I am at a small college and we need to work with disaggregated achievement data. If we choose to collapse racial and ethnic groups for institutional effectiveness planning and for other reporting purposes, what are some things we should be conscious of? We want to avoid identifying individuals who might belong to an underrepresented group. We basically have 3 groups: White, unknown and everyone else. What are some appropriate labels for “everyone else”?
1 Like
I think an earlier question you might need to ask is why you’re collapsing the categories. At least in my work (usually with community colleges), we’re trying to identify equity gaps when we disaggregate the data. But I’ve seen colleges basically create the white and “everyone else” categories and find that there was almost no difference (and then pat themselves on the back for not having equity gaps), because they had a substantial Asian population that was outperforming the white students (obviously Asian students are not always the highest performing group, but in these cases they were). So they actually had a huge equity gap score between their white and Black students that was not showing up in the “disaggregated” data. So I’m not actually sure of the value of using white, unknown, everyone else as groups because it likely won’t really tell you anything valuable. That probably wasn’t very helpful (sorry!), but this is fresh on my mind because I just encountered this with a college.
4 Likes
That’s exactly what we are trying to do - identify equity gaps. You are right, that the collapsed data doesn’t tell us anything. The problem is that our population (500) is so small that the disaggregated data isn’t very helpful either. In any given cohort we will have just a few students who identify as anything other than white or, now that I really look at it, Hispanic. So, if we choose to collapse, we need at least 4 categories. What about this?
- White
- Hispanic
- Other underrepresented races/ethnicities
- Unknown
Other than that, do we just rely on qualitative data to identify equity gaps?
I agree with Dawn’s post, I don’t think reporting it out in the aggregate is going to be very useful. I’d suggest reporting it out as close to however you presented the question to participants. Also, if population comparisons are going to be important at any point, that would be another reason you would not want to report it out in the aggregate. This may be more of a question of how you weight participant responses based on their racial/ethnic identity. Pew has a good article about that: 1. How different weighting methods work - Pew Research Center Methods | Pew Research Center
I’m not sure if anything I said helps your problem at all. lol. Sorry.
I’m a little late to the game here, but in this case, qualitative data can be extremely useful, if you have or can make the resource to collect it, and can keep it de-identified. Also, you may be able to disaggregate data for more specific groups if you aggregate it over the years. So you may not be able to talk about the 4 Black students who you had this year, but you might be able to talk about the 40 you’ve had over the past ten years. You can also look at trends over time. You still need to be mindful of not identifying individuals, but it helps people understand that gaps do in fact exist.
2 Likes
@sarahferrency I was thinking about this too. Would it make sense to combine cohorts so you had a larger pool and could make some larger comparisons.
1 Like
@Pspencer Without revealing too much, any chance you could give a few examples of the kinds of gaps you are looking at?
That can be super helpful in determining useful disaggregation strategies, qualitative vs quantitative approaches, and as @sarahferrency mentioned, whether it’s valuable to start collecting nuanced categories for use in the future as your long term sample sizes increase…
Examples: are we preparing graduates in under-represented communities to find jobs; are we providing the right services to help students graduate.
1 Like
Cool, thanks @Pspencer so are you wanting to look at gaps in the proportion of job matches for different student groups? Or are you looking for to measure a different gap? (Sorry for the slow reply - came down with COVID 
