Sunday, April 6, 2014

Is bad research unethical?

Suppose I were to decide to fudge some data in a publication. Maybe I refrain from reporting some

inconvenient data that don't support my argument, or perhaps I push my pet interpretation hard and don't bother to acknowledge counter-arguments or contrary data. Are these things unethical?

I refrain from this kind of activity. These activities seem deeply problematic and against my fundamental beliefs. I will usually make a stink if I catch such activities by colleagues or students. They are certainly against the standard ethical canons of science (see On Being A Scientist, something I often assign in graduate seminars). But I can't find anything suggesting they are against archaeological ethics. There is nothing in the Society for American Archaeology's Principles of Archaeological Ethics about this kind of scientific misconduct. The various collections of articles on archaeological ethics on my bookshelf don't say much about these problematic research practices. I would guess that the SAA journals would not accept papers with problematic methods like those I mention above, but the SAA style guide does not mention this at all.

This all seemed pretty normal until this afternoon. While reading up on rational choice theory, I found a reference to "post hoc theorizing" as something considered very negative in political science. Because I've been on the lookout for a concise description of what Binford called "post hoc accommodative arguments," I followed out some citations. Binford accused a lot of authors of this sin, but he never describes it in detail, or precisely what is wrong with it. I have always tried to get students to avoid this practice, but I've been looking for help from the published literature.

So I found some work on the problems of post hoc theorizing (see sources below). The main difficulty is that this practice prevents testing of one's hypothesis, since the interpretation is dreamed up after the data are gathered. It makes it difficult to know when an interpretation might be wrong. Post-hoc theorizing also opens up one's interpretations to random variation: the results are more likely to be due to chance. But then I was surprised to find these authors suggesting that this practice is also unethical.


“Graduate students in psychology are routinely taught the importance of delineating one's hypotheses in advance (i.e., prior to collecting data). Established researchers continue to regard it as questionable and possibly unethical to theorize after one's empirical results are known." (Baumeister and Leary 1997:313).

Post hoc theorizing jeopardizes the experimental method of psychology (and of much political science), and this is evidently considered an ethical lapse in that field. But I have never heard anyone suggest that the use of post hoc accommodative arguments was an ethical lapse in archaeology. Why not? I am no expert on ethics, but my guess is that the lack of a well-established methodology of data analysis in archaeology is the reason. We have canons of proper excavation technique, and if I were to screw up a dig and damage a site without proper documentation, it may violate archaeological ethics (I'm not really sure here; does incompetent excavation violate the principle of stewardship?). But my interpretation of Binford's critiques of post hoc accommodative arguments is that he was criticizing methodologically bad science, not unethical practices.

Should faulty argumentation, or other stronger cases of scientific misconduct, be considered violations of archaeological ethics? I'm not sure about this, in part because I haven't bothered to think much about it before this afternoon. But I admit that I have to admire the field of psychology if faulty experimental methods are considered an ethical breach. Perhaps archaeology needs stricter codes of ethics.

--------------------------------------------------------------
NOTE: I added this following material April 7, partly in response to Robert Mahaney's query.

Here is Kerr's (1998) list of the problems with HARKing ("Hypothesizing After the Results are Known"):

  • Translating Type I errors into hard-to-eradicate theory
  • Propounded theories that cannot (pending republication) pass Popper’s disconfirmability test.
  • Disguising pot hoc explanations as a priori explanations (when the former tend also to be more ad hoc, and consequently, less useful).
  • Not communicating valuable information about what did not work.
  • Taking unjustified statistical license.
  • Presenting an inaccurate model of science to students.
  • Encouraging “fudging” in other grey areas.
  • Making us less receptive to serendipitous findings.
  • Encouraging adoption of narrow, context-bound new theory.
  • Encouraging retention of too-broad, discomfirmable old theory.
  • Inhibiting identification of plausible alternative hypotheses.
  • Implicitly violating basic ethical principles.

 As for the ethical issue, Kerr notes that this practice is not mentioned in the codes of ethnics of the American Psychological Association, or the National Academy of Sciences. But he continues (p. 209):

·         “I think a case can be made that HARKing violates a fundamental ethical principle of science: the obligation to communicate one’s work honestly and completely. Albert Einstein states this principle well: ‘The right to search for truth implies also a duty; one must not conceal any part of what one has recognized to be true.” HARKing can entail concealment. The question then becomes whether what is concealed in HARKing can be a useful part of the ‘truth.’ ... The content of what is concealed or misrepresented in HARKing is undoubtedly less crucial than what is misrepresented when results are fabricated, but the damage done by widespread and recognized HARKing to mutual trust among scientists may be qualitatively the same.”  (p. 209)


Baumeister, Roy F and Mark R Leary  (1997)  Writing narrative literature reviews. Review of general psychology 1(3):311-320.

Green, Donald and Iam Shapiro  (1994)  Pathologies of Rational Choice Theory. Yale University Press, New Haven.

Kerr, Norbert L.  (1998)  HARKing: Hypothesizing After the Results are Known. Personality and Social Psychology Review 2(3):196-217.

Leung, Kwok  (2011)  Presenting Post Hoc Hypotheses as A Priori: Ethical and Theoretical Issues. Management and Organization Review 7(3):471-479.

4 comments:

Robert Mahaney said...

Hi Michael. This seems like a very interesting observation, but could you speculate on what an ethical and unethical post hoc interpretation would look like?

Michael E. Smith said...

Robert - Thanks for the question. I added some more details from Kerr's paper, the main source on this practices. As for archaeology, our standards of inference are quite sloppy, with little methodological or epistemological agreement. Therefore it would be hard to claim that a practice like this violated the ethics of archaeologists, beyond the general notion of scientific ethics.

Marcus said...

I really fail to see how post-hoc theory is problematic in a discipline that deals with history. Isn't everything after the fact for us? I fail to see how I am doing something unethical in ditching my initial theory chapter as I encountered data that contradicted it.

You want archaeology to be like experimental social science, but it isn't. Binford was wrong about that. In a science dealing with historical patterns post-hoc arguments are just fine.

Michael E. Smith said...

The "after the fact" refers to the analytical process, the structure of inference, and not to history. One paper in psychology likened post hoc theorizing to the farmer who shot some holes in the side of his barn, and then painted bulls-eyes around the holes. This made it look like he was a very accurate shot. But in reality he was not a very accurate shooter, but was good at covering up his poor shooting.

A lot of archaeological argumentation is like this. There is nothing wrong with coming up with a new idea after the fact, IF one then goes ahead and tries to test that idea with a new study using new data. But when the post hoc theory is the end of the line of inference, then it is a poor way to advance knowledge.