Abstract:
The purpose of this study is to develop and standardize an instrument to measure job satisfaction among teachers using the Rasch model. This was against the backdrop of the use of classical test theory (CTT) by many researchers to assess the psychometric properties of measurement instruments. Item Response Theory(IRT) moves steps further from CTT in determining the psychometric properties of an instrument by providing calibrated measures of equal interval, ensuring unidimensionality of the items of a scale, sieve out misfitting items and detect items that function differentially. Design of the study is Instrumentation. Eight research questions were posed and two hypotheses were formulated to guide the study at p < 0.05 level of significance. The sample was made up of 972 teachers from 36 schools in eight LGAs in Enugu state of Nigeria. Multi- stage sampling procedure was employed. A 90- item draft instrument, TJSS was developed by the researcher and trial tested. The responses were subjected to factor analysis; 35 items were dropped after factor analysis and 55 items emerged. It was found out by using principal component analysis (PCA) that the 55 items of the instrument are unidimensional. Conditional Maximum Likelihood Estimation technique of the WINSTEPS 3:80 software (Linacre, 2010) was used to analyze the data in order to answer the research questions according to the Rasch model (IRT) procedures. Independent sample t-test analysis was applied to test the hypotheses at p¬ < 0.05 level of significance. 52 out of the 55 items fit the Rasch modeled expectation, that is, 3 items were misfitting and were dropped. Item and person separation indices showed good alignment of item hierarchy and that persons have been separated into higher and lower levels of job satisfaction. The item threshold and person parameter estimates have accompanying standard errors of measurement (SEM) that were far below 0.5, meaning that the parameters were measured precisely. Differential Item Functioning (DIF) effect was discovered in approximately 29% of the items, precisely 16 items, which were also discarded. With the exclusion of 19 items from the 55 items that survived factor analysis, made up of 3 misfitting ones and 16 for DIF effects, 36 items finally emerged for the TJSS. It was recommended that IRT techniques be adopted by psychometricians so that the psychometric properties of a scale can be determined item by item and for the calibrated items to be used to measure Teachers’ Job Satisfaction in all primary and secondary school settings.
