The Role of Usability in Business-to-Business E-Commerce Systems: Predictors and Its Impact on User's Strain and Commercial Transactions

Table 2

Summary of fit indices.

Model	df	𝜒^2	𝜒^2/df	CFI	NFI	RMSEA	90% CI	SRMR
Fit indices for full mediation model
Full mediation model	86	82.71	ns.	1.00	0.97	0.00	.00, .02	.04
Fit indices for full partial mediation models
Partial mediation model 1	82	81.90	ns.	1.00	0.97	0.00	.00, .03	.04
Partial mediation model 2	82	80.37	ns.	1.00	0.97	0.00	.00, .02	.05
Partial mediation model 3	78	78.79	ns.	1.00	0.97	0.01	.00, .03	.04
Fit indices for direct effects models
Direct effects model 1	87	92.56	ns.	0.99	0.96	0.01	.00, .03	.06
Direct effects model 2	86	97.74	ns.	0.99	0.97	0.02	.00, .03	.06
Direct effects model 3	82	92.46	ns.	0.99	0.96	0.02	.00, .03	.06

Note: 𝑁=491. CFI: comparative fit index; RMSE: root-mean-square error of approximation; CI: confidence interval; SRMR: standardized root-mean-square residual. Full mediation model equals the hypothesized structural model (the best fit model; see Figure 1). The partial mediated model 1 for additional paths from organizational support (OS), information policy (IP), and subjective server response time (SSRT) on strain, the partial mediation model 2 for additional paths from OS, IS, and SSRT on purchase, the partial mediation model 3 for additional paths from OS, IS, and SSRT both on strain and on purchase (i.e., fully recursive model, where every structural path was estimated). The direct effects model 1 for additional paths from usability on strain (i.e., the mediated paths from OS, IS, and SSRT to usability was constrained to zero), the direct effects model 2 for additional paths from usability on strain on purchase, the direct effects model 3 for additional paths from usability on strain both on strain and on purchase.