We've been talking a *lot* about collecting demographic data for the purpose of #DEI work at @CultureAmp.

We're updating our internal collection & storage practices, and collecting feedback to provide more targeted guidance to our customers.

This topic is nuanced af. THREAD.

*Lots* of companies focus on creating exhaustive lists for each category of demographic data.

I especially see this for gender and sexual orientation. I strongly, strongly believe that this is a BAD IDEA.

You should use the *minimum* number of categories you can, honestly.

Why? Because for the purposes of firm audits and examining employee experience trends, you're NOT trying to capture each individual's identity perfectly. You're trying to create groupings that allow you to see important between-group variations.

Fundamentally, while agender and nonbinary people have *meaningfully different* individual experiences, optimizing for perfect identity captures means you lose your entire ability to examine (and thus act on) differences between their experience & cis (etc.) folx.

This is a *very* different problem than Facebook (where 52+ genders make sense), because you are *not* optimizing for individual expression (they are).

You're looking to be able to create patterns you can respond to!

Now, on collecting race / ethnicity data. If you're based in multiple countries, it's complex (and I'm not your lawyer, plz talk to yours!).

This seems to be the most complicated issue I hear about, for all the reasons.

First, race is an inherently unstable construct (that sometimes but not always overlaps with ethnicity) that's sensitive to the particular historical, political, and social context in which it arises.

BUT there are global contours and categorizations that are meaningful.

There are tradeoffs with creating a global ontology (larger n sizes! ease of collection! vs. language that's a bit foreign to every region) vs. optimizing for local customers (employee familiarity vs. fragmented data).

Multiraciality adds MOAR complexity. It's an art to decide.

Disability is another that causes a LOT of questions, discussion, and frankly consternation.

There's not an agreed upon global framework. I see lots of companies collecting specific information (e.g., uses mobility aids, mental disability, autism, etc.)

That gets complex not only b/c of different local customs for referring to types of disability and/or needs, but also because you may need to think about healthcare privacy law like HIPAA (does your HRIS comply? Do your processes?)

My personal recommendation is to focus on the impact of the disability (mental, physical, both, none) and leave information on specific to an articulated process (higher privacy, provides the support individuals deserve).

Same balance of individual vs. trend capture as gender.

Often: "This categorization is othering to me"

The solution there is to be *explicit* in your communication what you're optimizing for. It's about responding to org patterns, not describing individuals.

Say this. Often. It helps.

#AMA. I think about this all day rn.