Supplemental Reliability and Generalizability Assessments
In an effort to provide additional evidence for the reliability of the new eWT-S (besides split- half reliability, etc.), the following research question was asked:
RQ 7: Is there a significant and considerable correlation between the measurements of eWOM trust on two different assessment occasions?
Responses to the eWOM trust scale were collected on two occasions separated by three weeks from a convenience sample of undergraduate business students. On the first occasion, a paper- and-pencil questionnaire was administered to the respondents during five minutes of class time. The students were informed that the survey related to their perceptions of online customer reviews in general, and on the front page of the questionnaire a definition of customer reviews was provided. A similar description was given at the second point of time. No additional information was provided concerning the reasons for the second administration and no student questioned it. All respondents participated for course credits. Because of absences and not assignable questionnaires, a final sample consisted of 50 persons. The estimates of internal consistency reliability (Cronbach’s alpha) ranged from .74 to .94 at the first phase, and from .70 to .92 at the second phase (see Table 44).
The correlation between the first and the second administration for the overall eWOM trust score was .73 and was significant on the .001 level. The test-retest correlations for the five subdimensions ranged from .64 to .84. All of these correlations were significant (p < .001), and thus they provide some evidence for the stability of the scale over time.