Because all data are self-reported and collected using a
cross-sectional research design, common method variance
(CMV) may confound the true relationships
among the theoretical constructs of interest (Chang,
Witteloostuijn, and Eden 2010). Consequently, we followed
both ex ante (procedural) and ex post (statistical)
remedies to control for CMV. Regarding procedural
remedies, as Podsakoff et al. (2003) suggest, we adopted
different response formats and a counterbalancing question
order. For example, we asked respondents about
their WTB products before asking for their PJ to avoid
priming effects. Regarding statistical remedies, in the
pretest, we also included a ten-item measure for social
desirability taken from Netemeyer, Burton, and Lichtenstein
(1995) and originally developed by Crowne and
Marlowe (1960). All correlations between the social
desirability scale and our construct measures were not
significant. In addition, in the main study, we employed
Lindell and Whitney’s (2001) marker variable assessment
technique to assess CMV using “Germany is a
country of my dreams” (measured on a seven-point
scale) as the marker variable. We selected this item
because it was conceptually unrelated to both our
dependent and predictor variables. All correlation coefficients
that were significant on a bivariate basis
remained significant after we partialed out the marker
variable. In light of these analyses, we conclude that
CMV does not seem to pose a major threat in our study.