Is psychology a science? [Part 2]

By | November 14, 2014

[Part 1 of this discussion is found here.]

The original proposition that sparked this discussion was put forth in a recent study by a couple of psychologists that claimed that “academic math-intensive science is not sexist”. This study has many methodological flaws, and I’m not sure that its rather sweepingly generalized conclusions are really justified. But arguing that this particular study isn’t first-rate science doesn’t require arguing that psychology as a whole isn’t a science. Here, the problem isn’t with psychology as such, so much as it is with the formulation of the problem, the operational definitions of key terms, and the nature of the data used. There’s nothing in principle that forbids the formulation of a proposition regarding sexist behavior within academia in terms that are both clear and measurable, or that won’t allow the formulation of a model, the collection of relevant data, or an analysis that interprets relationships found within those data as meaningful comments on the degree to which that proposition is or is not supported.

Of course, there are many different ways in which the proposition could be operationalized. The choice among them ought to be based on the potential utility of the answer, just as the choice of a model in physics is based on utility. In the behavioral sciences, “utility” of necessity includes some understanding of politics and economics – but then, these factors affect research the physical and biological sciences as well. There is no shortage of examples of where research in these fields has been notably skewed by investigator status, institutional influence, and the presence of various “ideas in good currency” that turn out to be fundamentally flawed. And all brands of science – physical as well as behavioral – There’s no question that psychology and other behavioral sciences have had their share of sloppy definitions, poor analyses, and inappropriate claims and generalizations.

Science requires clarity and measurability – epistemological qualities that describe the communicability and intellectual coherence of a concept, in physics and chemistry no less than any other domain. No concept is “self-defining”, and there are usually alternative representations. For example, “pi”, “π”, “3.14159…”, and “the ratio of the circumference of a circle to its diameter” all describe the same concept, and the choice of one representation or another depends on its use and the intended audience. Note that in this case the verbal description of the concept is actually more exact than the numerical representation, since a transcendental number can only be approximated digitally.

The biggest single problem in any science is determining the range of generalizability of a finding – that is, what is the range of other phenomena to which our result might apply? Sometimes this can be tested empirically, by replicating an experiment or other kind of study with different samples. However, given the reluctance of journals to publish replicant studies, the general shortage of research funding, and the unending pressures on scholars to come up with new and original findings, replication is seldom actually carried out. And when it is, it’s not infrequent for the original finding to fail in confirmation. Generalizability is always a leap, and it’s important for scientists of all stripes to be quite careful in their claims. Certainly this is one of the areas in which the sexism study noted earlier fails to convince.

The key point in this argument is that science is a methodology, not a set of particular content domains. 400 years of the philosophy of science going back to Bacon support this perspective. This is certainly the current consensus in the “science establishment”. The National Academy of Sciences is just about as establishment as it gets; their working definition of “science” is “…the use of evidence to construct testable explanations and predictions of natural phenomena, as well as the knowledge generated through this process.” NAS has sections for psychology, economics, and social and political sciences, along with physics and the rest. Clearly the consensus in the scientific community is that behavior is scientifically researchable and that finding in this area can be credible.

Both the National Academy and the American Association for the Advancement of Science have both debated the validity of social and behavioral sciences as legitimate and worthy of inclusion within their ranks. on numerous occasions and concluded that they are sciences. The official history of NAS published in their Proceedings (; it contains the following excerpt:

“Although psychologists and anthropologists had been well represented among the membership from early on, only in 1966 was the first sociologist, the demographer Kingsley Davis, elected to membership in the Academy. Political science, sociology, economics, and social psychology all were added beginning in 1966. However, when I arrived to manage the behavioral and social science division of the NRC, relatively few members of the Academy were social and behavioral scientists.

Most members of the Academy were still getting used to the fact that there were sociologists, political scientists, and economists around. However, the acceptance on the part of the Academy of all of the social and behavioral sciences has been quite remarkable over the last 35 years.”

It’s worth noting that the new members just nominated to the National Science Board include two survey researchers closely associated with the University of Michigan’s Survey Research Center, which has for fifty years been a pillar of hard-headed behavioral science research. Since I did my graduate work at Michigan, I feel a certain justification in the presence of these scholars on the Board governing the agency where I worked back in the 1980s. Our interdisciplinary team – made up of three psychologists, an industrial engineer, an economist, a sociologist, a political scientist, and a couple of others – produced an extremely well received and widely circulated comprehensive literature review on the processes of technological innovation, as well as a portfolio of research projects that helped shape the national agenda on innovation during that rather time.

The public record is clear that the systematic study of human behavior is a “science”, according to the consensus of the best scientific minds in this country. Disciplinary chauvinism has been definitively rejected across the board, although it continues to be occasionally stubbornly reiterated.