Both questionnaires were completed by 292 respondents, but we included a “quality check” item in the first questionnaire to detect “blind checking,” where survey respondents provide responses to questions without reading or understanding them. The item read: “if you read this item, please do not respond to it” (Dollinger and DiLalla, 1996). Thirty‐four respondents provided responses to this item so we removed them from the data set prior to analysis. Thus, we tested the hypotheses using complete data from 258 respondents. We also analyzed the data with all 292 respondents included and detected no impact of the “suspect” responses on the results, but chose not to include them in the reported analysis so as to mitigate the possibility that blind checking might contaminate the findings. The sample (n=258) was large enough to give the statistical tests adequate power (>0.80) to detect the moderate‐sized effects we anticipated, and the ratio of subjects to paths estimated in the model (258/51) was approximately 5:1, which is reasonable large enough to mitigate potential multicollinearity among the exogenous variables (Grewal et al., 2004).
Both questionnaires were completed by 292 respondents, but we included a “quality check” item in the first questionnaire to detect “blind checking,” where survey respondents provide responses to questions without reading or understanding them. The item read: “if you read this item, please do not respond to it” (Dollinger and DiLalla, 1996). Thirty‐four respondents provided responses to this item so we removed them from the data set prior to analysis. Thus, we tested the hypotheses using complete data from 258 respondents. We also analyzed the data with all 292 respondents included and detected no impact of the “suspect” responses on the results, but chose not to include them in the reported analysis so as to mitigate the possibility that blind checking might contaminate the findings. The sample (n=258) was large enough to give the statistical tests adequate power (>0.80) to detect the moderate‐sized effects we anticipated, and the ratio of subjects to paths estimated in the model (258/51) was approximately 5:1, which is reasonable large enough to mitigate potential multicollinearity among the exogenous variables (Grewal et al., 2004).
การแปล กรุณารอสักครู่..