Log in / Register
Home arrow Engineering arrow Measuring Electronic Word-of-Mouth Effectiveness: Developing and Applying the eWOM Trust Scale

Scale Application: Segmenting Online Consumers

Segmenting Consumers on eWOM/Online Ad Trust

Another critical issue of this research is to demonstrate the usefulness of the newly developed eWT-S in a greater research context. Therefore, the last research question concerns the applicability of the new eWOM trust scale for developing an empirically-based typology of online consumers in respect to their level of trust in online customer reviews (eWOMTmst) and online advertising (oADTrust). Specifically, the following question was proposed:

RQ 13: (a) Can online consumers be meaningfully segmented according to their trust in online customer reviews and online adverstising? (b) How can segments be profiled?

As demonstrated earlier, research suggests that the levels of trust that consumers have developed towards eWOM and online advertising as the two major sources of market communication are likely to determine their responses. Answers to this research question not only shed light on the influence of these two critical forms of online market communication (e.g., persuasiveness, purchase influence) but also will enable marketers to implement target- group specific communication instruments in order to stimulate positive market responses. As it seems reasonable to group consumers according to the extent to which they rely on these information sources, cluster analysis was employed to segment US online consumers according to their responses to the new eWOM trust and online advertising trust scales. In its basic form, cluster analysis is a multivariate statistical procedure that meets the objective of exploring groupings of individuals in reference to certain characteristics they possess (Aldenderfer & Blashfield, 1984; Hair et al., 2010). In this research, the mean scores of generalized trust in online customer reviews as well as online advertising for each of the 517 respondents were used as a basis for a two-step clustering procedure (Punj & Stewart, 1983). For the initial step, the Ward’s hierarchical clustering method with squared Euclidean distances was applied in order to get an insight into a potential clustering solution. For an adequate representation of the real structure within the data, the elbow criterion was employed to determine the number of clusters. It was concluded that the sample is best described by four clusters. During the second step, a non-hierarchical, k-means clustering procedure (MacQueen, 1967) was used for developing a four-cluster solution. Here, the group centroids calculated in the hierarchical procedure were designated as the initial clusters for the subsequent k-means clustering. Tables 51 and 52 describe the identified empirical segments.

In general, the four distinct segments can be labeled as follows. Pure review trusters (19% of respondents): consumers in this cluster express a high level of trust in online customer reviews while showing a low level of trust in B2C market communication. Non-trusters (10%): the smallest group in the sample is consumers who have trust neither in online customer reviews nor online advertisements. It appears that these respondents tend to rely more on alternative information sources, including internal (i.e., their knowledge) and external sources (e.g., friends). Moderately trusting customers (41%): the largest segment is composed of consumers who score moderately on both trust in online customer reviews and online advertising. It can be assumed that in this group the impact of and trust in eWOM/Online advertising is here mainly triggered by situational circumstances. High trusters (30%): consumers in this group show the highest level of eWOM trust while at the same time have high trust in online advertising. Together with moderately trusting consumers, in terms of size these two clusters represent more than two thirds of the sample (71%). Further analysis demonstrated that within each cluster, eWOM trust was significantly higher than Ad trust. For instance, a paired-samples t-test for the high-truster segment indicated that scores were significantly higher for the eWOM scale (M = 5.13, SD = .50) than for the Ad trust scale (M = 4.79, SD = .68), t(153) = 5.42, p < .001.

In accordance with the suggestions made by Maute and Dube (1999), the cluster solution was subject to a multivariate analysis of variance (MANOVA) to test its internal validity. More specifically, the four clusters were compared in terms of the newly developed 22 trust-in-online- customer-reviews items (a = .94), as well as the 9 trust-in-online-advertising items (a = .97). A review of the Hotelling’s trace statistic showed significant differences between the identified clusters (F(6,1022) = 463.25, p < .01; Hotelling’s trace = 5.44, partial T|2 = .73) which therefore represents reasonable empirical evidence for the internal validity of the proposed four-cluster solution.

Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
Business & Finance
Computer Science
Language & Literature
Political science