This report presents the results of tests conducted in February 2020 using five television and three radio advertisements for the 2020 Census. The tests used quantitative procedures to better understand the effectiveness of the ads, as well as how the Census Bureau might use this methodology for ad testing in intercensal years and in preparation for the 2030 Census.

Each ad was embedded within a four-minute clip from a television program. Respondents were randomly assigned to either watch a census ad or a 30-second “control” ad before being asked about attitudes and intended behaviors related to the census and the advertisement. We then compared the answers by ad shown to evaluate the census ads.

After watching or listening to the clip of the show with the commercial break, respondents were asked a series of questions about their attitudes and intended behaviors toward the census. We compared census ads to the control ad across several questions including intention to respond to the census, intention to recommend others to fill out the census questionnaire, Census Bureau corporate image, attitudes toward the census, and efficacy to respond to the census, among other measures.

At the end of the questionnaire, all respondents were reshown a Census Bureau advertisement and asked to evaluate the advertisement. We used answers to these questions to compare the likeability, clarity, offensiveness, and teachability of different census advertisements.

Significance testing was done using a linear model predicting the outcome variable from the ad shown, with age, sex, Hispanic origin, race, marital status, household size, renter status, income, and region as covariates.

In total, 10,000 respondents completed this survey. Respondents were recruited from a nonprobability panel, and quotas were imposed for age, sex, race, income, education, and region to obtain an experimental sample more generalizable to the U.S. population.