By Susan Rosegrant
December 2012 (updated in April 2013)
Dana Garbarski is the winner of the 2012-2013 Charles Cannell Fund in Survey Methodology.
Suppose your spouse had a serious illness today with very low chances of survival. First, what if he were mentally intact, but in severe and constant pain? Would he want to continue all medical treatments or stop all life-prolonging treatments?
So reads a question in the Wisconsin Longitudinal Study. It’s one of a series of questions about end-of-life treatment that comes towards the end of the hour-long survey, which has followed the lives of a random sample of Wisconsin high school graduates for more than 50 years.
Researchers use this information to study how the elderly view end-of-life choices. But for Dana Garbarski, who recently completed her Ph.D. in sociology from the University of Wisconsin, Madison, the questions provided an ideal opportunity to learn more about how interviewers and respondents interact. Could an interviewer’s skill in handling such potentially sensitive questions at the end of a long interview affect whether a respondent would stay involved in the survey? And what if respondents paused and sounded uncertain? Did that mean their responses were less likely to agree with that of their spouses?
To answer these questions, Garbarski transcribed and coded audio recordings of 500 respondents from the 2004 study. Her analysis is focused not on the content but on how people answered.
“I think that respondents who seem to be displaying some sort of discomfort or uncertainty when answering these questions about end-of-life planning will be less likely to participate in the future, particularly if the interviewer isn’t able to establish a sense of rapport,” Garbarski says. “If the interviewers are responsive to respondents, if they are able to keep respondents’ motivation in answering these questions, I think that’ll be related to participation in the future.” To see if respondents stuck with the survey, Garbarski has tracked how many returned a DNA kit mailed to them a few years after the study.
Garbarski concedes that the second question, looking at whether hesitation or uncertainty signal a less accurate report of a spouse’s preferences, is a harder one to parse. For example, cognitive problems could affect the answers of one or both members of a couple, or respondents might never have discussed the issue with their spouses, and might simply report what they want for their own care.
Still, Garbarski expects that respondents who are hesitant will be less reliable reporters—a hypothesis she can check, because in each case she has the spouse’s response for comparison. Knowing that hesitant behavior is a warning flag could help researchers interpret the data from surveys that rely on respondents to report on the actions or beliefs of others.
Garbarski used the money she received from the Charles Cannell Fund in Survey Methodology for the transcription and coding of the interviews, and she has started analyzing the data. She is now a post-doc at the University of Wisconsin Center for Women’s Health and Health Disparities Research, and is also the mother of two young children—Lucy, 3½, and Levi, 2. “If you think of parenting as my primary job, then what I do for fun is my work,” she says. “And if my work is my primary job, then what I do for fun is parenting.”