
at worst, is positively harmful.
Finally, let me mention something of a "Catch 22" situation. For any research project, the choice of methodology depends on the question at hand, but due to the inadequacies of all methods is best approached using a barrage of different techniques. This, however, costs. To be useful and successful, then, requires good funding. But good funding is usually only available for projects adopting methods with sound track records. And because the people holding the purse strings feel happiest with experiments and questionnaires, these methods are likely to remain dominant for a good few years yet.
Argyle, M. (1979) Sequence in social psychology as a function of the situation. In G. P. Ginsberg (ed). Emerging Strategies in Social Psychological Research. Chichester: John Wiley.
Clarke, D. D. with Crossland, J. Y. (1985). Action Systems: an introduction to the analysis of complex behaviour. London: Methuens.
De Waele, J.-P. & Harre, R. (1979) Autobiography as a psychological method. In G. P. Ginsberg (ed). Emerging Strategies in Social Psychological Research. Chichester: John Wiley.
Ericsson, K. A. & Simon, H. A. (1980). Verbal reports as data. Psychological Review, 87, 215-251.
Gergen, K. J. (1985). The Social Constructionist Movement in Modern Psychology. American Psychologist, 40, 266-275.
Harre, R. (1977). The Ethogenic Approach: Theory and Practice. In L. Berkowitz (Ed) Experimental Social Psychology, Vol 10. New York: Academic Press.
Harre, R., Clarke, D. D. & De Carlo, N. (1985). Motives and Mechanisms: an introduction to the psychology of action. London: Methuen.
Henwood, K. L. & Pidgeon, N. F. (1992). Qualitative research and psychological theorising. British Journal of Psychology, 83(1), 97-111.
Jonas, K. (1992). Modelling & Suicide. British Journal of Social Psychology, 37(4), 295-306.
Jones, R. A. (1985). Research Moethods in the Social and Bevioural Sciences.
Sunderland, Mass: Sinauer.
Manning, M. R. & Oswald, J. S. (1989). The relationship between absenteeism and stress. Work and Stress, 3, 223-235.
McClosky, M. & Egeth, H. E. (1983). Eye witness identification: what can a psychologist tell a jury. American Psychologist, 38, 550-563.
Nisbett, R. E. & Wilson, T. D. (1977). Telling more than we can know: verbal reports on mental processes. Psychologocal Review, 84, 231-259.
Potter, J. & Wetherall, M. (1987). Discourse and Social Psychology: beyond attitudes and behaviour. London: Sage.