Statistical Seats at the Table

Discussing diversity is hard. I've written about it several times here on Gamasutra, always with an eye toward applying reason and objectivity, along with respect for individuality along with the business drivers that make diversity efforts both important and practical. As with everyone who enters into these conversations, I expected and encountered heated differences of opinion, but my goal was always to keep conversation possible.

The recent Google Manifesto turned up the heat on the always-simmering diversity debate. I watched as that short screed successfully made dialogue difficult again, exactly the opposite result anyone would seek who was sincerely trying to inspire meaningful change (as the writer claimed). I refrained from leaping into the Manifesto debate because I was asked to avoid stress for health reasons... but I need to speak out about one aspect of it--the aspect I consider most harmful in terms of hindering meaningful discussion.

Some of the most heated arguments around the Manifesto are over assertions that women are statistically less likely to enjoy engineering/tech and whether they're also statistically less likely to be biologically suited to succeed in engineering/tech. That aspect of the debate is a distraction from more meaningful, relevant conversations--a derailment of progress toward any kind of understanding.

More important, it's meaningless in this context. It makes us so mired in arguing about whether data is valid that we don't stop to think about whether it's relevant and applicable. Statistics are irrelevant when you’re dealing with individuals, especially once the individual’s position in the statistical pool has already been established. I'm a living example.

When I had a second cancer that recurred after treatment, statistics showed the main drug in the clinical trial I entered was effective in less than 15% of patients. I've been in the trial for almost a year. In the data just published, there were fewer than 3% complete responses to the trial drug combination, representing a single patient. I’m cancer-free for the moment. That single patient with a complete response is me. 

Imagine debating with me now whether I should continue trial treatments. “Only 15% respond at all, and less than 3% have a complete response,” you tell me, “so I don’t know if we should put any effort into keeping you in the trial.” That would be absurd, right? We already know I’m in the 3% complete response pool, so it’s meaningless to debate now whether I’m suited for the trial. Yet that’s exactly what’s happening when we discuss statistics about whether women are “suited” for engineering as a part of a conversation about diversity hiring and support.

Even if it were true that the majority percentage of women are inherently less suited for engineering/tech*, becoming an engineer--or professionally engaged in any tech occupation--isn't accidental or trivial. It takes years of training, hard work, keen problem-solving skills, and a particular way of thinking. It takes dedication and effort to go through a tech job hunt, nail an interview, and then do the kind of good work that leads to further opportunities.

In conversations about diversity hiring and support for women in our tech workplaces, we’re discussing an already-narrowed field: the percentage of women who ARE suited for engineering. We're not walking down the street, randomly pulling women into tech interviews. We're not cold calling random households to invite women to sit next to you and write full-stack code. In the context of diversity hiring and supporting diverse candidates after hiring, the process of segmentation has already occurred. It happened in her first algebra class, in Intro to Computer Science, in the three game jam projects she coded, in her decision to apply to your company, and in the company’s offer for her to work alongside you. She’s already passed all the “tests” and burdens and barriers to “prove” she’s in the percentage that’s suited for tech work.

Whipping out statistics about what percentage of women are “suited” for tech is meaningless when you're talking about efforts to hire women into technical roles or how to better support them after hiring. It's irrelevant because we already know they're in the segment that qualifies, just like I’m already known to be in the 3% segment with a complete response to the cancer clinical trial. And even if the statistics were accurate and valid* and only a subset of all women were "suitable" for engineering roles, that “suitable” percentage of women represents a huge pool of qualified, talented potential employees. 

Assumptions based on statistics--even valid ones*--are no better than stereotypes, and using them as an excuse to treat others with less sincerity or respect isn’t “scientific.” It’s also not good business. The truth is that any individual who has the talent, discipline, and skill to enter a tech career deserves equal treatment, equal consideration, equal support, and equal respect. If you’re a tech manager, you know how hard it is to hire and retain talent. Why would we not make every effort to refine our hiring and retention policies to be effective for every pool of qualified individuals available?

In my opinion, it’s no coincidence that a manifesto trying to rely so heavily on statistics also includes a call to “de-emphasize empathy.” Reducing human beings to statistics makes it easier to exclude them. Reducing empathy makes it easier to follow through on that exclusion in the name of “objectivity.”

Like cancer surgery, statistics operates in a world of clean boundaries, and that binary look at the world is often appealing to tech-minded folks like us. Resist the urge to segment people into binary pools of “valid” and “invalid” or “suited” and “not suited.” We’ve all put in the work to sit at the table and we all belong here. We are all in the relevant segment.

* There is no credible, widely accepted scientific evidence that women are biologically less suited for or talented in engineering/tech.