More on Milgram’s Methods of Research

In a previous post, I introduced Stanley Milgram’s experiments on obedience and authority. We watched a short video clip in class and students responded to questions about Milgram’s research methods. Upon realizing that the unwitting test subjects were all males, one student wondered whether that would have biased the results in a particular direction. The students hypothesized that women may have been much less likely to defer to authority and continue to inflict increasing doses of pain on the test-takers. While there are good reasons to believe either that women would be more or less deferential than are men, what I wanted to emphasize is the broader point about evidence and theory as it relates to research method and research ethics.

The 'sophisticated' machinery of the Milgram Obedience Experiment

The ‘sophisticated’ machinery of the Milgram Obedience Experiment

In the video clip, Milgram states candidly that his inspiration for his famous experiments was the Nazi regime’s treatment of Europe’s Jews, both before and during World War II. He wanted to understand (explain) why seemingly decent people in their everyday lives could have committed and/or allowed such atrocities to occur. Are we all capable of being perpetrators of, or passive accomplices to, severe brutality towards our fellow human beings?

Milgram’s answer to this question is obviously “yes!” But Milgram’s methods of research, his way of collecting the evidence to test his hypothesis, was biased in favour of confirming his predetermined position on the matter. His choice of lab participants is but one example. This is not good social science, however. The philosopher of science, Carl Hempel, long ago (1966) laid out the correct approach to  producing good (social) science:

  1. Have a clear model (of the phenomenon under study), or process, that one hypothesizes to be at work.
  2. Test out the deductive implications of that model, looking at particularly the implications that seem to be least plausible,
  3. Test these least plausible implications against empirical reality.

If even these least plausible implications turn out to be confirmed by the model, then you have srong evidence to suggest that you’ve got a good model of the phenomenon/phenomena of interest. As the physicist Richard Feynman (1965) once wrote,

…[through our experiments] we are trying to prove ourselves wrong as quickly as possible, because only in that way can we find progress.

Did the manner in which Milgram set up his experiment give him the best chance to “prove himself wrong as quickly as possible” or did he stack the deck in favour of finding evidence that would confirm his hypothesis?