Apart from using path analysis to test the individual hypotheses, PLS was used to simultaneously examine the dependence relationships outlined in the conceptual model, identifying new paths, and optimising the model for the data.
163 The results of the three path analyses helped confirm the findings discussed in Sections 6.2.2 -6.2.6 and identify two non-hypothesised relationships (discussed in Section 6.3.1).
Additionally, the analysis of the models did not confirm the applicability of the ELM for exploring consumer responses to incongruity. Specifically, it was found that incongruence did not directly affect respondents’ Attention and Elaboration levels. Furthermore, Message Sidedness and Argument Quantity were not found to function as either central or peripheral cues, as Attention and Elaboration did not moderate the relationship between either Message Sidedness or Argument Quantity or Attitude towards the Corporate Brand.
Furthermore, the PLS analysis indicated that the data had a good fit to the conceptual model regardless of whether this was the original model or the one with the new relationships added in. The analyses revealed that only a small amount of Attention and Elaboration was explained, and that in the conceptual model it did not directly affect Attitude towards the Corporate Brand. This provides further indication that the dual process theory of the ELM was not as highly relevant to the model as predicted.
6.3.1 New Identified Paths
The analysis revealed two significant non-hypothesised paths. Specifically, it was found that Argument Quantity had a positive effect on Attitude towards the Ad, where higher number of arguments were found to negatively affect Attitude towards the Ad, which contradicts the effect of Argument Quantity on Attitude towards the Corporate Brand. This indicates that respondents were overwhelmed by the clutter or large amount of information caused by the high number of arguments. One reason for this could be that fifteen arguments (used for the high Argument Quantity condition) is too much to process and leads to information overload.
Previous research indicates that this number of brand attributes can lead to information overload and cause consumer confusion (Jacoby, 1984; Malhotra, Jain, & Lagakos, 1982).
As discussed in previous sections, Attention and Elaboration did not affect Attitude towards the Corporate Brand. However, it was found that Attention and Elaboration had a significant positive effect on Attitude towards the Ad, where ads were perceived more favourable with increased message processing. Consequently, as Attitude towards the Ad has been found to affect consumer attitudes toward the brand (Dröge, 1989; Homer, 1990; Lutz et al., 1983;
MacKenzie, 1986; Muehling & Laczniak, 1988), it is likely that Attention and Elaboration affects Attitude towards the Corporate Brand through Attitude towards the Ad. However, this
164 relationship did not affect how much of the variation in consumer attitudes (both towards the ad and the corporate brand) was explained by the data.
6.3.2 Effects of Covariate Variables
The effects of five covariate variables were examined in this research, which included Prior Brand Attitude, Need for Cognition, Need for Change, Enduring Involvement and Brand Familiarity. Additionally, the effect of Argument Quality was accounted for, due to its impact on brand attitudes through the central route of persuasion (Petty & Cacioppo, 1986b). The model was run for different levels of Argument Quality and the results indicated that, overall, Argument Quality had little effect. Regardless of Argument Quality, Congruence and Argument Quantity affected brand evaluations positively, while Message Sidedness affected brand evaluations negatively, as aligned with the general findings of this research.
The main result of the covariate analysis was the effect of Prior Brand Attitude on the dependent variables. Specifically, the ANCOVA analyses revealed that Prior Brand Attitude had medium to large positive effects on Corporate Credibility, Attitude towards the Ad and Attitude towards the Corporate Brand. Further PLS analysis involved testing the whole model for different levels of Prior Brand Attitude. This analysis revealed that the effect of incongruence on attitudes differed depending on respondents’ prior attitudes. Specifically, the results indicated that incongruent messages positively affected consumer brand attitudes for respondents with a negative (low) Prior Brand Attitude, while for respondents with a positive (high) Prior Brand Attitude, congruent messages were perceived more favourably.
Furthermore, for respondents with a positive (high) prior attitude, it was found that Message Sidedness did affect Attention and Elaboration, where two-sided messages led to increased processing.
The effects of Enduring Involvement was accounted for, as this has been found to affect motivation to process the message, and consequently, the amount of elaboration the recipients put into processing the message (Petty & Cacioppo, 1986a, 1986b; Petty et al., 1981). The ANCOVAs revealed that Enduring Involvement had a small positive effect on Attention and Elaboration and Attitude towards the Corporate Brand. Specifically, it was found that highly involved respondents paid slightly more attention, engaged in increased message processing and had slightly more favourable attitudes. Furthermore, the model was tested for different levels of Enduring Involvement (see Section 5.7 and Appendix 8.16). The
165 analyses revealed that for less involved respondents, two-sided messages were effective in increasing Attention and Elaboration levels, while this had no effect for highly involved respondents. For highly involved respondents, Message Sidedness no longer had an effect on Corporate Credibility. Additionally, the model was tested for respondents who had high average involvement levels (above 5) and it was found that congruent messages led to more favourable brand attitudes. Specifically, it was found that the hypothesised interaction between Congruence and Message Sidedness was now significant. Overall, this indicates that Enduring Involvement did indeed affect the persuasion route taken.
Though PAK’nSAVE was chosen as a familiar brand, the model was run for respondents who were very familiar (with an average score of 6 or above). Two-sided messages were found to increase Attention and Elaboration for respondents with high Brand Familiarity.
Furthermore, the literature indicates that for highly familiar brands, incongruence led to higher brand attitudes (Dahlén & Lange, 2004). In this research, high Brand Familiarity did lead to a significant relationship between Congruence and Attitude towards the Brand, but results indicate that congruent messages led to more favourable brand evaluations.
Finally, Need for Cognition had a small positive effect on Attitude towards the Corporate Brand, meaning that respondents with a high Need for Cognition had slightly more favourable brand attitudes. Furthermore, Kao (2011) found that two-sided messages yielded more favourable results for respondents that had a high Need for Cognition. However, the further analysis revealed that there was no difference in effectiveness of Message Sidedness regardless of the level of Need for Cognition. However, for respondents with a low Need for Cognition, Message Sidedness no longer affected Corporate Credibility.