The Evaluation Of Domestic Scales On Occupational Stress

Objective: The research of occupational stress in China started later, andmost of its testing tools were adopted from foreign scales. With the differentscales, the research results cannot objectively reflect the stress of the sameoccupational group in China, thus hinder the research of occupational stress inChina. This research adopts some occupational stress scales which arecommonly used in domestic to evaluate the occupational stress in teachers,doctors and nurses. Meanwhile, by scientific and objective evaluation andcomparing the validity and reliability of these scales, the author of thedissertation shall provide a scientific reference for further and normativeresearches on occupational stress.Method: Totally 354 volunteers are randomly chosen as subjects: theyare teachers from the middle or primary schools and doctors and nurses fromtertiary hospitals in Nanchong. The OSI-R, JSS and ERI scales were selectedto evaluate the occupational stress in these subjects. The data analyses oftesting results were conducted by the t test, correlation analysis, exploratoryfactor analysis and confirmatory factor analysis to evaluate and compare thereliability and validity of the scales.Results:(1)The analysis on item discrimination and item-total correlationcoefficient showed that in OSI-R scale, about 73% of all totally 140 items and100% in JSS scale of all 60 items and about 86% in ERI scale of all 22 itemshad higher discrimination and item-total correlation coefficient (P<0.01).(2)The correlation coefficients of all the 14 items with their own subscales inOSI-R scale were ranged from 0.527 to 0.917, among which the correlationcoefficients of ORQ and PSQ with their own total table were 0.709 to 0.843, and that for their own items were ranged from 0.347 to 0.843 (P<0.01), theitems in OSI-R scale were moderately correlated with each other; thecorrelation coefficients of items in JSS scale with its own questionnaires wereranged from 0.462 to 0.941, the correlation coefficients of two questionnaireswith total score were between 0.906~0.919, the correlation coefficients ofitem-total were between 0.742~0.955, the items in JSS scale were highlycorrelated with each other; the correlation coefficients of items in ERI scalewith the total score were ranged from 0.484 to 0.751 (except Rewardquestionnaires), the correlation coefficients of each dimension were between0.440~0.870, and that among the items were ranged from -0.602 to 0.836, thecorrelation coefficients of items in ERI scale were highly correlated with eachother. (3) The Cronbachαcoefficients of the total table and the subsidiarysubscales of the OSI-R scale were 0.862 and 0.864, 0.922, 0.891 respectively;the Cronbachαcoefficients of the two subscales and the total table of JSSscale were 0.901, 0.917 and 0.962, the Cronbachαcoefficients of the eachdimension and the total table of ERI scale were 0.885, 0.897, 0.787, 0.913 and0.902 respectively. The fact that all the Cronbachαcoefficient we got in ourstudy were higher than the recommended value of 0.70 indicated that thescales we selected were of good reliability. (4) The split-half reliability of thethree subscales and the total table of OSI-R scale were 0.917, 0.924, 0.913and 0.907 respectively; the split-half reliability of the two subscales and thetotal table of JSS scale were 0.932, 0.948 and 0.816; and that of the fourdimensions and the total table of ERI scale were 0.878, 0.831, 0.750, 0.784and 0.855. The result that all the split-half reliability we got in our study werehigher than the recommended value of 0.70 demonstrated that the scales weselected were of good reliability and in line with the homogeneous requirements of psychometrics to the scales. (5)The exploratory factoranalysis and confirmatory factor analysis suggested that the OSI-R, JSS andERI all fit indices in the acceptable range, OSI-R, JSS and ERI scale modelswere fitted well with the theoretical models, meanwhile, they were of goodclustering validity, discriminant validity and content validity.Conclusions: (1) The author thinks that exploratory factor analysis andconfirmatory factor analysis indicated that the structure validity of OSI-R,JSS and ERI are basically consistent with their theoretical models; (2)Judging from the evaluation of item discrimination, correlation coefficientand Cronbachαcoefficient, the author draws a conclusion that JSS is optimal,second is ERI and OSI-R is relatively poor; while judging from theperspective of split-half reliability evaluation, the author also reach aconclusion that total scale of OSI-R is optimal, ERI comes the second, andJSS is relatively poor; (3) The study shows that OSI-R, JSS and ERI are ofgood aggregation validity, discriminant validity and content validity;(4) Thestudy also shows the OSI-R, JSS and ERI are of good reliability andvalidity;(5)Judging From the total analysis and validity and reliability analysis,the author of the dissertation holds that JSS is optimal, the second is ERI,OSI-R is relatively poor.