Every research study is shaped by how it is conducted. The 2023 Leading with Intent series
rests on a survey of nonprofit chief executives and board chairs, complemented by
external data sources that provide additional insight into organizational and leadership
characteristics. The four core findings — who serves on boards, how boards understand
their work, what they prioritize, and how they prioritize their time — are grounded in this
methodology, which determines what questions could be asked, how responses were
captured, and how results were analyzed.
This Methodology brief explains the design of the study: how the survey was distributed,
who was included, and how results were analyzed. It outlines the safeguards that ensured
responses remained voluntary and anonymous, and the criteria used to focus the sample
on public charities headquartered in the United States. It also describes the integration of
IRS Business Master File data and Candid’s Demographic API, which strengthened the
dataset by adding externally verified information on mission, finances, and leadership
demographics.
BoardSource conducted a self-administered online survey using a non-random,
convenience sample of nonprofit chief executives and board chairs. We shared the 2023
Leading with Intent (LWI) survey along with partner organizations using a general
anonymous Uniform Resource Link (URL). The survey was open to all chief executives and
board chairs of nonprofit organizations, and responses were collected between June and
December 2023.
Although the sample was non-random, efforts were made to distribute the survey broadly
to enhance sector-wide representation. The CEO survey included 96 questions covering
board composition, culture, meetings, policies, and board and organization performance.
The board chair survey shared many of the same questions as the CEO survey, and
included additional questions about oversight of the executive, but was limited to 64
questions.
Survey questions were developed based on prior iterations of the LWI study and refined
through internal review and feedback from nonprofit governance experts to ensure clarity,
relevance, and alignment with current sector challenges.
In exchange for completing the survey, respondents were offered a free PDF publication of
their choice. Survey responses were collected using Qualtrics.
Participation was voluntary. Respondents were informed of the purpose of
the study and their rights. All data were stored securely in accordance with BoardSource’s data protection policies.
Respondents provided their organization’s employer identification number (EIN) which,
after an additional checking process, was used to match organizations in our sample with
external supplementary data sources.
Data on each organization’s annual income, assets and mission area were obtained from
the Internal Revenue Service Business Master File extract (IRS BMF) sourced from the
National Center for Charitable Statistics (NCCS).1 NCCS performs additional data
checking on IRS BMF files.2
For organizations that participated in our survey, we obtained additional data – on board
member demographics, the CEO’s gender, gender identity, sexual orientation, and
disability status – from Candid’s Demographic API. At the time of analysis, Candid data
was last updated between August 2019 and January 2024 for the organizations we
surveyed. In their Candid profile, nonprofit organizations voluntarily fill out information
about their board size, CEO and board demographics. Due to the voluntary nature of this
data collection, a considerable number of organizations either did not complete their
profiles or only partially provided this information.
To be included in our final sample, participants had to complete at least half the survey
and indicate that they led a public charity headquartered in the United States. We also
verified organizational classification using IRS BMF data to ensure alignment with public
charity status. Because this study focuses on the governance of public charities, we
excluded all associations and private foundations/grant-making foundations.
To maintain independence between individuals in our sample, responses from the CEO
were kept in cases where both the CEO and the board chair from the same organization
completed the survey. For organizations with multiple responses from the same executive
or board chair (17 cases), we kept the most complete response or the earliest response.
We examined variable pairs using one of the Kruskal-Walli’s test, Chi-Square test,
Spearman correlation test or the Cochran-Armitage trend test depending on the nature of
the variables examined. More complex relationships were assessed using logistic
regression, proportional odds regression or partial proportional odds regression depending
on the nature of the variables and checked for assumptions. All statistical analyses were
conducted in R version 4.2.1. Unless otherwise stated, all results included in this report
are statistically significant at the p<0.05 level.
As with many research efforts, there are limitations to our methodology. The use of a
convenience, non-random sample may limit the generalizability of our findings to all public
charities in the United States. In addition, due to the survey’s cross-sectional nature, we
cannot determine causation in any relationship between two or more variables. We had a
minimal level of missing data for variables collected directly in the survey (less than 5.4%
for board chairs and less than 7.1% for chief executives in all cases) and for variables
obtained from the IRS BMF (less than 5%). However, for demographic data obtained from
Candid, there were many cases of missing or partially missing data. According to Candid,
organizations that filled out their profiles were less likely to respond to more sensitive
questions regarding board members’ disability status, sexual orientation, and gender
identity.3
Despite its limitations, this study benefits from several methodological strengths that
enhance the depth and reliability of its findings. First, by surveying both chief executives
and board chairs, we capture dual perspectives on board governance, enabling a more
comprehensive understanding of leadership dynamics. The survey instruments were
robust, with a wide range of questions covering board composition, culture, performance,
and oversight. Additionally, the integration of supplementary data from the IRS Business
Master File and Candid’s Demographic API allowed for triangulation and validation of self-
reported data, enriching the dataset with externally verified organizational and
demographic information. Clear inclusion criteria ensured that the final sample focused
exclusively on public charities headquartered in the U.S., maintaining relevance to the
sector. Statistical analyses were conducted using appropriate tests and regression
models, with all reported results meeting a significance threshold of p < 0.05.
The 2023 Leading with Intent series is grounded in survey data from executives
and board chairs, supplemented with IRS and Candid data that enrich understanding of
organizational size, mission, and demographics. This combination creates a dataset broad
enough to capture sector-wide patterns and detailed enough to surface gaps in equity and
representation.
Like all studies, this one has limits. Its non-random sample constrains generalizability, and
voluntary demographic reporting leaves gaps. Yet the strengths — dual perspectives from
executives and board chairs, robust survey instruments, and triangulation with external
datasets — make the results a meaningful mirror for nonprofit governance today.