Since 2014, Candid has been collecting demographic data about the people who work at U.S. nonprofits through Candid’s nonprofit profiles. To date, over 54,000 organizations have shared some data about how their staff and/or board identify by race/ethnicity, gender, sexual orientation, and/or disability status.i
We anticipate a steady increase in organizations sharing demographic information as a result of our recently launched Demographics via Candid campaign. This initiative aims to standardize and centralize nonprofit demographic data collection. It also seeks to provide a common baseline of the diversity of the field, as well as ensure that demographic data is available to those who can make use of it to evaluate their programs and assess progress around equity.
To help uncover what this demographic data can tell us about diversity and representation in the U.S. nonprofit sector, we want to better identify who, exactly, is sharing demographic data with Candid and what information they are sharing.
Here we share four key insights from our recent analysis of Candid’s nonprofit demographic data and what they mean for the nonprofit sector today.ii
1. Demographic data sharing varies by nonprofit subject category.
Turns out that nonprofits’ subject areas play a role in which organizations share demographic data. The chart below compares the proportion of nonprofits by subject area overall (in blue) with that of the subset of nonprofits sharing demographic data (in orange).iii
For example, approximately 36% of nonprofits overall work in human services. A slightly higher proportion, about 39%, of nonprofits sharing demographic data also work in human services. Therefore, we can say that human services organizations are slightly overrepresented in the demographic dataset. Nonprofits supporting the environment and animals are similarly overrepresented. They represent about 5% of overall nonprofits but nearly 10% of those sharing demographic data.
In contrast, education-focused organizations make up more than 15% of all nonprofits, but they only comprise 10% of those sharing demographic data. Given these results, we can say that organizations working in all subject areas are sharing demographic data with Candid but that there are discrepancies in demographic data sharing rates across these various subjects.
2. Nonprofits are most likely to share data at the leader level and on gender and race/ethnicity.
To explore what type of demographic data is typically shared, we can look to nonprofits’ leaders. In the chart below, we compared sharing rates by different staffing levels and demographic categories, including race/ethnicity, gender, sexual orientation, and disability status.
Here, we see a clear trend: Over 90% of nonprofits shared some demographic data about their leader.iv In comparison, the sharing rate for all other staffing levels is below 60%.
Among demographic categories, the sharing rates are notably higher for race/ethnicity and gender.v The main takeaway: To have a more comprehensive view into the demographic makeup of the nonprofit sector, more information beyond the leader level is needed.
3. Nonprofits that depend more on contributions or have more financial resources and/or staff are more likely to share demographic data.
Dollars and cents might help predict the likelihood of demographic data sharing. As part of our analysis, we considered multiple characteristics that impact whether an organization shares demographic data with Candid.
Here’s what we found: Financial factors, such as revenue, assets, or contributions (as a proportion of overall revenue)vi, are among the most important characteristics. In other words, the higher a nonprofit’s financial resources or the more it depends on contributions, the higher the chance it also shares demographic data.
In addition, organizations’ size (by number of employees) and whether they serve racial/ethnic populations are also influential factors. vii
4. Nonprofits serving racial-ethnic populations are more likely to share comprehensive demographic data.
The populations served by today’s nonprofits may be an indicator of the amount of data they share. To dive deeper into the factors that positively affect the sharing of demographic data, we created a metric called “survey completion level”, ranging from zero to four. These levels correspond to how much demographic data any given nonprofit shares with Candid. (Here, a higher number means that more data has been shared).
When looking at how many nonprofits are serving racial/ethnic populations, we can see a 2-5% incremental increase at each level. This trend indicates that nonprofits serving racial/ethnic populations are not only more likely to share demographic data with Candid, they’re also more likely to share more demographic data.
What nonprofit demographic data sharing illuminates
These insights highlight the untapped potential that nonprofit demographic data sharing promises for understanding the diversity of today’s nonprofit landscape. As the amount of the data shared increases, we will be able to understand more about the resourcing of BIPOC (Black, Indigenous, and people of color)-led organizations and the collective progress made on diversity, equity, and inclusion (DEI) efforts.
To join the over 50,000 nonprofits already participating, simply claim and update your Candid nonprofit profile, including the section on demographic data.
i To view the questions and categories on Candid’s demographic survey, see How to collect and share demographic data.
ii We limited our analysis to registered 501(c)(3) public charities that filed at least one 900 or 990 EZ between 2017-2021 and had at least $50,000 in annual expenses according to their latest filing, as of November 2022. Of the resulting set of organizations meeting this research criteria, 32,133 (11%) shared some demographic data with Candid.
iii Here, subject area is defined by the ten broad categories under the National Taxonomy of Exempt Entities.
iv Respondents may share demographic information for one leader or two co-leaders (if applicable). In the latter case, an organization is counted as sharing leader-level data if they have shared some demographic data for at least one co-leader.
v In the chart, the category for “Gender” combines responses to two separate questions on the demographic survey. The first asks whether respondents identify as male, female, or non-binary. The second asks whether respondents identify as transgender or not. In both cases, respondents have the option to decline to state.
vi Figures for contributions come from Part VIII, Line 1h on IRS Form 990 and represents the total of “Contributions, Gifts, Grants, and Other Similar Amounts” listed in Section 1.
vii The x-axis represents how important each factor or characteristic is in the machine learning model we used. It works by randomly shuffling the values of a factor and measuring how much the model’s performance decreases, giving a score that represents how much that factor contributes to the model’s accuracy. The higher the score, the more important the factor is in predicting the target variable.