Verified by 
This is a guest post by Michael McCarthy, a lab manager and researcher at the Bias Investigation and Applied Science (BIAS) Lab at University of the Fraser Valley in British Columbia. He is applying for graduate school in psychology and neuroscience and seeking a supervisor.
As usual, the views he expresses here are his, not mine, and I need not agree with everything here. I do think it is a very good essay, though.
-----------------------------------------------------------------------------
It is naive to assume that using the scientific method guarantees scientific progress. Although the scientific method is often characterized as an iterative, self-correcting process which leads to advances in knowledge, in practice this whiggish characterization of science does not necessarily hold true when accounting for the cognitive and behavioural biases of the people actually doing science. This idea has been explored to some extent recently in discussions of the replication crisis in psychology and the need for political diversity in social psychology. However, I argue here that greater emphasis needs to be placed on the relationship between the psychological biases of scientists, the validity of scientific research, and the advancement of scientific progress.
Much of the current discussion about psychological biases’ role in the replication crisis has focused on how various external pressures have created an environment which enables—or even incentivizes—researchers to act on their biases (purposely or not), leading to many of the methodological problems at the root of the crisis (e.g., questionable research practices, researcher degrees of freedom, etc.). From this perspective, psychological bias has primarily been treated as a moderating or mediating variable in the replication crisis rather than a root cause in itself. Indeed, the lack of emphasis placed on psychological bias’s causal role in the replication crisis becomes apparent when looking at proposed solutions, which have focused overwhelmingly on addressing the technical aspects of scientific practice (e.g., open science, methodological rigour, etc.) and underwhelmingly on addressing the human aspects of scientific practice (e.g., self-selection bias, confirmation bias, etc.).
Getting the technical aspects of scientific practice right is necessary for establishing the trustworthiness of scientific observations and analyses. Nonetheless, it is not sufficient for establishing the validity of scientific research or advancing scientific progress (even though it might give off that illusion). Instead, it is the human aspects of scientific practice—the imposition of theory onto data, the refinement or generation of ideas through social interaction, and so forth—that are necessary for establishing the validity of scientific research and advancing scientific progress, since they are the foundation of the scientific method's self-correction process.
However, getting the human aspects right takes just as much (or maybe more) effort as getting the technical aspects right. It is not sufficient to simply have humans do science and hope the self-correction process happens naturalistically; humans tend to be rationalizing and irrational rather than rational (e.g., confirmation bias, cognitive distortions, etc.), and organize socially based on similarity (e.g., homophily, self-selection bias, etc.) which can further reinforce this rationalizing tendency, create echo chambers, and so forth. Thus, getting the technical aspects, but not the human aspects, of scientific practice right might only lead to the creation quasi-scientific fields filled with highly capable experts whose work reflects their shared psychological biases instead of reality.
Just as researchers have adopted standard methods for preventing the psychological biases of participants from confounding the validity of research (e.g., counterbalancing & order effects, deception & demand characteristics, etc.), there is a need for researchers to adopt standard methods for preventing their own psychological biases from confounding the validity of fields of research. Viewpoint diversity offers a start in this direction, as although it does not remove the bias blind spot and confirmation bias of individual researchers, it addresses these biases by creating an environment in which the psychological bias of one researcher might cancel out the psychological bias of another. Viewpoint diversity is not a perfect solution to our biases, but it is the driving force behind the scientific method’s self-correcting process. As such, it should be considered a necessary condition when determining the validity of a field, and a necessary condition for advancing scientific progress.
