Statistical Modeling Study For Occupational Stress Of Doctors And Nurses

This subject studied doctors and nurses's occupational psychological problems.Modern statistics modeling techniques were applied to show the plentifulinformation from different sides. Strategies were provided in order to solve theoccupational mental hygiene problem which the doctors faced at present and thehospital's occupational pressure Managements. Main content as the following:(1) On the foundation of data-cleaning and missing value estimation, the surveyquality of data was evaluated systematically. Reliability analysis results ofoccupational stress inventory revised edition (OSI-R) showed that theinvestigation was reliable. Variance component model was employed to estimatethe individual validity and find the individual validity mean of the entire clinicaldepartment besides surgical department were higher than those ofemergency-treatment, assistant, administration departments, etc. The results ofindividual validity satisfied the purpose which doctors and nurses were treatedas main subjects.(2) Construct validity of occupational stress inventory revised edition (OSI-R)was tested by structure equation model. Single factor analysis results showedthat every one item (or latent variables) of three questionnaires has a finevalidity, but some subitem's factor loading was low. It implied that thesesubitems need be revised or verified further in other studies. Multi-factoranalysis results showed that the validity of Personal Strain Questionnaire (PSQ)and Personal Resources Questionnaire (PRQ) can be accepted as a whole andOccupational roles questionnaire (ORQ) is not too well. The main reasonsmaybe lie in cross-measurement between latent variable and observed variable.It implied that Chinese edition of OSI-R need to be improved further. 3山西医科大学博士学位论文(3) Multilevel confirmatory factor analysis model can improve the fitting effectthan single level model when the data have hierarchical structure character.MLwiN statistics package and LISREL statistics package were utilized toexemplify Multilevel CFA or SEM model with single factor and multi-factor.The model's parameter's convergence was related to dimension of the variables,when there are many variables, it is difficult to obtain robust covariance matrixestimation. It is one of the bottlenecks for multilevel SEM model.(4) As the middle outcome of confirmatory factor analysis model—latentvariable score, it can be applied for further statistical analysis. The standardizedestimated value of latent variable score was obtained with LISREL 8.53 in thispaper. Some meaningful results can also be acquired by analyzing latent variablescore.(5) On the basis of latent variable score, Multivariate variance component modelwere fitted and the results showed that the data's characteristics of hierarchicalstructure was obvious. Especially the three variable of Vocational Strain (VS),Interpersonal Strain (IS) and Physical Strain (PHS) were comparativelyoutstanding among them, it is more reasonable that the multilevel theory wasemployed to analyze the data.(6) Results of multilevel multiple regression model showed that roleinsufficiency (RI), role ambiguity (RA), role boundary (RB), responsibility (R)and physical environment (PE) caused the increase of vocational strain (VS),psychological strain (PSY) and interpersonal strain (IS) in varying degrees. ROdoes only have statistical meaning with physical strain (PHS). In addition,physical strain (PHS) had close relation with role insufficiency (RI), roleambiguity (RA), role boundary (RB), and physical environment (PE) exceptresponsibility (R). Proper recreation (RE) can alleviate the individual'sPsychological strain (PSY) and physical strain (PHS); self-care (SC) was theeffective good medicine on decreasing physical Strain too. Four variable ofpersonal strain questionnaire (PSQ) decreased with the increase of socialsupport (SS) and rational/cognitive copin...