Analysis procedures

The proposed hypotheses were tested using structural equation modelling (SEM). Two models were tested. Model 1 (Partial mediation model) was used to test H1, H2, H3, H5 and H6. The combined results of Model 1 and Model 2 (Direct effect model) were useful in testing H4. Measures for the key constructs included in the models are assumed to be reflective. in nature consistent with previous research. Hence, a reflective specification of the first-order dimensions is assumed. By definition, consumer-based brand equity is “reflected in the dimensions of brand awareness, brand associations, perceived quality and brand loyalty”. These four dimensions constitute a higher-order consumer-based brand equity construct. Hence, consistent with previous research in the area of brand equity (e.g. Yoo and Donthu, 2001) a reflective specification of consumer-based brand equity is appropriate. Similarly, consistent with previous research in the areas of brand signalling (e.g.Erdem et al., 2006), and source credibility (e.g. Ohanian, 1990), measures for brand credibility and endorser credibility are assumed to be reflective in nature.
Endorser attractiveness, expertise and trustworthiness were modelled as observed independent variables, which measured endorser credibility. Brand expertise and trustworthiness were modelled as observed independent variables, which measured brand credibility. Brand awareness, brand associations, perceived quality and brand loyalty were modelled as observed dependent variables which measured consumer-based brand equity.

Results for the main effects
Model 1 (Partial mediation model).
Endorser credibility, brand credibility and consumer-based brand equity were the latent variables in Model 1.
a. H1 predicted that endorser credibility would positively affect brand credibility
b. H2 predicted that brand credibility would