My
research philosophy is very much in keeping with Feyeraband, who in "Against
Method", counseled doing "what works" when it comes to research
methodology. I believe in having and using a large methodological toolkit such
that I can fit the method to the hypothesis/research question and the (often underestimated)
research opportunity. Research is fundamentally about problem solving. It starts
with questions, real world observation, and literature reviews, proceeds to theory
and hypothesis, and prepares to iterate as objective observations (in the form
of experiment, quasi-experiment, archival data analysis, meta-analysis, ethnographic
observation, or some other form of systematic testing confirms or denies our hypotheses
and provides the basis for better theory.
I have many colleagues
who do brute force experiment-based theory development. Each new experiment results
in a set of ANOVA or t-tests, the confirmation and/or disconfirmation of a set
of hypotheses, and the creation of a new set of hypotheses which sets up yet another
experiment. I've done that kind of research. It works, but I often find it inadequate
to the questions I find myself asking. I am not a fan of ANOVA. I sometimes wish
every ANOVA ever done was reconceptualized as a multiple regression or (preferably)
a path model. I tend to push my students to learn regression, if only for the
real world flexibility it gives you to allow the N of various conditions to model
the real world, to consider continuous measure covariates, and to shape the interaction
effects tested to the hypotheses. Yet I have performed ANOVA's and will do so
again if it is appropriate to do so.
I have many colleagues
who do nothing but participant observation. They develop research questions and
occasional hypotheses. They observe behavior in the real world and use thick description
to explore what they observe, the relationship of those observations to their
research questions/hypotheses, and to suggest new theory. I've done that kind
of research (in my dissertation). It works, but I often find it inadequate to
the questions I find myself asking. There is a point at which it is appropriate
to apply formal content analysis and statistical hypothesis testing to our real
world participant observations. Indeed, Internet media make this particularly
easy to do. A little measurement can be a good thing, even in participant observation.
Yet I have done participant observations in which I have done no measurement at
all and will do so again if the method fits the need.
I could go on, but in the end my research philosophy is simple. First, find a good question. Second, find an opportunity to observe the question in the real world (or at least a good thought experiment). Third, check the literature to see what others have done. Fourth, if you can find any patterns, build theory. Fifth, to the extent you can, build a diverse methodological toolbox so that you can pick the most appropriate methods for the theoretical questions when you find an appropriate real world research opportunity. Lou Brock, the veteran base stealer of the St. Louis Cardinals said it best: "Luck is where opportunity meets preparation." Build your toolbox. Be prepared.