Polygon Research Blog

How do you measure diversity?

MCGI Mondays Data Science
We have seen many leading companies and organizations in the U.S. financial services and housing industries announce a commitment to diversity and inclusion (D&I) and/or and diversity, equity, and inclusion (DEI). A number of prominent organizations have started their implementation of these initiatives.

In this post we discuss how stakeholders who have announced D&I or DEI initiatives could use data to inform goal setting and progress measurement. And the first obvious question is,

What data?

If your goals pertain to workforce or homeownership diversity, the starting point is simple - use U.S. Census data. But there are so many data sets and choices on the Census Bureau site that people often get confused. Which survey to use - the decennial census? The Current Population Survey (CPS)? The American Community Survey (ACS)? And if ACS, which one - 1YR, 5YR? PUMS or summarized, and if the latter, which tables to use? Out of the gates, there's a lot to sort out here.

Polygon Research Data Choices

In our interactive dashboards we do not use summarized data - we use microdata (PUMS data). This decision allows us to unlock the data and answer multiple questions and answer them rapidly. Our starting point is ACS-1YR data set because we want to track changes over time, for example, year-over-year changes. ACS 1YR is a large sample set of ~3.5 million households every year, and we've included 5 years of it in our CensusVision app. However, among its many secondary disruptions, the pandemic interrupted the normal data collection practice of the U.S. Census Bureau in 2020 causing the Bureau to release (only) experimental weights for the year, cautioning that these should not be used for decision-making as usual.

The good news is that the Census Breau also releases the Current Population Survey (CPS) on a monthly basis to inform demographic trends as well as labor trends. We model the monthly CPS microdata in CPSVision. In addition, to inform other use cases (e.g. homeownership and income analysis), we also include the Annual Social Economic Supplement (ASEC). Finally, we also model and include the annual Veteran Supplement to inform analysis of the U.S. veteran population.

Here are the industry sectors that are available in CPS that are relevant to the broader housing industry and can be interactively analyzed in CPSVision:
  • Banking and Related Activities
  • Construction
  • Insurance Carriers and Related Activities
  • Non-Depository Credit and Related Activities
  • Real Estate
  • Savings Institutions, incl. credit unions
  • Securities, Commodities, Funds, Trusts

What metrics?

Racial and Ethnic Diversity

Racial and ethnic diversity is a combination of race and ethnicity and in CPSVision it is analyzed using the "Minority Detail" variable. A key metric for measuring Diversity is the minority population as a percent of the labor force. When you compare this metric to the minority population as a percent of society, you can see if your industry (or your own company, or even both) is on par with the rest of society, or if it tends to be skewed to a particular demographic group. For example, let's assume that your company competes in the Non-Depository Credit and Related Activities industry sector. Using CPSVision to crunch through the data, within seconds we perform the following analysis (figure 1). In this industry, 62.4% of all labor participants are Non-Hispanic White, and the rest, 37.6%, are racial and ethnic minority population. From here, more comparisons can be made by each race/ethnicity depending on the D&I and DEI strategic goals of individual industry stakeholders. Moreover, this type of metric can be filtered further for Occupation (e.g. Loan Officers), or geography, or many other dimensions available in CPSVision.

Figure 1: Racial and Ethnic Minority Diversity

Source: Polygon Research: CPSVision

Age Diversity

A second metric for measuring Diversity is the share of a certain age group or "generation" as percent of the labor force. For each such metric, you can narrow your focus based on occupation, geography, level of education, or any other meaningful dimension for this type of analysis.

Ethnicity Diversity

Ethnicity, part of the Minority Detail dimension described above, can be analyzed on its own using the Census designation "Hispanic or Latino" vs. not "Hispanic or Latino". The metric for measuring Diversity here is the Hispanic population as a percent of the labor force. Here too you can narrow down your focus based on occupation, geography, level of education, or any other meaningful dimension for this type of analysis.

Race Diversity

Beginning in 2000, the race question on the Census questionnaires changed substantially to allow respondents to report as many races as they felt necessary to describe themselves.  In CPSVision, we model the race field into 6 categories:

  1. American Indian and Alaska Native
  2. Asian
  3. Black or African American
  4. Native Hawaiian or Pacific Islander
  5. Other 2 or more races
  6. White

Here too, the metric for measuring Diversity is the population of a certain race as a percent of the labor force. Similarly, you can narrow down to focus on occupation, geography, level of education, or any other meaningful dimension for this type of analysis.

Sex Diversity

The metric available in the Census data here for measuring Diversity is the share of Male and Female as percent of the labor force. Again, you can narrow your analysis to focus on occupation, geography, level of education, or any other meaningful dimension for this type of analysis.

Going beyond all of these ways to measure demographic diversity, Census data allows you to explore diversity based on veteran status, immigration status, educational attainment, and other characteristics of our population.

Deriving metrics for your own organization from data analyses of industry diversity and society diversity will help your organization:

  1. Keep track of progress of D&I and DEI strategy implementations.
  2. Improve accountability for results of D&I and DEI projects.
  3. Tell your story to potential recruits and gain their commitment to your organization.

How to get started?

Anyone can download and start to model Census microdata. But for the fastest time to value, the easiest and most efficient way is to subscribe to CensusVision and CPSVision. You will immediately access rich demographic, industry, and housing information delivered in a series of interactive dashboards. 

Sign up to receive our blog posts right into your inbox and to keep ideas and perspectives flowing into your organization, and/or schedule a 15-minute free brainstorming session with us.