leadership and motivation in hospitality

 RMSEA is 0.000 (upper limit 0.059; lower limit 0.000 with a pclose value of

0.899 (i.e.≥0.05) indicating that this finding for RMSEA is robust);

 CFI is satisfactory at 1.000 (i.e. >0.96); and

 SRMR (0.0314) is satisfactory at <0.09.

Hoelter’s CN (indicating the adequacy of sample size for the model) is now

considerably above the recommended value of 200 (for CFA1:1 CN was 85 and for

CFA 1:2 CN was 185).

AVE values are all above 0.5 (having improved considerably on those for CFA 1:1)

and all of the CR values are above 0.8 (in the range described by Kline (2005: 59)

as ‘very good’ to ‘excellent’). All of the inter-factor correlations remain

statistically significant and the estimates for these are illustrated in Table 7 - 7.

Factor correlation Correlation estimate
ML→JP 0.409
ML→DSB 0.377
JP→DSB 0.320

Table 7-7 Correlation estimates for CFA 1:3

The measures for discriminant validity are satisfactory with none of the squared

correlation estimates exceeding the lowest of the AVE (average variance

extracted) estimates (Table 7 - 8 ).

ML JP DSB

ML (^) 0.767 - -
JP 0.167 0.618 -
DSB 0.142 0.102 0.733
AVE values are on the diagonal and squared correlation estimates below the diagonal

Table 7-8 Discriminant validity estimates for CFA 1:3

An examination of the standardised residual covariance matrix reveals no

standardised residuals exceeding 1.96.

A final consideration regarding the parameter estimates for CFA 1:3 is the issue of

multivariate non-normality. West et al. (1995) provide details of using the

bootstrap procedure in SEM analyses to evaluate the stability of parameter

estimates under conditions of multivariate non-normality – a condition which

Byrne (2010: 330) notes is common to most data sets in SEM analyses. In this