The importance of defining diversity (and how to do it)!
One of the great joys of social science (yes I am starting off that strong) is getting to be precise about what we mean by multidimensional, often abstract concepts.
My early-grad-school self would be shocked to hear me say this. When I first entered my Ph.D. program, I was excited to contemplate big-picture topics, like the inevitability of international warfare (I was, and remain, a joy to be around). What I encountered, however, was semester after semester of (smart) faculty members asking me: What do you mean by that?
At first it was infuriating. What do I mean by war? I mean a whole bunch of people killing each other. How many is a bunch? Does it matter whether the people killed are soldiers or civilians? Or why they are killing each other, or who told them to, or over how many days they do it? If a bunch of people kill 20 people on one day, is that a war? What if they kill one person per day for 20 days?
Are you glad you’re reading this series yet?
I eventually realized that if I ever wanted to finish my dissertation, I needed to go to therapy. I also realized I needed to define my terms. It was not easy. Putting limits on what I was studying felt like giving up my Quest for Truth. Studying “war” feels important. Studying “instances of militarized international disputes that led to at least one death of a soldier per day over as many days as there were deaths” totally sucks. There’s a reason it’s called War and Peace, not Militarized International Disputes (MIDs) and the Lack Thereof.* There’s also a reason no one talks to me at cocktail parties.**
Likewise, putting limits on the kinds of diversity we want to understand at companies feels like the opposite of what the spirit of diversity is supposed to be about. But with diversity as with war (that timeless comparison), we can’t hope to understand what causes it without defining what we mean by it. And if we can’t think carefully about what causes it, we are likely to waste a lot of time and money implementing policies that address the wrong problems.
Often, defining terms is actually pro-diversity. For example, many companies purport to mean all dimensions of diversity ever, but then only focus on gender. In this case, the failure to define terms actually makes them more exclusionary, because they are tacitly implying that gender is the only dimension of diversity that matters.
And, defining terms keeps companies from weaseling out of accountability on some dimensions. For example, the most common response I get to pointing out that, say, all of the company leaders are men, is, “Yes, but look at all the cognitive diversity!” (The second-most-common response is “Who are you, again?”)
So how do you define diversity in a concrete way without excluding people? The key is that no definition is perfect, it just has to be useful for at least one project.
(Of course, from a legal and ethical standpoint, the minimum non-discriminatory requirements need to continue to hold. But when it comes to trying to figure out how to improve diversity, you’re allowed to focus on just one or two dimensions at a time — as long as you’re clear and transparent that this is about focusing on one thing for now to understand the causes and improve lives of employees from those groups — and that this by no means indicates you value people from other groups less.)
In language that we’ll get to in future articles: You’re just defining the dependent variable for your future analysis.
To be concrete about it, here are the three questions I have found useful for companies when they form their own definition of diversity:
1. Diversity with respect to what?
There are lots of types of diversity that we could stand to improve. There are the talked-about ones like gender and race, and even those can be expanded into many more dimensions.**** And there are many more, like disability or veteran status, socioeconomic class, personality type, work style, age, native language, education experience, and effectively infinity more.
What’s the right set of dimensions? Part of it depends on why you want to be more diverse in the first place (next week’s topic), and part of it depends on what’s actually feasible for you (this might be controversial, but I’d rather see a company make progress on a few realistic goals than name every dimension of diversity ever and then do nothing about any of them). When in doubt, make your definition finite: For this project only, I want to focus on why women are leaving the company. Then, expand to something else after that.
If you’re really lost at sea, for company-wide priorities, I say start with five dimensions that span the following categories: Demographic identity (e.g., race, gender, age), cognitive identity (e.g., education, personality), financial identity (e.g., class, debt, financial oblications ), and family identity (e.g., parental or caretaker status). Why these four? Because they touch on many (but not all) of the possible reasons you might want diversity. Why five? Because six seems like too many, and four seems like too few. (A dirty secret about science is that it’s about precision, which at some point often boils down to making a relatively arbitrary judgment call.)
2. What’s the denominator?
Most companies measure diversity at the level of the entire organization, but this can be deceiving. For example, Citi (one of the few major companies that releases diversity data every year, and for that I applaud them), on paper has tremendous gender diversity. They are the only company I can name with more women than men overall. But if you look closer, it turns most of the women work in Citi’s call centers. Similarly, Nike’s numbers (again, bravo to them for sharing their data) on race are impressive. But it turns out, most of their black employees work in retail.
In these two companies, if you measure diversity at the organizational level, they’re doing great. If you measure it by function, they’re doing horribly. For many companies who purport to be interested in the business case for diversity, what they really mean is diversity at the team level. A “team” can be a very tricky thing to measure in the first place, but it’s where I would start if I were in charge.
3. What does success look like?
If I’ve learned anything about the corporate world, it’s that “OOO” does not mean “order of operations.” The other thing I’ve learned is that companies love benchmarking. (Also, no one drinks the office coffee, even though it’s free.)
Benchmarking is, I think, about trying to understand what “good” looks like in other companies, and then using that to set your own standards for what you should do. I think it’s ok to get a sense of the landscape, but I don’t think it’s particularly useful for thinking about what you’re capable of or what you’re looking to do. (This is basically the diversity/corporate version of stop comparing yourself to others, which, as we all know, is impossible thanks to Instagram)
Instead, I recommend every company think independently about what they’d like to achieve, what’s actually possible for them, and why they want to do it. Those are the harder questions that, if you can answer them, will help you be precise about how you’ll know if diversity is getting better in your company.
Again, if you’re lost, start with something like: Take a look at the current diversity on the dimensions you’re prioritizing at entry levels of the firm. Then look at that representation at the top, with respect to the unit of analysis you defined above. Then, shrink that difference by an amount you think you actually can without tokenism each year.
I’m finally done talking now
If all this sounds like SMART goals to you, you’re not wrong. Social scientists aren’t the only ones who’ve realized that being concrete and specific is the way forward. If your goal is to “get healthy” you won’t get anywhere. If your goal is to eat more greens and do yoga once a week (I am a walking parody of myself), you have a fighting chance.
To paraphrase the great Skip Lupia, science is a “commitment to precision.” My equally catchy version is, “let’s all commit to transparently and thoughtfully conceptualizing and operationalizing the things we want to understand and change.” Feel free to print that out and hang it on your wall! I made a meme for you already.
Next up: Why we might care about diversity (it’s not all social justice and business), and then after that: why we need to focus on inputs rather than outputs. In other words, now that I’ve made you define your terms, I’m going to ask you to set them aside and think about what’s driving them, rather than focus on changing the numbers directly. Are we having fun yet?
* There’s a whole other great social scientific literature out there about peace as the absence of conflict versus positive peace, which is a relationship where two countries work together to achieve more than they could alone. DM me for deets. Is that how you say that?
** And it’s not just because no one invites me to cocktail parties. Please invite me to a cocktail party.
*** We have terms for this in social science called conceptualization and operationalization, but people’s eyes tend to glaze over whenever I talk about them (even though they’re amazing), so I’m relegating them to down here. Conceptualization means being clear about what you mean by your term. For example, by “democracy” do you mean some set of freedoms, the decentralization of power, or elections? That choice has big consequences for what counts as a democracy in your research, and ultimately what policies you’d encourage a democratizing country to focus on.
Operationalization means how you measure it. Is one election enough to call a country a democracy? Does it matter how competitive the election is? How do you tell if one election is more competitive than the other? There’s no right answer for how to measure anything, but a good rule of thumb is to look at the cases you do have and see what your measure is ruling in or ruling out. For example, if you require three fair elections in a row to consider a country as a democracy, are you ruling out countries you think ought to be considered democratic? In that case, try two fair elections in a row. This is just one example of how, despite our commitment to precision, we also are all using our own judgment at many stages in any kind of scientific research (this is not limited to social science, by the way).
**** For example, even a focus on “people of color”, while laudable, means ignoring the different experiences and challenges for black, Latinx, Asian, and multi-racial employees, and even that list ignores the differences between, say, East Asian and South Asian employees, and even that ignores the differences between, say, Chinese and Vietnamese employees, and even that assumes we can meaningfully lump nationalities of people together.
