OCISR Response to Public Comments for ACS NSC

American Community Survey Methods Panel Tests, Online Communications: Improving Survey Response Campaign

Response to Public Comments

The Census Bureau received one comment on the American Community Survey (ACS) Methods Panel, Online Communications Improving Survey Response Campaign Federal Register Notice (OMB Control Number 0607-0936). The response received had two main comments. First, the commenter recommended that the test oversample for racial and ethnic groups “in order to learn how communities of color (including Asian Americans) will respond” to the test, especially given “vastly different internet access rates” among subgroups. Second, the commenter raised a concern that the test methodology “assumes that people will not click on banners that contain direct messaging about the Census” and requested that a fourth experimental panel be added with banners with promotional messages about the 2020 Census. The Census Bureau considered these comments but don’t feel that they can be accommodated for this test, as indicated below.

Demographics of Test

The ACS is the Census Bureau’s largest survey and has a sample big enough to support this research. Every month, the ACS selects a random sample of addresses and subsequently collects sociodemographic information about the people who live there. Since this test is part of the production ACS the sample is already set. To effectively oversample any groups would require additional sample which is not in the budget for this test. The Census Bureau is committed to reaching traditionally hard-to-count communities and continuously researches the best methods to reach these populations. Therefore, during the analysis phase, we will analyze the results (dependent on sample size) by race, Hispanic origin, educational attainment, and other variables related to a person’s likelihood-to-respond. Overall, this test will allow us to determine if digital advertising can positively affect response rates for the ACS in a cost effective manner. An increase in response rates saves taxpayer money and leads to increased data accuracy for all populations, including the hard-to-count.

Lack of Clickable Banners

First, we must distinguish between 2020 Census messaging and Census Bureau messaging. The Census Bureau collects information throughout the decade, not just during the decennial census. This test is unrelated to the 2020 Census. Please note, however, that the 2015 Census Test in the Savannah, GA area did include digital advertising and targeted digital advertising for hard-to-count groups with successful results. We chose to embark upon this experiment to see if we could leverage those results from a simulated census environment in a sample environment, namely using the ACS.

Additionally, past research yielded two important findings, 1) Increases in awareness lead to increases in response, and 2) The “Census Bureau” name is more recognizable than the “ACS.” While we are using the ACS to test digital advertising, our hope is that we can use it for all surveys conducted by the Census Bureau. Therefore, the advertisements contain Census Bureau awareness building language rather than identifying the ACS by name. There are three advertising concepts and each one focuses on the impact data collected by the Census Bureau has in your community, on roads, jobs, schools, healthcare, etc. The ads also depict responding by Internet, paper, or phone and illustrate specific statistics generated by the Census Bureau. By increasing awareness of the Census Bureau and asking respondents to participate if selected for a Census Bureau survey, it is our hypothesis that when the viewer receives the ACS mailing, he or she will recall the positive association and be more likely to complete the survey.

In conclusion, the ads do feature Census Bureau specific information but not 2020 Census specific messaging as this is not a test for the decennial census. Furthermore, while we only hope the ads are viewed and recalled, people who see them can also click on the ad and will arrive at a specially designed landing page that is a gateway to all of the vast information the Census Bureau produces as well as highlighting the page where a respondent can link to online response.

11/16/2016 1