Portal de Periódicos da CAPES

Efficiency of Different Recruitment Strategies for Web Panels

2011; Oxford University Press; Volume: 24; Issue: 2 Linguagem: Inglês

10.1093/ijpor/edr020

1471-6909

Kasper M. Hansen, Rasmus T. Pedersen,

Expert finding and Q&A systems

The use of CAWI (Computer-Assisted Web Interviews) has increased greatly over the last decade, partly driven by technological development and increased Internet penetration, and partly by falling response rates in the traditional modes of polling (Baker et al., 2010; Curtin, Presser, & Singer, 2005; Couper 2000; Hansen, 2007). While some studies have found potential disadvantages to CAWI, e.g., higher levels of "don't know" answers and high levels of break off (Heerwegh & Loosveldt, 2008; Peytchev, 2009), research also shows many advantages of CAWI compared to traditional modes: They are cost-efficient, allow automatic correction of errors and omissions during the interviews (Alvarez & Beselaere, 2005), and they lessen problems with social desirability bias towards interviewers (Baker et al., 2010; Kreuter, Presser, & Tourangeau, 2008). Furthermore, using web panels to conduct recurring CAWI with the same group of respondents let us build true time-series data which mitigates the problem of endogeneity inherent in so many public opinion studies. Additionally, modest differences are found when comparing results from web panels with traditional modes of surveys (Sanders, Clark, Stewart, & Whiteley, 2007), and web panels even display higher levels of data reliability than telephone surveys in some studies (Braunsberger, Wybenga, & Gates, 2007). One of the key challenges for web panels is recruitment of members. While numerous studies have investigated strategies to increase response rates in web surveys, relatively little research has looked specifically at recruitment strategies for web panels (Rao, Kaminska, & McCutcheon, 2010). Through an experimental design, this article tests the efficiency of various recruitment strategies in a probability sample for a web panel. Efficiency is here mainly measured as response rate. A low response rate does not necessarily induce nonresponse bias (Groves, 2006), but it can nevertheless be a serious problem, if nothing else because it increases the cost of recruitment. Furthermore, in addition to response rate, the article also considers issues of general demographic representativeness, cost effectiveness and speed of recruitment.