Studying Political Behavior: A Comparison of Internet and Telephone Surveys

Foreign policy beliefs and support for Stephen Harper and the Conservative Party

“…However, other research concludes that there are very few differences between inferences made from high-quality online survey data gathered via "active" sampling methods such as those employed in our survey and those gathered by other survey modes (i.e., in-person, telephone or mail-back questionnaires) (see Ansolabehere and Schaffner, 2014). Similarly, analyses in Canada (Stephenson and Crête 2010) and Great Britain (Sanders et al 2007, Clarke et al 2008 have found that differences between probability and online data are very small, and substantive conclusions about relationships between variables of theoretical interest are identical across modes. For a recent review, see Baker et al (2013).…”

Can we augment web responses with telephonic responses to a graduate destination survey?

“…Since it is impossible to determine the actual extent to which effect size was due to group differences rather than mode effect, suffice to conclude that web responses to alumni or graduate destination surveys can perhaps be augmented with telephonic responses if necessary. Yet, this conclusion should be seen in light of the following; As mentioned before, some studies do suggest that web is equally if not more viable and reliable than telephonic mode (e.g., see Braunsberger, Wybenga and Gates [2007], Kreuter, Presser and Tourangeau [2008, 864], and Stephenson and Crête [2010]). Some contextual and practical considerations also favour web, particularly for alumni or graduate destination surveys.…”

“…Moreover, in the messy world of applied research, modes are often mixed by augmenting one mode with another pending response rates. Stephenson and Crête (2010) found that although web and telephonic responses differed in a survey of political behaviour in Canada, analyses from one set or the other resulted in few inferential differences and that the overall effect resulting from the difference in mode appeared to be minimal. However, more research is needed in different contexts about the effect of mixed-mode, and sequential mixed mode surveys in particular, so that researchers can make more informed decisions whether to use mixed-modes, and, if so, the extent to which results can be analysed as a single sample and inferences be made about a population.…”

Sample Representation and Substantive Outcomes Using Web With and Without Incentives Compared to Telephone in an Election Survey

“…Regarding the accuracy of reported political behavior and preferences, Stephenson and Crête (2011) find the number of differences in point estimates to be relatively high. Malhotra and Krosnick (2007), comparing web data with the 2000 and the 2004 American National Election Study (ANES) face-to-face surveys, find accuracy to be higher for the face-to-face probability-sample data than for the online-panel sample data in the majority of the comparisons, in particular for voting turnout and party choice.…”