What’s the correct breakdown of software developers by race / gender / etc? I think if we can answer that question we can then accurately assess whether or not those attributes are acting as filters. However, without knowing those breakdowns, taking any action to correct those breakdowns is wrong. We can’t accurately say there is a problem.
The best we can do is remove any bias we find at an individual level (ie a blind hiring process). Looking at aggregate level metrics is nearly useless because we cannot accurately account for all the factors contributing to them.
The best we can do is remove any bias we find at an individual level (ie a blind hiring process). Looking at aggregate level metrics is nearly useless because we cannot accurately account for all the factors contributing to them.