Thursday, September 20, 2018

Implicit bias is another political perversion of science

The phenomenon of unconscious or implicit bias training has reached New Zealand. Companies that offer the training make the claim that "scientific research has demonstrated the existence and prevalence of unconscious bias – unconscious beliefs and attitudes that go beyond our conscious perceptions of ourselves and others" and that businesses "need to give people the platform and tools [i.e. their training] to begin to mitigate bias". Some of my clients have purchased this training for their staff so I decided to look into the science behind its claims.

The foundational research for implicit bias was a 1995 study [PDF] undertaken by American psychologists Anthony Greenwald and Mahzarin Banaji. They established the Implicit Association Test (IAT), which purports to show, amongst other things, that white people regard black people as more of a threat than those of their own race. The research has received much criticism in recent years and its conclusions have been challenged in numerous other studies, most notably a 2016 meta-analysis of more than 500 studies over 20 years involving 80,000 people using the IAT, which concluded that:
    1. The correlation between implicit bias and discriminatory behaviour is weaker than previously thought
    2. There is very little evidence that changes in implicit bias have anything to do with changes in a person’s behaviour.
    Interestingly, one of the authors of the meta-study was psychologist Brian Nosek, who worked on the original implicit bias research with Greenwald and Banaji and helped create the IAT. Nosek warns in a recent article in the Chronicle of Higher Education of the "very weak overall" connection between implicit bias and discriminatory behaviour. The researchers concluded that "IAT provides little insight into who will discriminate against whom, and provides no more insight than explicit [i.e. self-admitted] measures of bias."

    The popular Canadian psychologist, Jordan Peterson, also warns that the original study did not control for the less pernicious psychological trait of novelty aversion. This is the tendency of all human beings to prefer that which is familiar. It explains why we prefer the foods, music, clothes and people that we know and that we grew up with, rather than those we don't know. It is also why most people tend to marry within their social, economic and ethnic groups. This trait is universal and applies equally to all races, and in order to control for it when testing for racial bias, researchers would need to use only subjects who were brought up and live amongst equal numbers of people of other races, which they certainly did not do in this case.

    It is clear that this is another area where science has been perverted for political and personal ends. Implicit bias is another pseudo-science act in the political left's playbook, while at the same time becoming a nice little earner for consultants who exploit well-intentioned or politically-fearful business owners and managers. It is notable that, in the Chronicle article linked above, both Greenwald and Banaji resort to the common behaviour of all politically-biased scientists, using both ad hominem attacks (e.g. Greenwald says of the lead author of the meta-study, Hart Blanton, that "He’s not a great scientist") and defamatory generalisations (e.g. Banaji likens IAT doubters to climate-change deniers) when their conclusions are challenged. That is a sure sign that their science has become too politicised and is on shaky ground.

    No comments: