A recent article has been making waves in psychology in which scientific replicability in the field is placed in doubt.
Even in social
psychology reproducibility can be important, as in the famous work of Solomon
Asch on conformity that was confirmed in multiple replications. However, with a
slight theoretical shift, his results became part of a much larger perspective
on opinion-formation by Serge Moscovici’s idea of minority influence. That is, Moscovici
does not “correct” Asch’s work, it enlarges it and gives it nuance. I use this well-known
example because it seems that the problem of “truth” in psychology is not necessarily
statistical: it has more to do with fine distinctions and the complexity of behavior.
Another problem has to
do with all the tests that are supposed to measure things like intelligence, attitude,
personality, school assessment and the like. There are well-known criticisms of
this kind of testing that deal with things like cultural exclusion and the
limited definitions of the field being evaluated (for example, musical ability
is excluded from intelligence tests). I mention this as just another issue in
which complexity makes any broad statements about the utility of these tests
impossible.
I don’t want to go
into other issues where objectivity is not even a goal in psychology. Rigor is
always a problem, and evidently there must be rules that make absolutely clear
what is being discussed and to what degree it might be representative of
something. As the multi-authored, re- published article from Science says in
its conclusions: “Any temptation to interpret these results as a defeat for psychology
or science more generally, must contend with the fact that this project
demonstrates science behaving as it should”.
Reference:
Alexander A. Aarts, Joanna E. Anderson, Christopher J. Anderson, Peter R. Attridge, Angela Attwood, et al (August, 2015) Estimating the reproducibility os psychological science. Science
Vol. 349, no. 6251. Disponible en: http://www.sciencemag.org/content/349/6251/aac4716.full
