A recent blog article by UGM Consulting in Australia has highlighted some important research from the USA regarding the impact of programs to counteract unconscious bias in the workplace. Unfortunately, the provocative title of the article and the conclusions and recommendations which the authors seek to draw, are in no way supported by the research that they cite!

Their proposition is simply that unconscious bias training should be jettisoned in favour of other forms of workplace intervention, lest the unconscious bias training itself do more harm than good. We will respond to this proposition in this post.

The first piece of research that the authors cite by Cialdini (2005), does not deal with biases as such but rather on the ways of optimising impact when seeking to change undesirable social behaviour. After conducting experiments in attempting to deter visitors to a national park from stealing petrified wood, the research concludes that a ‘descriptive normative’ message – that theft is commonplace – is actually less likely to reduce theft (and may in fact increase it) compared to an ‘injunctive normative message’ – that such thefts are disapproved of.

The authors suggest that the reason for this is that when endeavouring to bring about change to objectionable behaviour, one should focus on the desired outcomes rather than simply emphasizing that the bad behaviour is widespread. Simply focusing on the bad behaviour has two disadvantages: firstly it projects a mixed message (“this behaviour is bad, but everyone is doing it”) and secondly it may be inaccurate (the actual absolute quantity of stolen wood was relatively low).

The later research study by Duguid and Thomas-Hunt (2014) that was cited in the blog post does deal directly with stereotyping of the conscious and unconscious varieties. This study consisted of a series of tests to determine how respondents’ stereotypical attitudes towards certain groups would be impacted by advance information about whether stereotypes about these groups were either widespread or rare.

Consistent with the earlier study by Cialdini, they found that where respondents proceeded on the premise that stereotypes were widespread, they tended to exhibit more biased attitudes than when advised that stereotyping was very rare. It did not make any difference if the participants were admonished not to let themselves be influenced by stereotypical thinking.

The critical qualification, however, which is ignored in the UGM blog post is the following conclusion by Duguid and Thomas-Hunt: “…Our findings suggest that awareness of the general pervasiveness of stereotyping behaviour does not mitigate stereotypic expression, and in fact may have the opposite effect of increasing stereotyping. Nevertheless, stereotyping behaviour may be mitigated by heightened awareness to other’s efforts to work against stereotyping.” (in other words, by focusing on the efforts of others to achieve unbiased behaviour). Duguid and Thomas-Hunt go on to acknowledge that there is existing research in the literature which shows that raising unconscious bias awareness can cause a significant reduction in individual stereotyping behaviour, and they suggest that their study “qualifies” this previous research.

This rider furnishes the ideal basis for a solution to the potential negative consequences of providing generalised information about the fact that we all have bias in a training program. Indeed, no unconscious bias program worth its salt would simply inform participants that bias exists and leave it at that. That would be a job half done. Furthermore one could easily argue that the very fact that a group of 10 or 20 leaders are gathered in a workshop for the express purpose of finding ways to counteract bias, is in and of itself a clear message by the organisation  to every leader regarding the undesirable nature of biases.

Unfortunately, UGM concludes (without any empirical data to support their proposition) that most unconscious bias training must be increasing the levels of bias in the workplace. Accordingly they recommend that taking an “influencing approach” rather than examining the nature and prevalence of bias is the preferred route to addressing, for example, gender imbalances and stereotyping in the workplace.

This is an astonishing and unwarranted conclusion. There is a wide variety of programs designed to modify objectionable practices in the workplace in areas such as resource wastage, sustainability, unethical financial behaviour, workplace hazards, sexual harassment, bullying and so forth – most of which are built on the pillars of:

  1. Explaining the nature of the undesirable practice;
  2. The degree to which it occurs;
  3. Its effects, costs and general undesirability;
  4. Inculcating cultural and behavioural change

Taken to its logical conclusion, UGM’s argument suggests that it is always counterproductive to bring to the attention of employees that levels of unacceptable behaviour are high, because the inevitable consequence is that the bad behaviour will only become worse.

A more plausible approach suggested in this article written by thought leaders Sheryl Sandberg and Adam Grant in response to the same research, is to acknowledge the relevance and appropriateness of pointing out the stereotyping and biases, but warn that this should not be seen as “legitimising prejudice”. Instead, it should be information which provides a foundation to move to: “…reinforcing the idea that people want to conquer their biases and that there are benefits to doing so…”

Sandberg and Grant assert that people should be encouraged to correct for biases and this can only be done if we acknowledge that all actors in the workplace are subject to explicit and implicit biases.

In conclusion, we at Symmetra would contend that there is no reason to abandon unconscious bias training that is properly designed as part of a multi-pronged behavioural change initiative. Our experience with thousands of leaders across the globe over the last few years shows just the contrary – that starting with unconscious bias training as the first step of a culture change process towards embedding inclusion can “turbo boost” change.  When leaders actually become aware of the extent to which unconscious bias can undermine the quality of their decisions, we have found they become highly driven to develop their skill to counteract bias, as well as to (re)design systems and processes to render them more impervious to biases, as they see this as a direct means of optimising their business performance.