Suggestive Surveys

Simply asking survey participants whether they intend to exercise or use illegal drugs in the near future can result in increases in both behaviors, according to researchers at Duke, the University of Pennsylvania, and Baruch College. These findings should serve as a warning to public-policy and health-behavior researchers who routinely use surveys as a method to study risky behaviors, the researchers say.

"We expected to find that students asked about exercise would exercise more, because that fits with past research regarding people's motivation to conform with socially desirable behaviors," says Gavan Fitzsimons, associate professor of marketing and psychology in Duke's Fuqua School of Business. "What we weren't expecting to see is that the students asked about drug use actually reported more, not less, use of illegal drugs."

In a study published in the journal Social Influence, marketing professors Fitzsimons, Patti Williams of Penn's Wharton School, and Lauren Block of Baruch's Zicklin School of Business asked 167 undergraduate students how likely they were to either exercise or use illegal drugs over the next two months.

Two months later, the students were asked to report how often they had exercised and how often they had used illegal drugs since the initial survey. Students who were initially asked about their intention to exercise reported exercising an average of 15.7 times during the two months, compared with students not asked about exercise, who reported exercising an average of 11.8 times during the same period.

Likewise, students who initially were asked about predicted drug use reported using drugs an average of 2.8 times in two months, while students not asked reported using drugs an average of 1.1 times over the same period.

The team conducted a follow-up analysis of only those participants who reported any exercise or drug use during the two months following the initial survey. Among students who reported any exercise at all, those who were initially asked about their intentions to exercise reported doing so an average of 20.4 times, while those who had not been asked about it exercised an average of 13.9 times. Participants who were initially surveyed about intended drug use and consequently used drugs at least once reported doing so an average of 10.3 times, compared with an average of four times for participants who were not surveyed about drug use.

Based on this secondary analysis, the researchers suggest that people already predisposed to a certain behavior, whether negative or positive, may be even more likely to increase that behavior when asked about it. "This effect seems to work both ways," Fitzsimons said. "It's always great if we can somehow encourage a healthy behavior, but the dilemma sets in when we suddenly see that researchers may inadvertently cause people to increase their dangerous behaviors."

In follow-up research to appear in the June 2007 Journal of Consumer Research, Fitzsimons, Williams, and Joseph Nunes of the University of Southern California suggest that people may have a blend of positive and negative attitudes about certain behavior. This follow-up research also directly observed risky behavior to be certain the results of the initial study were not driven by the fact that drug use was reported, not observed.

"We're working hard right now to better understand when questions may lead to these negative side effects," Fitzsimons says. "For example, it appears that warning respondents in advance that asking questions can influence behavior may be a successful way to inoculate them against the potentially harmful effects of asking questions about risky behaviors."