In less than a year, the U.S. Census Bureau will conduct the largest peacetime mobilization in the United States to conduct the 2020 Census. Old or young, rural or urban, no matter who you are, you will be counted. But every day, the Census Bureau field representatives are knocking on doors to conduct over a hundred smaller-scale surveys. These household surveys collect important data to inform policies and decisions that affect our communities.
Yet too often, people refuse to respond.
Our team of Harvard students is working with the Census Bureau to increase response rates so the Census Bureau can collect complete and accurate information. We conducted 27 interviews with Boston-area residents and discovered five major reasons why people don’t respond to Census surveys: lack of awareness about household surveys, concern over the legitimacy of the survey and Census Bureau representative, distrust in government, fears about privacy or data breaches, and people simply not being in the mood to respond.
Brainstorming Over 100 Ideas
We brainstormed with an open mind and came up with some wild ideas — from partnering with celebrities to endorse surveys to getting rid of household surveys altogether. We thought of ways to augment what the Census Bureau is already doing, as well as completely new ideas they had never considered before. In the end, we came up with over 100 ideas!
We grouped our ideas into similar themes and showed them to our Census Bureau partners, Lisa Clement and Ruth Chan, who helped us narrow them down to the five most interesting, implementable, and impactful solutions.
When we suggested creating new educational curricula involving the Census Bureau to teach students in school, we learned that the Census Bureau already has these resources. Yet they are not well-publicized and the public-school teachers we spoke to had never heard of them, though they would be excited to take advantage of existing lesson plans.
We ended up with four major ideas:
Redesigning the letter the Census Bureau sends to notify survey respondents
Improving respondent support pages on Census.gov
Following up with a thank you message for survey respondents
Notifying 2020 Census respondents about the existence of other surveys
After creating simple prototypes, we took them out into the hands of potential respondents – everyone around you! — to see whether they would make a difference.
If you are selected for a household survey, you will first receive a letter addressed to your home, explaining which survey you are selected for and notifying you that a field representative will visit sometime soon. In our research, we realized that many people found the letter confusing, difficult to read, or just threw it out before opening it. We wanted to design a new letter incorporating many ideas, from adding more colors and graphics to testing behavioral science theories—psychological, cognitive, and social factors in how humans make decisions.
We tested four different behavior change theories:
Vividness: Humans tend to skim rather than read—and a letter from the U.S. Census Bureau would seem to be particularly susceptible to people skimming it. Therefore, we highlighted in bold text the most critical information in the letter.
Loss framing: Humans are motivated by avoiding economic and other losses. Loss framing messages, in this context, would attempt to motivate respondents who are unaware of the societal issues associated with not completing the survey.
Social norms: Humans are motivated by peer behavior and pressure. When we know everyone else is participating in an activity, we’re influenced by that fact. We tested messages that informed the respondents that X% of the neighbors had answered the survey. This puts additional pressure on the respondents as would not want to be seen as the very few who do not answer.
Plan making: Behavioral science shows that humans are more likely to follow through on a course of action if they have explicitly made a plan. We wanted to test messages that encouraged people to make a plan to respond to the survey.
Additionally, we included a QR code (a machine-readable matrix barcode that smartphones can decipher, and load a website) to prompt potential respondents to visit the website.
We asked interviewees whether they preferred the original letter or the new one and why. We spoke with Kyla Fullenwider, the former Chief Innovation Officer at the Census Bureau, who has conducted extensive research on survey materials. Kyla’s advice and our own research will inform our next prototype, as we continue to iterate and test our new letter design.
Respondent Support Content on Census.gov
The Census Bureau’s website has several pages of support resources available to help respondents along the survey process, but this information may not be clearly communicated to people browsing through. We rearranged and reformatted the content on one of these support pages to test what details stand out most to a respondent looking for help. In our prototype, the first major section helps a person figure out whether or not they have been selected for a legitimate survey. Meanwhile, the current website emphasizes pictures and links to answers to common questions about legitimacy and data security, as well as Respondent Advocates, who are the primary Census Bureau members responsible for addressing respondent concerns. The Respondent Advocates are currently revamping this page as well, and moving forward, we will incorporate aspects of their drafts into our designs to test.
The top of the current webpage for respondent support and the top of our prototype for the same page.
From our preliminary tests, the Census Bureau got it right. Potential respondents appreciate large, direct links that address their most important questions right away. However, it is not clear to them who “Your Advocate” is, but after learning more about the Respondent Advocate, they appreciated the existence of the service and felt that it would have been useful to know that information right away, rather than having to go to a separate page to find out more.
Following up after a survey
After completing a survey, some respondents report feeling as though their information disappears, with no idea about what happens with their responses. While some field representatives send thank you cards, there is no established communication to acknowledge the important service that respondents have completed and thank them for their time. To address this, we created prototypes of a text message and a thank you card to see whether they might encourage respondents to feel positively towards the Census Bureau, increasing their likelihood to answer another survey or tell their friends about household surveys.
Our text message is short and simple, including a link to a website with more information on the survey they just completed and a fun infographic about the results of the survey. This webpage emphasized that the impact that the survey had could not have been possible without the respondent’s contribution, hoping to show that their response made a direct impact on their own community. Since our key information was on this separate webpage, we tested whether respondents would click on the text message link to learn more.
For those who did not own a cellphone or might prefer a more analog, physical memento, we created a thank you card prototype. The card includes the same infographic as the website, again creating that direct connection to emphasize the importance of answering household surveys. We also tested whether a QR code would be helpful for respondents to get more information online.
Prototypes of the follow-up text message and the physical thank you card
Our initial tests revealed that the colorful infographic of the thank you card was more exciting and eye-catching than the sparse text messages and the lengthy explanation on the website of how the government uses survey data. The potential respondents we tested with so far both described themselves as “analog people” and preferred the physical thank you card over the text message, adding that they would only go to the website if they saw the text message at a convenient time. We’re interested to see how a greater diversity of potential respondents feel about the different message length between the two prototypes and their preferred interactions between the communication channels.
Outreach through the 2020 Census
Many people already know about the ten-year census, particularly the upcoming 2020 Census. Since everyone is a participant, there is more advertising and general awareness about the importance of this survey. Taking advantage of this knowledge, we created a short acknowledgment at the end of the survey to explain that the Census Bureau’s work does not stop there. In behavioral science classes here at Harvard, we’ve learned that actively acknowledging a task makes individuals more likely to remember to do it later. Since the new Decennial Census will include an option to respond online, we created the following prototype:
Our testing indicates that this message is informative but has an unclear effect on likelihood to complete to surveys. We learned that the prior question influences response and people are concerned about potential solicitations they may receive from acknowledging this message. We iterated on the prototype based on these findings. We are excited to speak to more people to see whether these patterns hold.
We plan to keep iterating and testing to create practical recommendations for the Census Bureau. Along the way, we will talk to more Census Bureau experts to understand how our solutions can be implemented. The Census Bureau is already exploring some of our ideas, including a website redesign and follow-up text message. We are excited to continue working together to make our dreams into reality.