Results

 < Day Day Up > 



Before the EC-adapted SERVQUAL can be considered a good measure of perceived service quality, it is necessary to assess its reliability and validity. It is hypothesized that e-commerce service quality is a second-order construct governing the correlations among such factors as tangibles, reliability, responsiveness, assurance, and empathy. In this sense, gap scores (Q1 to Q22) of the 22-item EC-SERVQUAL instrument were directly subjected to second-order CFA. After examining the multivariate normality and model identification, the parameters were estimated using the maximum likelihood (ML) method.

To find a proper level of model-data fit, the general guideline suggested by MacCallum (1986) and Anderson & Gerbing (1988) were used in the specification search process. Seven common model-fit measures were used to assess the model's overall goodness of fit: the ratio of Χ2 to degrees-of-freedom (df), goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), normalized fit index (NFI), non-normalized fit index (NNFI), comparative fit index (CFI), and root mean square residual (RMSR). The initial model's results (Model 1 in Table 4) did not fit the data very well. After examining the squared multiple correlations (SMC), standardized residuals, and t-values, six items (item 1, 2, 3, 4, 14, 19) with lower than .5 of SMC were dropped from Model 1. The model's fit after making these modifications is shown in Table 4 (Model 2). Model 2 provided acceptable model-data fit, as evidenced by goodness-of-fit indices. After examining the modification indices of Model 2, no cross-loading items were found.

Table 4: Goodness-of-fit Indices for Alternative SERVQUAL Models (n=260)

Goodness-of-fit Index

Model 1

Model 2

X2/df

2.66

1.34

GFI

0.87

0.94

AGFI

0.84

0.92

NFI

0.85

0.94

NNFI

0.88

0.98

CFI

0.90

0.98

RMSR

0.05

0.035

Next, a formal test of discriminant validity was performed. Empirically, such evidence can be obtained through the comparison of an unconstrained model that estimates the correlation between a pair of constructs and a constrained model that fixes the value of the construct correlation to 1.0. A significant difference in Χ2 between these models implies that the unconstrained model is a better fit for the data, thereby supporting the existence of discriminant validity (Anderson, 1987; Bagozzi & Phillips, 1982; Bagozzi et al., 1991; Gerbing & Anderson, 1988; Venkatraman, 1989). Table 5 contains the results of the pairwise Χ2 difference tests among four EC-SERVQUAL dimensions. As shown, all Χ2 differences are significant at p<0.001. Hence, each of the EC-SERVQUAL dimensions seems to capture a construct that is significantly unique from other dimensions, supporting the existence of discriminant validity.

Table 5: Results of Discriminant Validity Tests: EC-SERVQUAL Dimensions

Test

ML Estimate

T-Value

Constrained Model Χ2

Unconstrained Model Χ2

Χ2 Difference

Reliability with

     
  • Responsiveness

0.83

28.92 [***]

122.06 (27)

42.60 (26)

79.46[***]

  • Assurance

0.61

11.99[***]

173.25 (20)

30.13 (19)

143.12[***]

  • Empathy

0.60

12.12[***]

303.38 (27)

33.97 (26)

269.41[***]

Responsiveness with

     
  • Assurance

0.62

12.32[***]

154.46 (14)

15.91 (13)

138.55[***]

  • Empathy

0.60

12.01[***]

274.68 (20)

18.56 (19)

256.12[***]

Assurance with

     
  • Empathy

0.68

14.02[***]

167.18 (14)

125.17 (13)

42.01[***]

[***]Significant at p < 0.001

Reliability of the factors was estimated by composite reliability. The composite reliabilities can be calculated as follows: (square of the summation of the factor loadings)/{(square of the summation of the factor loadings)+(summation of error variables)}. The interpretation of the resultant coefficient is similar to that of Cronbach's alpha, except that it also takes into account the actual factor loadings rather than assuming that each item is equally weighted in the composite load determination. The composite reliability of each factor was as follows: reliability = 0.879; responsiveness = 0.872; assurance = 0.809; empathy = 0.871.

It is worth noting that the final 16-item EC-SERVQUAL (see Appendix B) completely excludes the tangible dimension (item1 to item4). The observation may also indicate that the tangible dimension is less relevant to the e-commerce service quality in virtual contexts where an e-business's entire activities for digital product marketing are conducted via the Internet.



 < Day Day Up > 



Advanced Topics in End User Computing (Vol. 3)
Advanced Topics in End User Computing, Vol. 3
ISBN: 1591402573
EAN: 2147483647
Year: 2003
Pages: 191

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net