National Academies Press: OpenBook
« Previous: 2 Background and Context
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

3

Participation

This chapter summarizes the presentations and discussion in two sessions (one on September 27, the second on September 28) that focused on survey participation. Michael Davern chaired the first session on Maximizing Respondent Retention, and Sunita Sah chaired the second session on Participant Engagement: Insights from Behavioral Science Research.

MAXIMIZING RESPONDENT RETENTION

Maximizing Response: The HILDA Survey Experience

Nicole Watson, a survey methodologist and associate professor at the Melbourne Institute within the University of Melbourne, has worked on the Household, Income and Labor Dynamics in Australia (HILDA) Survey since 2000.1 The HILDA Survey had 7,700 responding households in Wave 1 in 2001, and added an extra 2,100 households in Wave 11 in 2011. HILDA interviews everyone in the household age 15 and older, conducting interviews with around 17,000 people each wave. For a two-adult household, the interviews take 85 minutes total. About 90 percent of interviews are done face-to-face and 10 percent are done by telephone. Interviewers also ask respondents to do a self-completion questionnaire. Fieldwork starts in July and runs until February. Incentives have changed over time, but it is currently about $45 U.S.

___________________

1 For more information, see https://www.dss.gov.au/our-responsibilities/families-and-children/programmes-services/the-household-income-and-labour-dynamics-in-australia-hilda-survey.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

Watson provided a figure showing re-interview rates achieved in the HILDA Survey. Figure 3-1 shows the portion of respondents in one wave re-interviewed in the next wave by 5-year age groups; deaths and people who move abroad have been removed. She noted that, similar to other panel surveys, the first few waves have higher attrition and a lower rate of re-interviews. For example, the first black line in Figure 3-1, which are 15- to 19-year-olds in Wave 1, shows that 80 percent of this group were re-interviewed in Wave 2. By Wave 5 they were re-interviewing 95 percent of the previous wave respondents within that age cohort. Among respondents in their 50s through early 70s, the re-interview rate is around 96 to 98 percent by Wave 5.

Watson described the results of research looking at hard-to-get cases. They examined whether the hard-to-get cases who are interviewed are different from the other interviewed cases; whether the cases that require a lot of effort in one survey wave require a lot of effort in all survey waves; and whether easy-to-get cases can be re-weighted to eliminate the biases that arise from not interviewing the hard-to-get cases. Hard-to-get survey respondents were found to be distinctly different from the easy-to-get cases; however, 80 to 90 percent of those hard-to-get cases are easy to get in the next wave. Watson said that they could curtail the fieldwork without noticeably affecting the population estimates if they kept the number of calls to households to 12, and the biases that were introduced by doing that could

Image
FIGURE 3-1 Proportion of previous wave respondents who were re-interviewed at each wave shown separately by age group.
SOURCES: Nicole Watson workshop presentation, September 27, 2021; data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey.
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

be rectified through weighting. Curtailing other field activities resulted in greater cost savings. However, there were impacts on population estimates and estimates of change that were not well addressed by weighting.

Watson described another study that looked at how to re-engage previous wave nonrespondents using the HILDA Survey together with the British Household Panel Survey and the German Socioeconomic Panel. Around one-quarter of HILDA Survey nonrespondents are converted to being interviewed at each wave, she noted. The study examined different aspects of their survey fieldwork to see what encouraged people who are nonrespondents to come back and what encouraged people to continue participating from one wave to the next.

Watson highlighted the results (Table 3-1). As an example of what was learned, she pointed to the first line of the table, which shows that having the same interviewer for a person who was previously a nonrespondent has a negative effect on their likelihood of responding again but a positive effect on retaining participants from the previous wave. Watson said, therefore, they want to change the interviewer for nonrespondents but keep the same interviewer for previous respondents. The only measure that has a positive effect for both groups was to interview face-to-face rather than by phone.

Focusing on re-engaging nonrespondents, Watson noted that the longer since the last interview, the less likely these nonrespondents will return; refusals are harder to convert than other reasons for nonresponse; if the sample member changed address, that change in the household could result in a new situation that made them more amenable to being interviewed;

TABLE 3-1 Results of Different Data Collection Procedures and Events on Re-engaging Nonrespondents and Continuing Participation of Respondents

Re-engaging nonrespondents Continuing participation
Interviewer-respondent continuity Negative Positive
Change in survey mode No effect Negative
Face-to-face (vs phone) Positive Positive
Long time since last interview Negative n/a
Refusal (vs other reason for nonresponse) Negative n/a
Partial household response No effect Negative
Change of address Positive Negative
Experienced (vs new) interviewer if changed and no change of address Positive No effect

SOURCES: Adapted from Nicole Watson workshop presentation, September 27, 2021.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

and if an interviewer has to be changed, sending an experienced interviewer convinced more nonrespondents to re-engage than sending a less experienced interviewer.

In summary, Watson said that chasing hard-to-get cases in the HILDA Survey is worthwhile, because hard-to-get is not a persistent state, people tend to fall into the hard-to-get situation temporarily, and they will become generally easy-to-get in future waves. The hard-to-get cases are distinctly different from the easy-to-get cases, and weighting does not always correct for the biases in the estimates, especially when looking at estimates of change. She noted that re-engaging nonrespondents is a different decision than continuing participation, and different factors affect re-engagement versus continued participation. She suggested considering the feasibility of changing interviewers, although it is not always possible. The next best thing, she said, would be to arm the interviewers with a toolkit that provides them with a number of fresh approaches to the household, giving them different material to present to the sample member. She also suggested trying to re-engage the sample members as soon as possible, and that when respondents move, that can be a good opportunity to re-engage.

Panel Attrition in High School and Beyond

Eric Grodsky is professor of sociology and educational policy studies at the University of Wisconsin-Madison and co-principal investigator (PI) of the High School & Beyond Midlife Follow-up Study (HS&B FU). Rachel Canas is a senior research director in the Health Sciences Department at NORC at the University of Chicago, where she currently serves as the assistant project director for HS&B FU. Grodsky began by saying an advantage that HS&B FU has for understanding selection and response in surveys of aging is that sample members were contacted a long time ago as adolescents. The adolescents were also clustered in high schools, which provides additional understanding of who does and does not continue to participate in studies as time goes by. They also have standardized assessments of academic achievement and cognitive skill in adolescence, which lets them evaluate the relationship between prior cognitive skills and attrition, a process especially relevant for studies designed to understand cognitive change, cognitive decline, onset of Alzheimer’s, and Alzheimer’s disease-related dementias.

Grodsky said that the target population for HS&B was students enrolled in 10th or 12th grades in the United States in the 1979–1980 school years, so they were born between 1961 and 1964. In the clustered sample design, 1,122 schools were selected for the sample and 1,105 participated (with replacement), including public schools, nonreligious private schools, and religious private schools. The schools were oversampled to make

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

sure there were adequate numbers of Cuban-serving public and Catholic schools, other Hispanic schools, Black Catholic schools, and other non-Catholic private schools. From each school a maximum of 36 seniors and 36 sophomores were selected for the base year, and 84 percent of the sampled students are included in the data for that first year. The initial interviews were conducted in 1980.

Grodsky noted that the focus of the study is on attrition from the panel that was selected from the 1980 base year sample in 1982, and that panel included about 12,000 seniors and about 14,830 sophomores. This panel was followed up in 2014–2015 when the sample members were typically 50 to 55 years of age. They removed those who were out of sample due to mortality, illness, or institutionalization, but these sample members are carefully considered in the substantive work done on mortality as well as the nature of the surviving sample.

The attrition analyses did not distinguish among active refusals, passive refusals, and locating problems, Grodsky said. Predictors were the social background of participants, including sex, race/ethnicity, and socioeconomic status. He noted that the socioeconomic status variable was a combination of father’s occupation, father’s and mother’s education, family income, and the presence of household items such as books. Measures of academic achievement were also used, specifically math test scores and self-reported grades in the senior year of high school. They also used the school mean math scores, centering student math scores around the school mean to distinguish between differences in attrition that could be associated with differences in the types of schools students attended and differences related to the students’ own performance relative to their peers in schools.

The results of the analyses showed that women were more likely than men to respond by 6 percentage points, and relative to non-Hispanic White respondents, Black, Latinx, and respondents of other races were about 5 to 7 percentage points less likely to respond. Students who reported earning lower grades in high school were less likely to respond to the survey 35 years later, with the difference between those reporting earning A’s responding 6 percentage points higher than those reporting earning B’s and C’s, and 16 percentage points higher than those who reported earning D’s or below. Math test scores predicted participating in the survey at mid-life with each standard deviation increase in math test scores associated with about a 6 percentage point increase in the probability of response. In addition, each standard deviation increase in the school average socioeconomic status (SES) increases the probability of response by 2.8 percentage points. Within schools, there is a 1.9 percentage point bump for students for each standard deviation higher than they are relative to the school mean SES. So, he concluded, both the school mean and individual SES contribute a little bit to the probability of response in 2014–2015.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

Grodsky reflected on the results and stated that these differences speak to the challenges to consider in thinking about the distribution of early life cognitive skills and noncognitive skills among people contacted later in life. He concluded by noting that in addition to the biases he described, there is also panel conditioning bias that biases the sample members who are observed over time.

Canas described how the study team took into account what they learned in previous rounds to encourage participation for a new round of data collection in 2021. They do extensive pre-field locating efforts on HS&B FU panel members because they cannot do inter-round communications like other panel studies. Instead, they devote a large part of the data collection effort to pre-field locating efforts in which they email, mail, and call sample members to alert them of the upcoming study with the goal of learning where they are, updating contact information, and approaching them about the study that will begin in a few months. An additional level of effort individually searches for panel members using property searches and Accurint. They are also seeking IRB permission to reach out to sample members via social media on Facebook, as well as reaching out to alumni organizations to see if they are willing to let people in those graduating classes know that they may be asked to participate in a study done by NORC at the University of Chicago.

Canas said multiple modes of response for HS&B FU are offered: telephone, a web survey, or a paper survey. According to Canas, everyone starts with an assignment to a singular mode and then switches modes about halfway through the data collection; the opportunity to do a last-chance paper survey is offered at the end of data collection. She explained this ensures that they are targeting people in modes that they would prefer and enjoy participating in using the mode respondents used previously.

Canas also described the mix of pre- and post-incentives to gain participation for both the survey portion as well as new biomarker components. These two parts are treated separately so panel members receive an incentive for participating in the survey and then an incentive for participating in the biomarker data collection. Through an adaptive design, differential incentives are provided for people who are likely to participate based on past rounds and other characteristics, such as those Grodsky had noted were related to panel attrition. They are about to launch some increased incentive packages.

Because HS&B panel members have never participated in a biomarker data collection in this study before, a number of different efforts will be made to get people interested, agree to consent, and participate in biomarker data collection this round. Based on a pretest, they reframed the blood draw to a health visit, which includes a blood draw plus some health measures and a saliva sample. Reframing it in this way made people a little

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

more interested, she reported. The web survey included a video in which one of the PIs, Jennifer Manly, talked about the importance of the health visit. This was done to put a face to the research and emphasize its importance to address lower rates of consent on the web.

Canas concluded by noting that refusal conversion efforts have helped increase the consent rates and interest in the biomarker aspect of the study. If panel members say they are not interested at the first contact, a field interviewer team calls them back and lets them know about the health portion, answers questions, and provides an increased incentive.

Retention in Longitudinal Studies of Aging

Pamela Herd is a professor in the McCourt School of Public Policy at Georgetown University and a PI of the Wisconsin Longitudinal Study (2010–present). She described the Wisconsin Longitudinal Study (WLS), which has been tracking about one out of every three Wisconsin high school graduates from the class of 1957. Siblings were added starting in the 1970s and more fully in the early 1990s. The study participants are currently in their early 80s, so Herd emphasized that she is talking about retention in a life course longitudinal panel, especially for much older samples.

Herd noted that WLS response rates over time have been about 80 to 90 percent, with higher response rates among the primary graduates and slightly lower response rates among the siblings. The primary sample basically had a 100 percent response rate at baseline, so they have information on everyone in the sample frame. Roughly two-thirds of their living sample are still involved in the project.

Herd said that the data collected cover basically every aspect of the panel participants’ lives, and biological data added over time include saliva samples and data on the microbiome, as well as administrative data like Social Security earnings and Medicare data. Herd said that some of this information has been important for retention, as it helps the researchers track participants over time. She also noted that engagement over the years has fostered participants’ sense of involvement in the project.

Herd described a new study, the WLS dementia project. They have two waves of data collection over a 5-year period, and each wave includes a preliminary half-hour phone survey. They conduct a follow-up in-person survey with people who they believe are at risk for dementia. Because of COVID-19, the in-home survey has temporarily been converted to the telephone. She noted that although there is a lot of interest in assessing cognitive change and cognitive outcomes, it does add to the burden on study participants. The team is figuring out how to minimize the burden and keep participants engaged when some of what the study participants are doing may be more challenging for them.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

Herd described the characteristics of participants at higher risk of dropping out of the study, such as lower educational attainment. Similar to the HS&B FU described by Grodsky and Canas, the WLS has a measure of adolescent cognitive functioning, and it is a strong predictor of attrition over time, with those who have lower cognitive functioning more likely to drop out. This has implications for studying dementia and cognitive change in later life, Herd said, but having baseline data can help adjust for nonresponse bias.

Herd also highlighted the challenges of doing phone interviews and trying to assess cognition and hearing decline much later in life, because by the time people are in their 70s and early 80s, these conditions are fairly prevalent. In-person surveys are needed, and even they are challenging. Poorer health and hearing issues were related to higher attrition from the study. Another challenge Herd noted was that with older samples, some respondents have different residences in the summer and winter. She said this affects high-SES rather than low-SES participants, but it is more difficult to keep track of them because they move.

Herd also described an example of how they subsample for different projects within the broader sample and the implications of nonresponse. She noted that in the microbiome study, they found differences between respondents and nonrespondents within the subsample. She also said it is important to pay attention to this because the kinds of weighting adjustments that one might do are trickier with a small sample.

WLS researchers take a number of steps to retain participants. Herd stressed the importance of repeated contacts both to figure out where participants are living and to keep some kind of engagement or connection between them and the project. The materials communicate clearly to participants and provide meaningful engagement about what are they contributing to the research, giving them a connection with the project and what the researchers are learning from them. She said this matters for long-term panel engagement. She added that data on participants specifically geared toward figuring out where they will be for the next round is very important because people move. Older adults may not move for work, but they often move in with their children or into environments with more caregiving support. She also said that the quality of the interviewers matters in terms of effectively getting people to keep participating and engaged.

Herd emphasized that although response rates matter, it is selective attrition that is of concern, so they do targeted nonresponse follow-up among subsamples to reduce bias or better model to account for attrition. Targeted doorstep visits are effective when other modes have failed. More burdensome kinds of interviews make it harder to get people to engage again because they expect another long interview. Thus, they are very careful about minimizing burden.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

Herd shared examples of materials sent to participants and noted some participants contact them and request more information. Although they do not have explicit data, they receive feedback from participants that they feel loyalty toward the project the longer they have been involved in it, which matters for their continued participation. Similarly linked is a sense of belonging to something bigger than themselves and the feeling of making a contribution to the greater good. Herd said it would be useful to explore survey experiments to see how much these feelings matter for studies like WLS, HS&B FU, and others.

Herd concluded by describing challenges going forward. In terms of managing attrition and nonresponse in lifelong panel studies, she referred to two approaches already discussed in the workshop. The first is trying to keep people in, but she stressed the importance of drawing on administrative data to fill in when they lose track of people. The other approach is developing modeling techniques around selection to deal with nonresponse. She noted that longitudinal studies have extensive data on people who were lost due to mortality or attrition, and leveraging those modeling techniques to adjust for bias is worth continued work.

Discussion

Davern started the discussion period by asking the presenters what they saw as the most important emerging ways of engaging and retaining respondents over the last 4 or 5 years. As a follow-up, he asked what strategy did they consider likely to be the most important over the next 2 to 5 years.

Herd acknowledged the tradeoff between cost versus retention and response, and the high costs while maintaining sample quality. In her view, one of the most critical things is keeping some element of in-person recruitment, whether that is shifting toward in-person interviewing for people who cannot be reached through other modes, or very selective targeting of in-person efforts to manage nonresponse bias. She said the continued use of in-person data collection is critical, but needs to be targeted given how expensive it is.

Canas noted from a lower-cost perspective, email and texting in panel studies have proven to be reasonably effective. HS&B FU has a web survey, and emails have provided the biggest bump in completing those web surveys, in addition to texting. Grodsky concurred with Herd about the need to maintain contact and develop identity and rapport with sample members. He noted a challenge between funding cycles is the lack of a built-in mechanism for sending out birthday cards, newsletters, and occasional communications, which are not terribly expensive but could yield substantial rewards back in the field in terms of the representativeness of the sample and response rates.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

Watson agreed and suggested building rapport with respondents, showing them how relevant their stories are, and feeding back some of the results, all of which are becoming more important.

John Phillips returned to comments made by Herd and Carol Ryff (see Chapter 2) regarding loyalty or trust. He asked whether pamphlets about what the study is finding or other updates on the studies build trust. He questioned whether these efforts could have an impact on things like willingness to consent for a biomarker, apart from the loyalty built from being in the panel for a long time. He noted that HRS began doing biomarker collection after participants had already been in the study for quite a while, which may have a meaningful effect on trust just by itself. He asked the presenters for their thoughts on trust and the ways to build trust in a study to not only maintain a panel, but also to improve initial response rates.

Ryff responded that she would frame the issue more in terms of rapport than trust, because trust implies more complex interactions. She suggested they could study participants’ responses to the newsletters more systematically and what they think they are getting out of them. She noted that her study’s newsletters are very high quality, with a science journalist distilling the research information into engaging and readable four-page brochures. She was unclear whether the newsletters cultivate loyalty, but they do give the participants a sense that they are contributing to important science, which she said is a huge hook. However, Ryff said she does not favor distributing the newsletters to some people but not others to test effectiveness. She noted that giving back the science is an important part of taxpayer-funded research, and is also part of research translation. For that reason, she added, the MIDUS newsletters are posted on the study website.

David Weir commented that MIDUS has great results and produces excellent reports; however, other studies may not have the resources or breadth of interesting results to share. He has observed that HRS participants want information that will make them healthier. He suggested different studies could collaborate to generalize findings so that they can provide research on health, not necessarily tied to a specific study. He noted that while participants like to see that they are contributing directly, they also could be getting something back from the experts conducting the study. Herd countered that one challenge to this idea, which Grodsky noted earlier, is panel-conditioning effects. When her study sends out findings, she noted, they are thinking carefully about what they are conveying and how they are conveying it. She called for more research to understand exactly what participants want. She noted that conversations with participants in their microbiome study completely changed the strategy for how and what was communicated to them. Davern concluded the session by noting the importance of listening to the respondents to increase retention.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

PARTICIPANT ENGAGEMENT: INSIGHTS FROM BEHAVIORAL SCIENCE RESEARCH

Recruitment and Consent of Vulnerable Populations

Bettina Drake is a professor of surgery at Washington University School of Medicine in the Division of Public Health Sciences and the associate director of Community Outreach and Engagement at the Siteman Cancer Center. She is a cancer epidemiologist and health disparities researcher with expertise in community-based research. She began by noting that diversity in research participation is needed in order to conduct the most valuable research possible and that includes diversity by race, rural/urban status, and geographies, as well as individuals who have regular access to care and those who live in medically underserved areas. The reason, she continued, is to obtain generalizable results so that interventions and treatments can have the biggest impact in communities and patient populations, but also address disparities across race, ethnicity, and socioeconomic status.

Drake described studies focused on recruitment to biorepositories, but underscored these strategies apply broadly. She noted that consent for the secondary use of biospecimens is both a critical ethical issue and a policy issue. Patients are rightly concerned about how long the samples will be stored and what research is going to be conducted. She said that the literature provides little empirical evidence for donors’ preferences or different models of informed consent.

The first study Drake described was to understand barriers and strategies to improve participation in biorepositories for African American men, specifically for recruiting a prostate cancer cohort. Recruitment was made at the time of diagnosis with consent to grant access to their medical records, completion of a questionnaire, and collection of biospecimens. The biospecimen was collected at the time of treatment, and it was a consent to allow the specimens that were already being removed or blood draws that were already happening to have additional tubes of blood drawn to allow those specimens to be included in a biorepository. Initially they had very low minority participation. She said that about 70 African American men were included in focus groups, about half of whom were prostate cancer survivors.

Drake shared some of the themes and quotes that came out of the focus groups identifying strategies to increase participation. Participants suggested not recruiting at the doctor’s office because patients might be in pain or thinking about other things. The focus groups suggested talking about participation to people at younger ages so that when they end up in a health care setting, it is not completely new information. Mistrust and privacy were key themes that emerged, and some people were concerned

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

their specimens would be taken and used for personal gain. Henrietta Lacks was brought up in a couple of the focus groups, Drake reported.

Drake said that another theme was that education about what the biorepository is and what the researchers were trying to do was important. Some participants said that anything that can further research that can help their family members was important, while others commented that as long as there is not a lot of pain involved in the procedures, they would be willing to participate. Physician endorsement of the study was helpful to address the lack of trust in research.

According to Drake, this information led to changes in the recruitment strategy for the prostate cancer cohort. They provided information to all of the physicians about this study, and gave them fliers to put in their waiting rooms as well as little cards for participants to take with them. She said that one of the more important but also time-consuming strategies was to include an additional approach for each participant. The first attempt to recruit participants was between diagnosis and treatment, which is a small window. Prior to this time period, they would have an informational meeting where the study was introduced as an opportunity, and patients were told that their urologist or medical oncologist was aware of it, and they were encouraged to talk with their doctors about the study. The study team then offered to call the patients or meet them at their next appointment. She said this offer increased the comfort and the trust among many participants and increased the percentage of minorities by 7 to 12 percentage points over a couple of years, and that percentage is even higher now.

The next study that Drake described was designed to assess the impact of different consent models on intentions to participate in a biorepository for secondary research among a diverse sample of women. Four models of consent were used: (1) study-specific consent is when researchers ask a participant’s permission before each future study in which they would use their sample; (2) broad consent is when researchers only ask for the participant’s permission once for all future studies; (3) the opt-out option is when researchers inform the participant that their sample may be used for future studies, unless the participant directs them not to store the sample for future use; and (4) notice is when researchers tell participants that if they participate in this study, their sample will be used for future studies, without giving the participant an option to say yes or no.

Drake discussed the qualitative work done with a sample of 60 diverse women, of whom about half were African American and half were Caucasian. The women were asked to identify their most and least preferred options. They found that broad consent was the most preferred and that notice was the least preferred model of consent. Study-specific consent was the second most preferred and the second least preferred method of consent.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

Drake shared qualitative comments that reflected the participants’ reactions to each of the consent models. The notice model was perceived to have the benefits of being simple, efficient, and requiring little effort by the participant. However, concerns about the notice model were a loss of control over their sample, insufficient information, and it was discourteous not to ask for permission to use the samples. For the broad consent model, some of the benefits were that it was a good balance of the different model options, providing more control over the patient’s sample and less burden on the researchers. A few concerns raised regarding the broad consent model related to insufficient information about each individual future study. The study-specific model was perceived to provide information about all future studies and give the patient more control over their samples. Participants also raised concerns that it could slow down the research, and is more trouble for both the participant and the researcher.

Drake then described an experiment to assess the intent to participate after being presented a randomly selected consent model of three options: notice, broad, and study specific. The sample consisted of 357 women, of whom 56 were African American, about one-third had a high school degree or below, about one-third had a college degree or above, and about one-quarter had a limited health literacy level. Drake said that trust in medical research was a consistent and significant predictor of intention to donate for each of these consent models.

Although they did not observe any racial differences in intent to donate for any of the three consent models, Black participants reported wanting more control overall than White participants of their samples, Drake said. Individuals who participated in medical research in the past were comfortable with less control of their samples than those who had not participated in medical research. She noted an interaction between race and trust in research and doctors, such that trust in medical research, as well as trust in doctors, was significantly and negatively associated with the level of control, but only among White participants, not Black participants.

Drake summarized her research among men, particularly African American men, and a diverse sample of women as illustrative of how to better educate and work with patients and community members on consent models for biorepositories. She said that these lessons can be applied for other types of research as well.

Harnessing New Communication Methods to Support Survey Research in the 21st Century

Amelia Burke-Garcia is a health communications professional focusing on digital outreach strategies at NORC at the University of Chicago. She is also the author of the book Influencing Health, A Comprehensive

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

Guide to Working with Online Influencers. Burke-Garcia focused her talk on how digital and social media can be used throughout the survey lifespan to help with some of the challenges that the survey research field is facing, particularly related to engagement. Burke-Garcia noted that the survey and polling business has been in crisis with response rates falling for more than 30 years, and high-quality face-to-face surveys now rarely reach a 70 percent response rate.

According to Burke-Garcia, this trend is driven by a number of factors, which taken together make the whole world harder to survey. She pointed to displacement of people both within the United States as well as globally, explaining displaced persons are not necessarily living in their homelands and may or may not be counted as part of the populations being surveyed. She also noted that many Americans have direct access to the internet via mobile devices only; research has shown that with so much information coming every day, people tend to lose their concentration after just 8 seconds; and trust is waning, with just 18 percent of Americans saying that they can trust the government.

Burke-Garcia said these factors have implications for doing survey research because the devices and the technology that people use are critically important for how they access information, as well as participate in web-based surveys. She urged thoughtfulness in terms of engaging participants, and carefully considering the length of the questionnaire that respondents are asked complete. At least pre-COVID-19, people were spending more time in the workplace than at home, which makes it harder to contact and survey people at their homes, a key mode of data collection for many surveys.

Burke-Garcia also described developments with new types of data being made available, new ways of communicating with people, reaching people, and identifying them, which constitute a paradigm shift for survey research that uses multiple data sources, multiple modes, and multiple frames. For example, there are more than 3 million Google searches every minute; more than 300 million monthly active Amazon users globally; more than 2 billion monthly active Apple device users globally; and more than 2.2 billion people on Facebook globally.

Burke-Garcia advocated using new and emerging communication channels to support and amplify survey research, and broadening the conceptualization of what social and digital media mean for survey research. Burke-Garcia said that it is not just about the message, text, or images used to communicate, but that these platforms, these data, and the people that use them offer some additional context. Her focus is on how to use social and digital media throughout the survey life cycle, including sampling, identifying, reaching, persuading, and interviewing respondents.

In terms of sampling people, Burke-Garcia said that social media offers opportunities for leveraging and building on venue-based time-space

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

sampling. People congregate as groups and get together for virtual events on Facebook, but those events differ from in-person real-time events where some kind of survey data collection might take place. However, she said that leveraging groups that come together around similar issues or topics can be effective to get additional types of samples and subgroups to participate in research. In addition, she pointed to online groups and online influencer networks that offer opportunities to potentially do some additional sampling work.

In seeking to identify people, Burke-Garcia said work is being done that looks at how to leverage different variables shared through social media to both deepen an understanding of survey responses and supplement gaps in data collection. She cautioned about risks in terms of what data people make available and the accuracy, but suggested researchers could passively collect some data to reduce burden on respondents.

Burke-Garcia described ways to reach and engage people in social media to participate. One main way is to leverage paid social media advertising to promote surveys to people who may not otherwise know about them. Thinking about surveys as brands can improve understanding about how to message why data are being collected and what value these data will have in their lives or communities, she said, adding that the messaging also can help garner trust to participate. Paid social media ads are fairly easy to develop, launch, and manage, she commented. She noted that one can see fairly quickly what messages may or may not work well, which improves efficiency in trying to reach and engage people to participate.

Burke-Garcia also noted work to use social media as an additional contact channel for follow up to lower attrition. She suggested thinking about people’s social networks as additional opportunities to connect with them over time, as their address or phone often may change, especially for harder-to-reach populations. She said that their social media does not change as frequently, which provides another venue for contacting and tracing.

For persuading people, Burke-Garcia said social media advertising can also be very effective. She encouraged running ads in an A/B testing approach, which reveals which messages are more persuasive versus those that are less persuasive based on people’s engagement with the ads, how they click through to a survey, if they start a survey, and if they are eligible to complete it. In addition, online and social media influencers can be very persuasive and powerful in raising public awareness about the survey and establishing legitimacy of the research. She added that hearing about the survey from a friend can be quite persuasive because that person is a trusted voice.

In terms of interviewing people, Burke-Garcia noted the potential for passively collecting what people are doing and accessing some of their data can be highly valuable to deepen an understanding of what is collected in

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

the survey. She also noted opportunities for active data collection by soliciting feedback from particular groups in social or in digital media as a way to crowdsource information and get feedback perhaps less formally than a survey, but still add value.

Challenges exist with these approaches, which are in various stages of testing and implementation, Burke-Garcia cautioned. Social media platforms and the way people use them are changing rapidly, but many types of groups, especially harder-to-reach groups, congregate within different online communities. She encouraged experimenting with and exploring how to use social media to augment traditional survey methods. Ad blockers, inactive users, fake accounts, and bots pose additional challenges, and she urged having an understanding and awareness of users.

Transparency, privacy, consent, and appropriate use of the data apply to these data as they do to survey data, Burke-Garcia stressed. Tools for analyzing these data are improving but more work is needed to better understand respondents’ sentiments. Other considerations she briefly noted were the available budgets, threats to the validity of the study from human error and misrepresentation, and the risk for a Hawthorne effect by participating and engaging within these communities. Burke-Garcia urged considering these challenges to ensure that research is implemented correctly.

Burke-Garcia concluded by pointing toward future directions and the work that needs to be done, urging investments in pilot and feasibility studies to test if and how these approaches work and to identify weaknesses and gaps. She said that new frameworks need to be explored to integrate these different types of data with traditional survey research. She also noted that work is needed on error structures for these new data sources, and developing protocols for appropriately and successfully using social media data. She underscored the opportunity to reach new populations, suggesting that engaging people in different ways can hopefully create more engagement and participation in survey research into the future.

Discussion

John Phillips kicked off a discussion about using social media to provide awareness, comfort, or trust in participating in a survey, and he asked who the influencers are for aging studies. Burke-Garcia replied a large majority of online influencer spaces are comprised of moms and dads who become caretakers of their children and their aging parents. She suggested an opportunity to engage them to help bring the message home to their peers who may then influence their parents to participate. She added that older adult influencers tend to be heavy Facebook users and may participate in groups for social community or to look at photos and stay in touch with their families.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

Phillips also referred to a comment by Drake that suggests a theme running through several presentations: When people have some experience with the research enterprise, they tend to be a lot more comfortable with it. He said he is thinking about ways to use this observation to further scientific engagement for the community and to improve people’s willingness to participate and consent to research. He noted the opportunity in longitudinal studies to build that trust over time, but the initial ask to participate is in some respect the most important point.

Drake responded that trust is very important, but so is “pulling back the curtain” and helping people understand what is going on behind the scenes. She underscored the opportunity to educate and engage community members before they get to a point where they are being asked to participate in a study or a clinical trial. Drake said researchers often mistakenly take for granted that participants understand what is done with their specimens and data and why. Helping them understand that ahead of time can be important in gaining their trust and participation.

Phillips concluded by noting an interesting potential link between the two presentations in this session. He suggested bringing the experience with research that Drake described to a broader population through the types of social media that Burke-Garcia discussed, perhaps through testimonials to make people more comfortable with a study.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×

This page intentionally left blank.

Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 31
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 32
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 33
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 34
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 35
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 36
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 37
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 38
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 39
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 40
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 41
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 42
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 43
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 44
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 45
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 46
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 47
Suggested Citation:"3 Participation." National Academies of Sciences, Engineering, and Medicine. 2022. Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26481.
×
Page 48
Next: 4 Informed Consent »
Improving Consent and Response in Longitudinal Studies of Aging: Proceedings of a Workshop Get This Book
×
Buy Paperback | $25.00 Buy Ebook | $20.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

This Proceedings of a Workshop summarizes the presentations and discussions at the Workshop on Improving Consent and Response in Longitudinal Studies of Aging, which was held virtually and live-streamed on September 27-28, 2021. The workshop was convened by the Committee on National Statistics of the National Academies of Science, Engineering, and Medicine to assist the National Institute on Aging (NIA) with its methodological research agenda and inform the different longitudinal survey programs sponsored by NIA about practices and research to improve response and consent in other survey programs. The workshop was structured to bring together scientists and researchers from multiple disciplines and countries to share their research and insights on how to improve response and consent in large, representative longitudinal studies on aging.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!