On the Crisis of Reproducibility in Psychology

As we might expect, most results in psychology are not reproducible. As the authors who obtained these results say, ‘reproducibility is a defining feature of science’. From this, we could conclude, as have many in the field, that the answer is more experiments, tweaked statistics, metholodogical tinkering and the like. Or, we could make a point that is not so much epistemically radical as it is blatantly obvious, and that is that psychology is not a science at all. It does not resemble science except in the most superficial of respects. It isn’t just the failure of replication documented here, but the complete impossibility of findings in psychology ever being abstracted into formulae for precise prediction.

Meanwhile, whilst ‘behavioural scientists’ are unsuccessfully aping their colleagues in the physics and chemistry faculties, they are neglecting psychology’s basis in the humanities. These latter disciplines are capable of constructing arguments with evidence and rigour. The findings of sociology, history and the like have many ‘real-world’ implications. Yet nobody imagines that they constitute a science, and neither is such an assumption necessary for the survival of the discipline itself. By overlooking theory in the attempt to play science, the psychologists have ended up with a discipline that mainly consists of shonky metaphysics held together by correlational statistics.

As an aside, we should take heed of the folly of using statistics to aggregate human subjective experience, rather than taking subjects one at a time. If we were to be rigorous, it should be clear that not only are psychology results nor replicable, but the same applies to the experiments themselves. The only thing that is replicated, strictly speaking, are the techniques and jargon deployed in a given study. Rather than do the hard work of conceptualisation, ‘behavioural scientists’ have imagined that they can avoid difficult epistemological methods through technocratic perseveration. There’s a great deal of statistical significance, but not a lot of significance.

The other factor here, noted in some of the press coverage on this issue, relates to the bankruptcy of the academic/journal system as a means of undertaking study. Getting university staff to meet KPIs for publication could only ever be expected to bear an incidental resemblance to doing science. It isn’t just academics who are gaming the system, however, but policy-makers and others who wish to implement disciplinary methods onto subjects using techniques from psychology. Naturally, such an exercise requires the discourse of ‘science’ to do some ideological heavy-lifting.

Advertisement