Pushing Households to the Web: Experiments of a 'Web+Mail' Methodology for Conducting General Public Surveys

Benjamin L. Messer

Previous research found that a web+mail survey design, in which a web survey request is mailed to an address-based sample of households, followed by a paper questionnaire at a later date, can be effective at obtaining responses via the web. However, many questions remain unanswered about which mail procedures can be used effectively to `push more households to the web' in different populations. In this dissertation, I conduct several experiments to test multiple survey procedures for improving web+mail methods and gain a more comprehensive understanding of how these methods perform in different populations. The first study explores the effects of using $5 cash incentives, web instruction card, and special mailing in web+mail statewide surveys. Demographic analyses of web and mail respondents compared to estimates from the American Community Survey are also presented to identify potential nonresponse biases. The results suggest that web+mail methods are effective in a statewide population compared to mail-only control groups, particularly when used with $5 incentives. Additionally, socioeconomic status (SES) and Internet access in the household are found to be significant predictors of responding via web (vs. mail). The second study reports on a test of web+mail methods in three different states: Washington, a `local' population more familiar with the survey sponsor, Pennsylvania, a `distant' population less familiar with the survey sponsor, and Alabama, a `distant' and demographically different population in terms of lower SES and Internet coverage. Results suggest that both "distance" from survey sponsor and lower SES and Internet coverage have a negative effect on web response rates and can contribute to higher nonresponse bias (vs. mail). The final study examines the effects of using a web+mail design (2web+mail) whereby the mail option is not mentioned to sampled households until it is sent in the final contact (vs. a web+mail design that mentions the mail option in prior contacts) in an effort to increase web response rates in Washington, a `local' population, and Pennsylvania, a `distant' population. Results indicate that the 2web+mail design is successful at increasing web response rates vs. web+mail, especially in Pennsylvania, thus helping to potentially overcome the `distance' barrier.

Pushing Households to the Web: Experiments of a 'Web+Mail' Methodology for Conducting General Public Surveys

Files and links (1)

Abstract

Metrics

Details