Another way to evaluate the eWOM trust scale’s validity is to assess the extent to which the proposed operationalization is similar to (or “converges on”) alternative but theoretically similar operationalizations (Trochim, 2012). In other words, convergent validity is defined as the degree to which two independent measures which are designed to measure the same construct are related (Netemeyer et al. 2003). Evidence for convergent validity of the developed measure is offered by significant and strong correlations between responses obtained by maximally different methods of measuring the same construct (Churchill, 1979; Netemeyer et al., 2003; Peter, 1981). Hence, in order to investigate the convergent validity of the eW OM trust scale, the following research question is proposed:
RQ 3: Does the developed measure of eWOM trust correlate significantly and considerably with other methods to measure eWOM trust?
The aim of discriminant validity evaluations is to ascertain that a measure is indeed novel and does not simply reflect some other variable (Churchill, 1979). More specifically, discriminant validity is the investigation into the degree to which the proposed operationalization of the construct is not similar to (or “diverges from”) alternative operationalizations which should measure a similar, but conceptually different construct (Netemeyer et al., 2003; Trochim, 2012). Accordingly, evidence for discriminant validity is indicated where there are “predictably low correlations between the measure of interest and other measures that are supposedly not measuring the same variable or construct” (Heeler & Ray, 1972, p. 362). Discriminant as well as convergent validity are both operational types of validity (Peter, 1981) and are typically investigated simultaneously by using the Multitrait-Multimethod Approach (MTMM Matrix) proposed by Campbell and Fiske (1959) - and/or alternative approaches (e.g., Anderson & Gerbing, 1988). In the previous section, this thesis has discussed eWOM trust construct’s relationship with review credibility (RCred) and attitude towards reviews in general (RAtt). This thesis has concluded that these concepts are related but conceptually distinct. In order to verify the discriminant validity of the developed measure of eWOM trust, the scale should not correlate very highly with the measures of review credibility and attitude which represent similar or overlapping but different constructs. Therefore, the following question addressed:
RQ 4: Is the developed measure of eWOM trust significantly different from the measurements of (a) review credibility and (b) attitude towards reviews in general?
This thesis assumes that the new trust measure is constrained to measure confidence in OCR and not in other trust objects. Hence, it does not possess the ability to measure a more general tendency to trust other people or generalized others (i.e., a person’s disposition to trust) nor to quantify trust in other forms of market communications (e.g., salespersons, advertising). Hence, it is domain-specific and individuals develop different kinds of generalized trust depending on the trust object. In order to evaluate whether people develop trust that is specific to customer reviews, the following research question emphasizes the construct’s relationship with other trust objects:
RQ 4: (c) Is the developed measure of eWOM trust significantly different from the
measurement of dispositional trust? (d) Do consumers develop trust that is specific to online customer reviews?
In order to argument for construct validity, the measure has to be able to identify groups that are known to show differences in the focal construct (MacKenzie et al., 2011). According to Netemeyer et al. (2003), known-group validity targets the extent to which a measure differs as predicted between groups of people who should score low and high on the focal trait. Hence, evidence for this kind of validity is provided by significant differences in mean scores of the trait across these groups. In literature, diverse hints exist that for most market communications increases in psychographic as well as socio-demographic variables such as intelligence, self- esteem/self-confidence, education, income, and age (for a discussion see later) are all likely to have a negative effect on consumer persuasion. In line with earlier contributions in the field of advertising research (Obermiller & Spangenberg, 1998), it is assumed that such a resistance against persuasion is indistinguishable from eWOM scepticism which translates into minor levels of eWOM trust and potentially eWOM distrust. The author feels confident that the conglomerate of heightened levels of self-confidence, and earlier experiences (e.g., with other consumers, purchasing), amongst others, together shape a consumer’s generalized tendency to rely on review-conveyed information. Due to a lack of an adequate knowledge base, limited analytical abilities and alternative forms of socialization processes, younger consumers are theorized to have significantly higher levels of trust in OCR. This is also mirrored by earlier research that concludes that younger Internet users rate information on the Internet as more trustworthy than older users (Johnson & Kaye, 2002). Likewise, older Internet users are more sceptical towards the Internet in general (Metzger, Flanagin, Eyal & Lemus, 2003), which may result in lower levels of information trust. The fact that consumer scepticism is likely to increase with age is also supported in the offline literature (Boush et al., 1994). Various authors suggest a negative relationship between age and persuadability in general (Tyler & Schuller, 1991; Visser & Krosnick, 1998). Therefore, this thesis argues that eWOM trust is limited among faculty members compared to students because of several reasons. For instance, members of the faculty are typically older and as a consequence have other and more marketplace experiences (Obermiller & Spangenberg, 1998). They have learned to use the Internet differently from younger consumers. Additionally, they possess a higher educational level and are socialized during an era where the Internet was not available or was in its early phase. Younger people are used to socializing via the Internet, apply Web 2.0 applications more regularly and hence have more direct experiences with the trust object. Further, it is assumed that faculty members have a higher level of self-confidence and feel more comfortable in making critical product judgments on their own. In addition, faculty members are theorized to have a higher self-esteem and posses on average higher intelligence. Both circumstances should cause limited persuadability (McGuire, 1968; Rhodes & Wood, 1992) and hence eWOM trust.
RQ 5: Do students and faculty members differ significantly in their trust in customer reviews?
According to psychometrics literature, predictive validity is defined as the extent to which a new measure predicts scores of a criterion measure (Cronbach & Meehl, 1955). Netemeyer et al. (2003, p. 86) define this validity type as “the ability of a measure to effectively predict some subsequent and temporally ordered criterion”. In the context of this thesis, this criterion is referred to as trusting behaviours (i.e., diverse forms of eWOM usage). Prior research shows that trust and trusting or risk-taking behaviours are closely related (Gefen et al., 2003; Mayer et al., 1995; McKnight et al., 2002b). This is supported by the Theory of Reasoned Action (TRA), which advances that attitudes indirectly and behavioural intentions directly (i.e., both elements of trust) are the most influential predictors of behaviour (Fhishbein & Ajzen, 1975). An essential assumption of the TRA (as well as the Theory of Planned Behaviour (TPB) (Ajzen, 1991)) is that there is a positive relationship between beliefs, intentions and behaviours. Accordingly, numerous empirical studies - in different contexts - have found considerable correlations among these concepts confirming the presence of such a positive, strong relationship (Hansen, 2005; Sheppard & Sherman, 1998; Vijayasarathy, 2004). In a similar vein, trust researchers have augmented these findings and provide theoretical as well as empirical support that trust and trusting behaviours (i.e., risk-taking behaviours) are positively related (see earlier). In the context of eWOM, trust has been shown to elicit various forms of trusting behaviours. For instance, when trust is present, individuals liberally exchange, seek, and collect knowledge gained through online reviews and recommendations (Chen & Hung, 2010). Trust between the members of a virtual community has been demonstrated to affect members’ behaviours, such as obtaining and contributing information (Kankanhalli, Tan & Wei, 2005; Lu et al., 2010). Based on these prior findings, this thesis is interested in investigating the new measure’s ability to effectively predict subsequent and temporally ordered behaviours related to eWOM communications. Similarly, the measure should be able to postdict prior behaviours. Hence, the following research question is asked:
RQ 6: Does the eWOM trust measure significantly predict (postdict) eWOM-related consumer behaviours (e.g., opinion seeking, adoption, giving, passing)?