Data analysis
Prior to analysis, diagnostic and normality tests were conducted.
Scatter plots, residual plots, normal quantile plots and histograms
were generated, and all data were found to meet normality
assumptions, excluding the JCQ section scores. Decision latitude,
physical job demands and psychological job demands showed
a negative skew whereas, social support, organizational level, job dissatisfaction, and job insecurity showed positive skews. Transformations
were unsuccessful in achieving normality. The analyses
used here (regression) are known to be robust to normality
assumptions, therefore, these data were used in their raw format.
All analysis was done using the Statistical Analysis Software (SAS)
9.2 and JMP 7 software from SAS (SAS Institute Inc., Cary, NC).
Three multiple linear regression models were developed for
each risk factor category individually first to predict ODI scores.
Each model considered main effects and all two-way interactions.
The regression models were built using a stepwise selection
procedure, where the significance level to enter (SLE) ¼ 0.250 (to
include any potential factor or interaction) and the significance
level to stay (SLS) ¼ 0.100 (to remove any factors or interaction that
had little predictive value). Interaction terms were included in the
model only after including the individual variables in the model.
Additionally, a combined multiple linear regression model
including the risk factors and all possible two-way interactions was
developed using the same stepwise model building methodology
as described previously. All occupational and personal risk factors
were included in the model. In case of the psychosocial factors, to
check for multicollinearity issues, correlations between the
psychosocial factors were performed. It was found that PSS was
found to be significantly correlated with all the other factors and
therefore, only PSS was included in the final model. For all models,
adjusted R2 and the model building steps are reported.
Data analysisPrior to analysis, diagnostic and normality tests were conducted.Scatter plots, residual plots, normal quantile plots and histogramswere generated, and all data were found to meet normalityassumptions, excluding the JCQ section scores. Decision latitude,physical job demands and psychological job demands showeda negative skew whereas, social support, organizational level, job dissatisfaction, and job insecurity showed positive skews. Transformationswere unsuccessful in achieving normality. The analysesused here (regression) are known to be robust to normalityassumptions, therefore, these data were used in their raw format.All analysis was done using the Statistical Analysis Software (SAS)9.2 and JMP 7 software from SAS (SAS Institute Inc., Cary, NC).Three multiple linear regression models were developed foreach risk factor category individually first to predict ODI scores.Each model considered main effects and all two-way interactions.The regression models were built using a stepwise selectionprocedure, where the significance level to enter (SLE) ¼ 0.250 (toinclude any potential factor or interaction) and the significancelevel to stay (SLS) ¼ 0.100 (to remove any factors or interaction thathad little predictive value). Interaction terms were included in themodel only after including the individual variables in the model.Additionally, a combined multiple linear regression modelincluding the risk factors and all possible two-way interactions was
developed using the same stepwise model building methodology
as described previously. All occupational and personal risk factors
were included in the model. In case of the psychosocial factors, to
check for multicollinearity issues, correlations between the
psychosocial factors were performed. It was found that PSS was
found to be significantly correlated with all the other factors and
therefore, only PSS was included in the final model. For all models,
adjusted R2 and the model building steps are reported.
การแปล กรุณารอสักครู่..
