Publication Type: Conference Paper/Unpublished Manuscript
Review Method: Peer Reviewed
Abstract: Despite the widespread interest in utilizing web-based SAQ instruments as a primary mode of data collection, controlled experiments on the relative effectiveness are rare. The 2003 Survey of Doctorate Recipients (SDR) has implemented such an experiment and this paper will present findings on the effects of a web-first option and a “preference” (where the respondent chooses between web and mail modes) option on survey response rates compared to CATI-first and the standard mail-first modes. The SDR is a panel survey administered biennially to a national sample of some 40,000 doctorate-holders in the fields of science and engineering. Until 2003, the primary mode has been a mail SAQ with about 40 percent ultimately completed via CATI in follow-up efforts. In 2003, stratified random subsamples of about 2,000 cases each were allocated into the web-first, CATI-first, and preference experimental treatment. Stratification factors were sample member’s doctoral cohort, prior survey outcome (complete, refused), and sector of employment (academe, business or industry, government). The experimental groups will be compared with a control group of about 28,000 cases administered the standard SDR treatment of advance letter, first mail questionnaire, postcard reminder, second mail questionnaire, and then transfer to CATI mode. The experimental groups will also be compared with each other. An interesting subset of the web-first subsample are contacted via e-mail, in order to assess whether response rate improves with e-mail contacts. Response rate differences will be assessed using simple t-tests as well as logit models for estimating interaction effects of mode and stratification variables. The 2003 survey began in October and data collection continues until April 2004; final response rate outcome data will be available immediately after close of the data collection.