only engage users but would also produce data that could be “scraped” and analyzed by EIA, for example, to track trends in the behavior of the users.
A more involved approach that would work particularly well in combination with a web survey would be to provide each respondent with a brief analysis of his or her responses once the questionnaire is completed. This is a concept similar to one used by the National Health and Nutrition Examination Survey (NHANES), which provides survey participants with the results of their physical examination as an incentive to participate.1 The results provided to the RECS respondents might compare their homes’ energy use to energy efficiency standards or to the energy use of similar houses in the neighborhood, state, or climate zone. The most compelling, although also most resource-intensive way to implement this feature would be to generate the analysis “live” at the end of the survey. Alternatively, or in combination with the live online feedback, a process could be developed for mailing a respondent the report after the interview is completed.
One concern that arises is whether techniques such as these may be disproportionately more likely to engage sample members who are particularly interested in the topics measured by the survey—for example, in the subject of energy efficiency—and who therefore are likely to be different from the rest of the sample. This concern can probably be mitigated by aiming to maintain high overall response rates. In other words, even if the interactive tools attract a subset of the population which differs from the rest of the sample, intensive nonresponse follow-ups, including through different modes, should assure that any bias introduced by disproportionate interest in the incentives among a subset of the population is minimized after follow-up.
The question of how such feedback could affect participation in the proposed longitudinal survey deserves particular attention. An interactive feature could serve as an especially strong incentive if a longitudinal design is introduced. However, it is possible that the feedback provided will alter the behavior of respondents in the longitudinal sample, not only in terms of their likelihood of participating in a second survey, but also in terms of their energy consumption. The benefits and associated risks would have to be evaluated.
1 For an example of the Final Report of Findings prepared for NHANES respondents, see http://www.cdc.gov/nchs/data/nhanes/nhanes_07_08/ROF_07_08_eng.pdf [December 2011].