Value judgments

Very often we get angry when we observe real discriminatory behavior against certain groups of people. Just as often, however, we get angry and judge as discriminatory a choice or behavior that might not be. This is because we are sensitive to the way in which a certain situation is presented to us (it is the so-called framing effect), we tend to ignore relevant aspects to formulate a judgement, or we judge based on a general rule that seems to us "the right one". An article was recently published in the journal PNAS (Proceedings of the National Academy of Sciences) which demonstrates how much the perception of discriminatory behavior depends on the way in which it is represented, regardless of whether it is whether or not actually.

The authors conduct a series of experiments assigning the participants to one of the following conditions: “hired” – “rejected”. All participants are told that: 1) a certain pool of candidates contains 2 groups with the same qualification, but the groups are unbalanced in terms of composition (e.g. 25% of candidates are female and 75% male); 2) the decision maker accepted half of the candidates and rejected the other half. Furthermore, the participants in the "hired" condition are informed of the composition of the hired candidates only (e.g. 25% of the hired candidates are female), while the participants in the "rejected" condition are informed of the composition of the rejected candidates (e.g. 25% of rejected candidates are female). Finally, all participants are asked to express a judgment on how discriminatory the decision maker was.

It is important to underline that the information in the two conditions are equivalent: a) everyone is informed of the composition of the pool of candidates (eg 25% women and 75% men); b) everyone knows that the decision maker has accepted half of the candidates and rejected the other half. Thus, participants in the “hired” condition are informed about the proportion of minority group candidates who have been hired (e.g. 25% of hires are female) and can therefore easily infer the proportion of minority group candidates who have been rejected (in the example above, 25% of the rejected are female), and vice versa. Simplifying: if there are 8 candidates, of which 2 are women (25%), if the decision-maker does not discriminate by gender (he hires half of the female candidates and half of the male candidates) and if 25% of the hires are female, then 1 woman out of 5 total recruits was hired (that is, 25% of those recruited) and consequently 1 woman out of the total 5 rejected (which is always 25%, but this time of those rejected). In most of the experiments, the authors simplify the calculation process by setting both the minority group's percentage of hires and minority group's percentage of rejects equal to that group's proportion of the applicant pool (25%, in the example above). Furthermore, in most of the experiments there is no "objective" discrimination, i.e., continuing with the previous example, the decision maker will always hire half women out of the total candidates and half men out of the total candidates.

The results of the experiments show how people tend to judge the decision maker as more discriminatory towards the minority in the candidate pool if people see the composition of accepted candidates than when they see the composition of rejected candidates.< br/>
This difference in perceived discrimination occurs even in the absence of objective discrimination. The authors explain these results by saying that, when judging discriminatory behavior, people do not compare the composition of recruits with that of rejects or vice versa (as they should), but focus only on the information provided in their condition and neglect the information in the other condition, even though they could easily derive it. Instead, they tend to compare the information received about the percentage of the minority group that was hired or rejected with an ideal composition, and expect equal representation of the recruited groups (e.g., equal proportions of men and women): if the composition observed deviates from this expectation of equality, people are likely to view the decision maker's choice as more discriminatory.


Perluigi Conzo
NP December 2022