Customer image research Results Essay Example

relationships between the key variables.

Significant correlations were found for the client's acquiescence, the number of years the participant had been auditing the client, the auditor's experience, and the auditors' identification with their profession when the significance level (p) was 0.05 or less. Similarly, a significant correlation was found between client identification with images of clients and importance, the number of years the client had been audited by the participant, and auditors' identification with professions at a significance level of (p;0.01). The same was true for client image and client importance with the size of the client at a significance level of p;0.05. However, it should be noted that these correlations are not strong enough to suggest that the variables used in this study indicate a similar underlying construct.

Analysis of Measurement Models

The first step involved performing an analysis of factors for the study's latent variables: client identification, client image, and professional identification.

The purpose of the study was to confirm the structure of the items and determine if they align with the same underlying factor. To achieve this, an item from the client identification scale was removed based on loadings and modification indices. Anderson and Gerbing (1988) recommended deleting items that do not fit well with the latent factor. The results of the analysis show no discrepancy between the observed data patterns and the measurement model analysis. Therefore, the null hypothesis, indicating a good fit, is not rejected.

The chi square value above qualifies the measurement analysis model as acceptable. The RMSEA (Root Mean Square Error of Approximation) for this analysis is 0.07, which is less than 1, making the measurement analysis model acceptable. The estimate RMSEA is

0.07 with a 92% confidence level. The results of the confirmatory factor analysis are displayed.

When assessing the fit of measurement models, several indicators are considered. The RMSEA, as stated by Rigdon (1996), has reasonable approximation errors as long as its values are up to 0.08. The CFI, IFI, GFI, and AGFI are other measures used in this evaluation, and they can range from 0 to 1.0. Hair, Anderson, Tatham, and Black (1998) argue that values greater than 0.90 indicate a good fit.

Smaller standardized RMR values (0.05 or less) suggest a satisfactory model fit. The measurement model's fit is considered acceptable for all indices, except for AGFI. Evaluation of the validity and reliability of index items indicates that constructs are significantly related to items in all cases (p;0.01). Following guidelines by Hair, Anderson, Tatham, and Black (1998), items representing the latent construct should have high variance extracted values (above 0.50). It is evident that all three constructs' measures exceed this threshold.

Additionally, the composite reliabilities exceed the recommended threshold of 0.70. Next, a principal component analysis was conducted to assess the distinctiveness of the constructs. The analysis included items from three constructs, namely client identification, client image, and professional identification. Except for one item with a loading of 0.43, all client identification scale items loaded above 0.50 on their intended factor. No cross-loadings higher than 0.30 were observed.

From the above results, it is evident that the measures employed in this study assess three distinct constructs. Based on the factor loadings, validity and reliability, and the outcomes of the confirmatory factor analysis, it can be inferred that the measurement model is satisfactory. To examine if common

method bias influenced the findings, the original 9-factor model was modified into a 6-factor model. By comparing the two models, it was found that the data was a better fit for the original 9-factor model (X

²

=262.65, df=101) than the 6-factor model (with a chi-square value of 1058.68 and df=125).