Partition dependence bias, which is detailed well in this new HBR article, basically says that, in a context when people have to choose multiple options out of many available options, and the options are grouped together based on a given dimension, people tend to think, “Let’s choose some from each category.” Therefore, people tend to choose some options from each group, ultimately choosing more diverse options.
Here’s an example, pull-quoted from that HBR article above:
We conducted a study with 121 experienced HR professionals who had an average of eight years of HR-related experience. We asked them to download a zipped folder containing resumes of 16 job applicants who graduated from one of four top universities. In one version of the study, the order of the resumes was random and did not vary by university. In the other version, the resumes from each school were contiguous in the folder (i.e., the files in the folder were sorted alphabetically). All HR managers were asked to select four candidates to interview. We found that when the resumes were randomly interspersed, 14% of managers chose candidates from all four universities, but this number more than doubled to 35% when the resumes were grouped together by university. We found similar results when we grouped candidates by gender, either by listing them contiguously or by using a paper clip to hold their printed resumes together. We found similar results when grouping candidates by ethnicity and nationality.
Long pull-quote, and I apologize, but the general idea is “Let’s choose one from each!” The full paper is here.
Could this actually work towards diversity?
Somewhat. It’s definitely a good start on how the human brain tends to process things in front of it, i.e. grouping and patterns and whatnot. I could see it work in some organizations, but the challenges would be:
- How would it be executed? The people who wrote this paper say you should send hiring managers candidates in folders called “Batch 1” and “Batch 2” and so on. “Batch 1” would be diverse candidates, and “Batch 2” would be white guys from elite schools, or men in tech, or whatever. That way, instead of choosing everyone from Batch 2, maybe the hiring manager would grab some people from Batch 1. That could work, but I also think some hiring managers just want people who come from similar backgrounds as them, live in similar neighborhoods, have spouses that look similar to their spouse, etc. Even with science and strategy, we cannot always overcome that.
- Could HR manage this? This is the elephant in this particular room. HR people tend to automate lots of shit that should be human, sometimes because execs bought them software in hopes of eventually automating the entire department away, and HR people also tend to not manage these types of processes well, and instead roll over for whatever that specific hiring manager wants or thinks they need. Even if you have science and strategy, you need execution, and if the execution is at the HR level, I don’t know.
- Some hiring managers are just flat-out racist: Look at some of these screenshots.
We’ve been trying to solve for diversity for a long time, and…
… we ain’t really there yet. Some care, some claim to care and don’t really (probably the biggest bucket), some think that “social justice things” shouldn’t interfere with work (another large bucket), and so on. Plus, a lot of the research on bottom line diversity ties gets dismissed as “anecdotal” or “academic,” and I think we know executive-types don’t always respect academic types. Even the paper I linked above is academic in nature, and I could see ROI Ronnie getting that first link and being like “Gah, I have a 1:15 with Japan about Q4 plays to keep the lights on! No time!”
Remember: people are obsessed with being busy, cutting costs, and oftentimes just following the herd.
So could this stuff work? It could, and we increasingly need to try something, because a lot of the issues we are facing are issues of opportunity — even though we frame them as issues of other things — and if we could remove some of the biases that prevent access to those opportunities, that’s good. But: it does need to involve science, strategy, and execution, and if we whiff on that final step, we whiff on the whole thing.
