Nature Podcast：招聘中的歧视

Interviewer: Nick Howe

I think we can all mostly agree that looking for a job is pretty stressful. But for certain groups of people, it can also be an unfortunate reminder of the bias and prejudice that exists in the world.

Interviewee: Valentina Di Stasio

Discrimination is enduring and it’s pervasive.

Interviewer: Nick Howe

This is Valentina Di Stasio, a sociologist who has worked on the GEMM Project, a project that aims to identify discrimination across Europe.

Interviewee: Valentina Di Stasio

The research on discrimination goes back to the 60s and it has demonstrated over and over again that discrimination is pervasive. With regards to ethnic and racial discrimination, the general result is that there is discrimination in most of the contexts and also discrimination is enduring. It has not diminished over time and we have made comparisons of experiments that have been conducted in different decades and the alarming result is that there is no decrease.

Interviewer: Nick Howe

Research about discrimination has, up until now, largely relied on correspondence studies. These involve sending multiple fictional CVs to a job opening, all identical apart from one characteristic – something like ethnicity or gender. Here’s Valentina again.

Interviewee: Valentina Di Stasio

These allow us to then compare the invitations that they receive, the invitations to a job interview, for example, and then to basically understand whether there has been discrimination. These types of experiments are typically considered the gold standard in discrimination research, however this research stops at the moment of application, so we send the application and then we have no control over what happens next.

Interviewer: Nick Howe

Also, in these studies, you can’t send too many CVs or you risk interfering with real jobseekers. Scientists needed another approach and, this week in Nature, Dominik Hangartner and his colleagues have been pioneering a new method using large datasets from employment websites in Switzerland. Here’s Dominik.

Interviewee: Dominik Hangartner

So, what we’ve done in collaboration with the Swiss Ministry is that we track how recruiters on the website search which profiles of potential candidates they look at, how long they look at these profiles and who is contacted for a job interview.

Interviewer: Nick Howe

The advantage of this is that you have real-world data and a whole lot of it. Dominik and his team were able to leverage data from the Swiss Public Employment Service to get information about millions of jobseekers and hundreds of thousands of job searches by tens of thousands of recruiters. That’s a lot of big numbers. Now, things like nationality and ethnicity are not always included in job applications in Switzerland. But Dominik and his team could still make certain inferences in the same way it appears that recruiters do. In particular, they looked at three factors to make inferences about candidates – their name, their spoken languages and whether they had Swiss citizenship. In fact, Dominik was able to group jobseekers into ethnicities from nine different regions, with people who have a Swiss name, speak a Swiss language and have Swiss citizenship as a reference.

Interviewee: Dominik Hangartner

What we find is that if we compare these several applicants that appear in the same search, that people from minority ethnic and immigrant backgrounds face 4 to almost 20% lower contact rates compared to ‘native Swiss citizens’, that is people with a Swiss name who speak one of the Swiss languages and who have Swiss citizenship.

Interviewer: Nick Howe

Contact rate is a measure of how often a recruiter clicks on a jobseeker’s application to receive their contact details, and that click is directly linked to a jobseeker’s chances of being employed in the next few months. And within that 4-20% range, there appears to be a pattern.

Interviewee: Dominik Hangartner

What we found is a pattern that fits theories of so-called ethnic hierarchies, meaning that immigrant and minority ethnic groups who are more distant, be that in terms of culture or language or skin tone, face more discrimination than those groups that are more similar. In the Swiss context that we focus on, that means that we find almost no hiring discrimination for immigrant jobseekers from Italy or Spain, but a lot of hiring discrimination for immigrant jobseekers from sub-Saharan Africa or Asian countries.

Interviewer: Nick Howe

So, whilst immigrant jobseekers from western Europe may only face a small penalty in the eyes of recruiters – around 4% – people from sub-Saharan Africa and Asia face around a 20% penalty. These are from otherwise comparable CVs, what’s known as ‘observably similar applicants’. For Valentina Di Stasio, who you heard from earlier and who wasn’t associated with this study, these results aren’t particularly surprising.

Interviewee: Valentina Di Stasio

The findings from this study are quite consistent with a broader literature on discrimination.

Interviewer: Nick Howe

Dominik also used this approach to investigate gender discrimination, and it seems that this picture is a little bit more nuanced. Here’s Dominik.

Interviewee: Dominik Hangartner

Across the 70 million decisions to contact or not contact a woman or man applicant, on average we don’t see any evidence of discrimination. So, that was a bit of a surprise. But then we started to unpack it because we have so much data, and so what we find is that this average null result of no discrimination masks a lot of variation across occupation, and the pattern that we documented is very striking in the following sense. Women who apply in typically male-dominated jobs, say construction or forestry, face a lot of discrimination and much lower contact rates compared to equally qualified men. However, we also see exactly the symmetric or similar pattern for men applicants equally qualified who apply in women-dominated professions, for example in the care and the health sector.

Interviewer: Nick Howe

Dominik’s work only gives an indication of how often people are contacted for an interview. It doesn’t say anything about pay gaps or other aspects of the labour market that have been previously linked to gender-based discrimination. But to Valentina, this approach could still be very powerful to dig deeper into how different forms of discrimination interact.

Interviewee: Valentina Di Stasio

What I see as a potential of this study is that they could try to understand the role of ethnicity and gender in combination, so in interaction with one another, and this would be very interesting because we could study whether the same gender pattern can also be observed with members of disadvantaged groups, so ethnic minorities. Do we see the same pattern?

Interviewer: Nick Howe

To tackle discrimination there’s patently a lot to be done. Dominik and his colleagues’ data mostly shows associations, but it also lays the foundations for others to assess recruitment websites and perhaps try out different ways to reduce and hopefully eliminate the observed discrimination. What can we do, is what Dominik is focusing on next.

Interviewee: Dominik Hangartner

That’s one of the most important questions for me and all of my colleagues who work in this area to focus on. I think it’s important to document discrimination, it’s important to have evidence about the potential drivers of this discrimination because that has some consequences of how we’re going to tackle it. But at least my colleagues and I think that over the next months we are well advised to spend most of our time and effort and think about how we can overcome discrimination.

Interviewer: Nick Howe

That was Dominik Hangartner from the London School of Economics in the UK and ETH Zurich in Switzerland. You also heard from Valentina Di Stasio from Utrecht University in the Netherlands.