Labor market discrimination is rampant. Every research that’s done, at least here in the Netherlands, tells the same story. Even though the Netherlands is very egalitarian, open and multi-cultural, our unconscious bias when it comes to selecting candidates from resumes makes all the difference.
Earlier this year research from several different universities showed that even criminals convicted of violent crimes have a bigger chance of getting invited to a job interview than Mohamed or Abdul with a perfect record. For their research, they send out exact replicas of cover letters and resumes where the white names admitted in their cover letter to different types of crimes.
People that admitted to having served time for violent crimes got invited to an interview 28% of the time, whereas migrants with no criminal record only got invited 9% of the time.
Last year a Dutch Ph.D. student analyzed 440.000 resumes (he had access to all the anonymized resumes of one ATS provider, that has hundreds of clients in all kinds of markets) and had an algorithm do correlations. After teaching the algorithm the human behavior (he knew who got invited to an interview and who eventually got selected) he let the algorithm select from the second part of the sample. The algorithm was 80% correct when selecting based on a resume only.
The biggest correlation the algorithm found in human behaviour when selecting candidates was age.
Testing and data is the solution
Recently I came across some really interesting data from Harver. For those who do not know them, they build pre-selection testing for contact centers and retail sales staff. Basically, they let their applicants play a game an during this game they measure the quality of the candidate for the specific job. So for contact center applicants, they measure, among other things, the ear-hand coordination and spelling qualities (and about 20 more relevant skills for the job). Every candidate gets scored on several different aspects of the job, resulting in a total ‘fit for the job’ score based on testing. Many of their clients don’t even ask for a resume anymore since it has no predictive value what so ever and the test scores have lowered attrition at many of their contact center clients from over 100% to less than 20% a year.
They recently did an analysis of 130.000 Dutch applicants from whom they knew who was hired. The divided these candidates into 2 different groups: ‘native’ and ‘migrant-background’ based on the first names of the applicants. Basically the same way a recruiter looks at a resume. The results were staggering.
- Of all the native applicants, 22,585% were hired.
- Of all the migrant-background applicants, 22,505% were hired.
These results were not correlated with the actual scores in the test, but since most of Harver’s clients reply mainly on the test scores, it’s safe to assume they all scored well. The very small difference (0,08%) in the percentage of hires this shows that hiring based on testing and data rids the recruiter of his/her unconscious bias.
Replacing a resume with actual test scores and data about the quality of the candidate seems to rid us of unconsious bias.
The future of recruitment is away from the resume. This is a future is much more inclusive that actually selects candidates on their abilities instead of our biasses about their abilities.
This post originally appeared on LinkedIn. It has been published here with the author’s permission.