If you want to add equity to your data analysis you're going to need multiplication rather than addition.

The brilliant scholar Lisa Bowling is working on the cutting edge of applied intersectional analysis.  Her piece “When Black + Lesbian + Woman ≠ Black Lesbian Woman: The Methodological Challenges of Qualitative and Quantitative Intersectionality Research” has inspired a lot of our thinking on the topic of how to actually build statistical models that reflect an equity lens in data analysis. We’re working on demonstrating some hands-on ways to develop these models and add them to your toolkit and data products.
Check out the work of Dr. Lisa Bowleg
It is all amazing and worth your time. Homepage here.
But what about sample size?!? It's true that sample size can become an issue in quantitative intersectional analysis. As we use data about people more accurately, the amount of people in each subgroup usually gets smaller. This is a real issue. However, just for the record, it's not more real than the fact that using only additive models usually gives you the wrong answer. So. what to do about the sample size issue? 

In technical terms, this is called "sparse data bias" and basically is the bias in estimates when the data lack adequate numbers of observations for some combination of factor and outcome levels, which may arise even if the total sample size appears large.

Two ways of dealing with this that I have used successfully are using "Exact Statistical Methods" such as exact logistic regression and secondly, penalised regression. This method, using data augmentation is simple and understanable to most readers if explained carefully. It does include making judgement calls. But, realistically, so do all methods. There are several good R packages that can do this including the oem and the Penalized packages. I've been told that STATA and SPSS can also do this. If you have examples, send them my way and we'll add them.
Want to hang out IRL?

The We All Count team will be on the road this fall.
You can find us onstage at:

Consultation Chicago, Sept 9-13
Women in Data Science Conference in Seattle, Oct 3-5
The Social Finance Conference in Toronto, November 6-8
The AEA Conference in Minneapolis, November 11-15
Tableau Conference in Las Vegas, November 12-13
Conference on Statistical Practice in, Sacramento Feb 20-22

If you're in the area and would like to host a workshop, a lunch and learn, or a tea party let us know. We'd love to meet you and your team.


We're looking for equity problems and successes.

If you have a story or idea you want to share, send me a note by replying to this email.

Project for Equity in Data Science
Copyright © 2019 Datassist, All rights reserved.

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.