tructural equation model results indicated a good fit between the model and data. The
model’s GFI was 0.97, and the AGFI was 0.93. All other fit indexes scores were 0.99 or
higher. The chi-square test p-value was 0.16, meaning that there is no significant
difference between the model and the data. Also, the chi-square/df ratio was 1.5, well
within accepted guidelines. Similarly, the RMSEA index was 0.06; showing a good fit
between model and sample data collected for this study. Table II provides full fit
statistics results.
After model adequacy was established, variable relationships were examined.
Path tests showed a significant negative link between worker creative environment
perceptions and intent to turnover. This finding is in support of theoretical
expectations. The standardized path coefficient between these two variables was
0.77. In practical terms, this result means that for every 10 per cent increase in
creative environment perceptions, there is an expected 7.7 per cent decrease in
turnover intentions. All manifest variables significantly loaded on the expected
latent variables. Full path result information is presented in Table III, and
graphically in Figure 2.
In addition to the main hypothesis test, a comparison was made between non-skilled
and professional (skilled and managerial) worker results in order to check for
subsample differences. For this comparison, the sample was split between non-skilled
and professional workers, and the SEM model was tested for each group. Next, the
relationship between creative environment and turnover intentions for professional
workers was fixed to be the same as the relationship for non-skilled workers. This
fixed model was compared to the non-fixed model for the professional workers using
the Bayesian inference coefficient (BIC). If there were substantive differences (i.e.
discrepancies due to more than sampling variation) this number would be high. The
difference between the two model’s BIC scores was less than five (26 for the free
model, and 22 for the fixed model), and indicates that any differences between nonskilled
and professional workers can be attributed to random variance (Raftery, 1993).
tructural equation model results indicated a good fit between the model and data. Themodel’s GFI was 0.97, and the AGFI was 0.93. All other fit indexes scores were 0.99 orhigher. The chi-square test p-value was 0.16, meaning that there is no significantdifference between the model and the data. Also, the chi-square/df ratio was 1.5, wellwithin accepted guidelines. Similarly, the RMSEA index was 0.06; showing a good fitbetween model and sample data collected for this study. Table II provides full fitstatistics results.After model adequacy was established, variable relationships were examined.Path tests showed a significant negative link between worker creative environmentperceptions and intent to turnover. This finding is in support of theoreticalexpectations. The standardized path coefficient between these two variables was0.77. In practical terms, this result means that for every 10 per cent increase increative environment perceptions, there is an expected 7.7 per cent decrease inturnover intentions. All manifest variables significantly loaded on the expectedlatent variables. Full path result information is presented in Table III, andgraphically in Figure 2.In addition to the main hypothesis test, a comparison was made between non-skilledand professional (skilled and managerial) worker results in order to check forsubsample differences. For this comparison, the sample was split between non-skilledand professional workers, and the SEM model was tested for each group. Next, therelationship between creative environment and turnover intentions for professionalworkers was fixed to be the same as the relationship for non-skilled workers. Thisfixed model was compared to the non-fixed model for the professional workers usingthe Bayesian inference coefficient (BIC). If there were substantive differences (i.e.discrepancies due to more than sampling variation) this number would be high. Thedifference between the two model’s BIC scores was less than five (26 for the freemodel, and 22 for the fixed model), and indicates that any differences between nonskilledand professional workers can be attributed to random variance (Raftery, 1993).
การแปล กรุณารอสักครู่..