Methodology

The insights presented herein are the product of a collaborative effort by our cross-functional team of domain experts. From Left to Right: Richie Eaddie, Babacar Diatta, Hayden Mattick, and Dillon Duff

Methodology and Background

Leadership

Here we evaluate the gender composition of three levels of employees. We can access this data as the IRS requires organizations to report on leadership only. Note: some organizations also disclose their non-binary employees.

Pay

Equal Pay is a workplace issue that women care most about–according to research performed for Gender Fair by ABX, surveying 2,500 Americans. In addition to revealing gender pay gaps, our ratings show the ratio between average pay and top salary, which is significant as 20 percent of nonprofit employees struggle from paycheck to paycheck.

Diversity

The United States encompasses a diversity of cultures and ethnicities that are not typically represented in this sector. Nonprofit organizations are not required to report on workforce diversity as companies are, although a petition to the IRS is now circulating to change that. A small percentage of nonprofits voluntarily report this information to the Candid organization, and we rely on this reporting for our ratings.

Scoring Methodology

51 Points are based on Gender Diversity In Leadership:

33 Points are rewarded based on Pay Equality:

16 Points for Overall Diversity:

Percentile Scoring

To evaluate each organization we divided the performance data into five distinct groups based on percentiles—these are essentially benchmarks that split the data into five equal parts. Depending on an organization's performance in a specific metric, such as gender diversity or pay equality, it falls into one of these groups. Each group corresponds to a different score, ranging from lowest to highest. So, an organization performing in the top percentile group would score the highest possible points for that metric, reflecting excellent performance, while a company in the bottom percentile group would receive the lowest score, indicating significant room for improvement. This method ensures that the scoring is both fair and reflective of how each organization's efforts compare to others in the same field.

After individual metrics are scored and scaled, they are summed to produce a final score for each organization. This aggregate score represents an overall evaluation of the organization's practices in gender diversity, compensation equity, and transparency. This requires a balanced approach across all key metrics to achieve a high score.

How Can Organizations Be More Fair?

SOURCES

Our team sourced gender and pay data from IRS 990 Schedule J forms. Data on diversity is derived from Candid’s database; nonprofits can opt in to share their demographics, yet only 10 percent of organizations choose to do so. If a nonprofit does not share demographic data with Candid, their score drops by 16 points.

Four computer science students aggregated and conglomerated 50 Gigabytes of data (equivalent to about 1.3 million pages of paper) and used artificial intelligence (GPT) to categorize the organizations. Gender identity was determined probabilistically using a gender name dictionary from Harvard, found here: Harvard Dataverse. The work can be found at rhit-duffdl/GenderFair.

Contributors:

Acknowledgments:

Gender Fair would like to thank: the Faculty of Computer Science at Rose-Hulman Institute of Technology--notably, Dr. Gloria Liou; Dr. Jean Camp, Director of Center for Security and Privacy in Informatics, Computing, and Engineering; and Candid.

<