METHODOLOGY

INTRODUCTION TO EPOP

The Entrepreneurship in the Population (EPOP) Survey is a new publicly available resource for understanding the scope of entrepreneurial activities across the United States. The EPOP Survey includes a variety of measures of entrepreneurial behavior at a local level, providing policymakers and researchers with new information on the experiences and perceptions of entrepreneurs in their local community.

The EPOP Survey uses a nationally representative sample to measure current and former business ownership, whether individuals are currently taking or have in the past taken steps towards starting a business, the extent to which individuals engage in freelance work, and engagement with the “gig economy.” In addition to providing a characteristic profile of the individuals involved in these various entrepreneurial activities across the U.S., the survey also serves as a resource for understanding the behaviors, challenges, and resources available to individuals during the entrepreneurial process.

EPOP SURVEY INSTRUMENT DESIGN

The EPOP Survey allows individuals to qualify for multiple categorizations of entrepreneurship (e.g., Current Entrepreneur and Former Freelancer). Each category is derived independently and accommodates complex work situations individuals may have.  The EPOP Survey begins by capturing employment history and screening for entrepreneurial activity, the survey classifies individuals into an entrepreneurship category for follow-up questions.  For individuals that qualify for more than one entrepreneurship category, the following priority order is applied to assign a single category for follow-up questions and limit respondent burden: 

• Current Business Owners: individuals who currently own a business 
• Current Freelancers: individuals working for themselves as a freelancer, consultant, or independent contractor
• Nascent Entrepreneurs: individuals who are actively in the process of starting a business or other form of self-employment at the time of the survey
• Former Business Owners: individuals who report previously owning a business but are no longer business owners
• Former Freelancers: individuals who report they were previously a freelancer, consultant, or independent contractor but are no longer engaged in freelance work
• Withdrawn Entrepreneurs: individuals who considered starting a business, had a specific idea, and took active steps towards the endeavor, but did not ultimately start; this includes pre-entrepreneurship leavers, missed entrepreneurs, and former nascent entrepreneurs.
• Non-Entrepreneurs: individuals who have never been engaged in entrepreneurship or business ownership at any stage

The follow-up questions are structured so each entrepreneurship category receives similar questions worded in comparable ways to foster more robust analysis across the categorical assignments.

The EPOP Survey also captures incidence of gig work and asks if the primary or secondary job is gig work and an additional question is asked about any other gig work beyond the primary or secondary jobs to ensure all gig work activities are reported.  Given the potential for lack of clarity in what counts as gig work, the survey includes a definition of gig work in the main text of the question: 

“Some people earn money through short, paid tasks or jobs online or in-person that are conducted through companies that coordinate payment for the service. This is sometimes referred to as ‘gig work.’” 

The EPOP Survey includes follow-up questions about gig work which are asked of any respondent engaged in it regardless of their entrepreneurship category. 

SURVEY SAMPLE

Target Population and Estimation Objectives

The target population of the EPOP Survey includes noninstitutionalized adults 18 years or older in the United States.

The sample design supports the following estimation objectives:

• National estimates of entrepreneurial activity by demographics such as race/ethnicity, gender, age, and education, but not necessarily by the cross of these demographic variables,
• State-level estimates of entrepreneurial activity by race/ethnicity and gender, but not necessarily by the cross of these variables, and
• Metropolitan statistical area (MSA) level estimates of entrepreneurial activity for the top 50 MSAs by population by race/ethnicity and gender, but not necessarily by the cross of these variables.

Sample Design

The EPOP Survey uses a stratified sampling design to achieve the research objectives. States without a top 50 MSA constitutes a primary sampling stratum or a geography. For states with one or more of the top 50 MSAs, each MSA and the rest of state outside MSAs make a primary sampling stratum.  MSAs made up of counties from multiple states are divided into multiple primary sampling strata, one for each state.  To achieve the objective supporting estimation and analysis of entrepreneurship characteristics of underrepresented minorities, particularly Black and Hispanic individuals, within states and MSAs, each primary stratum is further divided into three secondary sampling strata: Hispanic, non-Hispanic Black, and non-Hispanic Other.  

Sample Sources

The EPOP Survey sample is selected from three frame sources: 

1. NORC’s AmeriSpeak® Panel, 
2. An address-based sample (ABS) frame built from the United States Postal Service (USPS) Delivery Sequence File (DSF), and 
3. Opt-in online survey panels. 

Samples selected from the AmeriSpeak Panel and the ABS frame are probability samples with explicit stratification and known sample selection probabilities while the sample obtained from the opt-in online survey panels is a nonprobability sample with unknown frame coverage and unknown selection probabilities. 

DATA COLLECTION

Timing 

For EPOP: 2022 data collection began on February 15 for the AmeriSpeak sample, February 28 for the ABS sample, and May 13 for the non-probability samples.  Data collection ended on June 6 for both the AmeriSpeak and ABS samples and on June 3 for the non-probability samples.

Methods

Differential data collection protocols were followed for each of the sample types.  

• The AmeriSpeak Panel sample was invited to participate by email and telephone prompting.
• The ABS sample was sent requests for survey participation by USPS letter, postcard, email, and telephone prompting.   
• The opt-in panel participants were recruited via a variety of methods including email and digital network and social media ads. 

Partially completed surveys in the AmeriSpeak Panel and ABS samples were followed-up with emails and prompting calls.

Data was primarily collected via an online survey.  Computer-assisted telephone interviewing was a secondary mode and available upon request. 

Participants were able to take the survey in English or Spanish. 

All participants were compensated for their participation.

Response Rates

The response rate varied by sample type. For the AmeriSpeak Panel sample, the response to the survey request was 38.2%, and the overall response rate was 4.9% which reflects the panel recruitment and retention rates.  For the ABS sample, the response rate was 6.0%. For the non-probability survey panels, the response rate is not reported. 

DATA COLLECTION & RELEASE SCHEDULE

The overall timeline for implementing this project is July 1, 2021 to December 31, 2026. The schedule of this project is as follows:

DATA PROCESSING

Data Review

The raw data file was reviewed and compared to the programmed survey specifications to identify data irregularities and develop any necessary code to transform raw data for consistency. Complete and partial survey records were all reviewed for data integrity (e.g., to identify respondents who completed the survey too quickly to have read question text and respondents who straight-lined responses) and those that were in violation were discarded. For a small number of cases where the data could not be repaired, the case was discarded. Similarly, cases were evaluated for item nonresponse. Those with high item nonresponse or missing critical pathway information were dropped from the final data set.

Weighting

Complete surveys from the three sample sources were combined using NORC’s TrueNorth® weighting method to generate a set of combined sample weights to support estimation. The survey contains two sets of weights:  

• Probability sample weights for the  probability samples (AmeriSpeak and ABS).  These weights are available only in the restricted use files (RUFs).
• Combined sample weights for the combined probability and nonprobability samples.  These weights are available in both the public use files (PUFs) and RUFs.

Statistical Disclosure Limitation (SDL)

Statistical Disclosure Limitation (SDL) techniques were applied to the data to protect the privacy of survey respondents including recoding, suppression, rounding and micro-aggregation. 

Appended Variables and Data File Creation

The following variables were appended to the EPOP Survey data file after processing: 

• Location variables - the nine-category census division and four-category census region variables, which are derived from respondent ZIP code
• Weighting variables – both the combined and probability weight were created to allow for calculating accurate standard errors
• Administrative variables - information relevant to survey administration, data editing, disclosure review, and sample information were appended or derived

The final PUF and RUF were created from the fully processed data set.

DOCUMENTATION

The following EPOP Survey documentation is available:

• Questionnaire
• Survey Item Crosswalk (shows pathways through the survey by entrepreneurship category)