Achieving More With Less: Intuitive Correction in Selection

Abstract

Choosing between candidates for a position can be tricky, especially when the selection test is affected by irrelevant characteristics (e.g., reading speed). One can correct for this irrelevant attribute by penalizing individuals who have unjustifiably benefited from it. Statistical models do so by including the irrelevant attribute as a suppressor variable, but can people do the same without the help of a model? In three experiments (total N = 357), participants had to choose between two candidates, one of whom had higher levels of an irrelevant attribute and thus enjoyed an unfair advantage. Participants showed a substantial preference for the candidate with high levels of the irrelevant attribute, thus choosing the less suitable candidate. This bias was attenuated when the irrelevant attribute was a situational factor, probably by making the correction process more intuitive. Understanding the intuitive judgment of suppressor variables can help candidates from underprivileged groups boost their chances to succeed.

Algorithms in selection decisions: Effective, but unappreciated

“…The results of Experiments 1 and 2 show that providing advice led to participants' selections being more similar to what a normative model recommends. Approximately 60% of the participants (in both studies) chose normatively when presented with advice, compared to less than 20% who did so without advice (in line with the results of Rabinovitch et al, 2020). We found this effect regardless of whether the participants had already committed to a decision or not.…”

“…To overcome this bias, DMs need to adjust their evaluations of the candidates accordingly. Algorithms such as statistical models can perform this correction, but Rabinovitch et al (2020Rabinovitch et al ( , 2022 demonstrated that people struggle to do it intuitively. Therefore, we examined how willing would DMs be to be advised and assisted by external sources and whether they treat advice from human experts or algorithms differently in this context.…”

“…Sample size was determined by an a priori power analysis using G*Power software (Version 3.1; Faul et al, 2007) with a .05 criterion of statistical significance, power of .90, a .5 correlation between repeated measures, and a moderate effect size (f ) of .25. This effect size is equivalent to a difference of .2 between two independent proportions and was used in previous studies (Rabinovitch et al, 2020(Rabinovitch et al, , 2022.…”

“…To discount the effect of the bias, DMs should consider the irrelevant attributes that affected the selection tests by lowering the scores of those candidates who benefited unjustly, revealing the candidates' true skill level. Rabinovitch et al (2020Rabinovitch et al ( , 2022 found that performing such a correction without the help of a model is difficult. Only 15-30% of the participants in their studies were successful in doing so.…”

Faster responders are perceived as more extraverted.

“…It has been well documented that people in charge of candidate selection decisions such as hiring or university admission decisions are swayed by features of applicants that are not directly related to job performance. Examples include height (Harper, 2000), physical attractiveness (Hamermesh, 2011), gender (Marlowe et al, 1996), and performance on unrelated tasks (even when explicitly informed that the task is unrelated; Rabinovitch et al, 2020). Most pertinent, recruiters seem to match applicants to certain jobs or careers based on mere stereotypes of the applicant and on the degree to which these stereotypes “fit” the job description, discounting the applicant’s actual capabilities and character.…”