Methods of Discovery: Heuristics for the Social Sciences, by Andrew Abbott

Methods of Discovery: Heuristics for the Social Sciences, by Andrew Abbott. New York: Norton, 2004.

Reviewed by Thomas J. Scheff

This book, a pint-sized paperback about 5 by 8 inches, has only 248 pages of text but packs a wallop. For $17.40 at Amazon, it's a steal. For all the social and behavioral sciences, it offers an exciting boost to our “craft or sullen art.” If you are in the business of research, don't hesitate: read it IMMEDIATELY. No matter your persuasion, you won't be sorry.

The otherwise helpful Glossary offers only an abstract definition of the key word heuristic: “A discipline that aims to facilitate invention and discovery of new facts and ideas in the sciences.” But Abbott actually uses the word in its dictionary meaning: “Involving or serving as an aid to learning, discovery, or problem-solving by experimental and especially trial-and-error methods” (Mirriam-Webster online, emphasis added).

This definition demotes heuristic from a high-falutin discipline to the lowly level of mere aid, like a shovel or screwdriver, to be used temporarily, trial and error, and in harness with other tools. You can't build a house with only a hammer. So much for sacrosanct loyalty to pure disciplines, sub-disciplines and “schools of thought.” Quantitative, qualitative, sociological, psychological, historical, experiments, surveys, ethnography, CA, feminism, Marxism, and all other methods, theories, and approaches are only aids against the sluggishness of thought. They are to be ignored whenever they don't help with the particular problem or data-set at hand. Hang loose, dudes and dudesses!

To illustrate the use of heuristics, Abbott draws upon brief profiles of particular studies in the social sciences. Examples from history, economics, political science and sociology hammer home the practical details of heuristics as aids to discovery, even when, as is usually the case, the authors didn't think of their work in this way. Abbott's meta-analysis makes new even the most familiar of studies, like any good heuristic, often to the point of eclipsing the point of view of the original authors. Perhaps this is the reason that in his review of Abbott's 2001 book, Charles Tilly noted that it varied “in tone from Don Quixote, to Cassandra, to Vlad the Impaler…”

One serious limitation of Abbott's use of examples is the absence of studies drawn from psychology, psychiatry and psychoanalysis. For reasons not explained, there is no empirical work by Solomon Asch, Freud, Colin Parkes, and many, many others who have made substantial contributions. Another book should be written about heuristics in these fields, since they are at least as important as those chosen from the social sciences.

Some of the aids that Abbott proposes are so simple as to sound like tricks or gimmicks. Running out gas on a research proposal? Try reversing a well known hypothesis, or re-labeling the concepts. Come to think of it, in my own work I have begun to “emotionalize” language that hides emotions under cognitive and behavioral disguises. E.g., “terror management” studies of “mortality salience” can be translated into the simpler and more universal language of fear, just as the innumerable studies of shyness, social anxiety, lack of self-confidence, self-esteem, etc can be outed from the embarrassment/shame closet (Scheff and Fearon 2004).

Many of the heuristics that Abbott proposes are fancier, however. What he calls lumping, splitting, and analysis of counterfactuals seems to be a middle level. Perhaps the most sophisticated is his celebration of fractals: “…the property of recurring at finer and finer levels, always in the same form.”(p. 250) Abbott has also published an earlier work (2001), mentioned above, that promotes fractals.

Fractal models might help explore a central issue in social science: how can we build links between the most microscopic levels of human conduct and experience up to the highest levels of culture, social institutions and society? In his 2001 essay, Abbott applied this idea to the reproduction of conflict at various levels between and within disciplines and sub-disciplines in the academy, and in the world of intellectuals in general. Perhaps the enigma that is Goffman's Frame Analysis (1974) might yield up its treasures to a fractal analysis of his otherwise dizzying arrays of boxes within boxes within boxes.

The fractal idea gives rise to interesting questions in social science. The elegance of fractal geometry in the physical world arises because of the EXACT duplication of forms at different levels, with no difference at all except in size. Snowflakes provide an example. Goethe, in his botanic studies, noted that in plants such as palms; the whorls of the trunk can be found repeated down to the smallest living sub-units.

But arguments over Marxism at various levels in the history of the socialist movement are similar in some ways and different in others. Certainly Proudhon's rebuff to Marx was never repeated, at least in so eloquent a form. The shape of snowflakes is elementary compared to the complexity of human discourse. Both sides of the conflict between the Leninist and Trotskyite lines probably varied with each argument, depending upon context, emphasis, choice of words, overt and/or covert emotional content, etc. To use the fractal heuristic effectively would require conceptual and operational definitions of each “line”, so that the extent of variation could be noted. The same reasoning applies to deciphering Goffman's frame analysis, since neither he nor any of his commentators provided a clear definition of frames. Heuristic analysis puts interesting demands on the researcher.

I promise that this book will help any researcher who reads it, but it's not easy. Since it is meta-analysis, the logical path is relentless and without digression or humor. There are no images of actual human beings or their voices. Although it has a hands-on feeling about conducting research, it provides only a bird's-eye view of the human condition. This reader found it necessary to take a break and a breath after each page as a heuristic for digestion.

At the risk of being impaled by Vlad, a final complaint. It seemed to me that a faint wisp of complacency about social science hung over the text. I could be mistaken about this, but it appears that in choosing the best of social science studies as examples, there was a tendency to equate best of a bad lot with good in some absolute sense.

One example is the Blau and Duncan (1967) study of occupational status attainment between generations. This study is mentioned several times, and given high marks each time. Innovative at the time, it provided the model for many subsequent quantitative studies. It is often seen as magnificent social science. The amount of variance accounted for is higher than most social and behavioral science studies: compared to studies based on self-esteem scales, it is gigantic (Scheff and Fearon 2004). But even so, it is still quite small, nowhere near half the variance. That means to me that the major causes of status attainment are still undocumented.

As far as I know, replications, like snowflakes, continue to produce copies of the original study without gaining ground on identifying major causes. What about luck, pluck, obsession, greed, and criminal/unethical proclivities?

Durkheim's study of suicide has been replicated many times, because it was innovative at the time it was conducted. Like Blau and Duncan's, Durkheim's study accounts for only a tiny proportion of the variance. Even the best of the subsequent studies didn't seek the major causes of suicide, content to be snowflakes. Innovative studies should be honored, not worshipped. After enough replications, they should become heuristics, not shrines.

Still, I find Abott's book the most exciting contribution to the craft of research for many a year. Don't argue; go buy, borrow, or steal it without further delay.


Abbott, Andrew. 2001. Chaos of Disciplines. Chicago: University of Chicago Press.

Blau, Peter, and Otis D. Duncan. 1967. The American Occupational Structure. New York: Free Press.

Goffman, Erving. 1974. Frame Analysis. New York: Harper and Row.

Scheff, Thomas and David Fearon. 2004. Social and Emotional Components in Self-Esteem. Journal of the Theory of Social Behavior. 34: 73-90

Abbott Nov 10 1337