Examination of the adaptation of the examination stress scale-and the measurement invariance

Within the scope of this research, validity and reliability studies are made by adapting Examination Stress Scale developed by Yao-Ting Sung and Tzu-Yang Chao (2015) to Turkish and a model for examination stress is tested. In the second phase of the study, the measurement invariance was examined in the sub-groups of gender, school type and class level of the examination stress model. The study group consists of 1617 students in the 9th, 10th, 11th ve 12th grades of science high school, anatolian high school, technical high school and social sciences high school in Balıkesir in 2016-2017 education period. Data collection tool SSÖ is a five-point likert scale consisting of 27 subject matters. The original form of the tool was translated into English by five experts during the adaptation of the SSÖ, which was first aim of the research, to the Turkish. The English and Turkish scales were applied to a group knowing both languages well, first the English form and then the Turkish form. Analyzes were conducted to examine the linguistic equivalence between the scores obtained from the original form of the scale and the scores obtained from the Turkish form. Correlations between the two forms of the scale were investigated by Pearson Moment Product Correlation Coefficient. Correlation coefficients obtained from the analyses result subscales were found to be ,93, ,87, ,96 ,respectively. The correlation coefficient obtained from the whole scale is ,97. The difference between the scores was examined via the t-test. As a result of the analysis, there was no significant difference between the scores and it was decided that the scale was equivalentAfter the linguistic equivalent was achieved, the scale for the actual application was carried out by means of 1617 students in a two week period of December 2016-2017 academic year. Test and item statistical analyzes were performed on the data. Subtest correlations calculated for subtest discrimination were not removed from the resultant scale. The reliability was calculated by the Cronbach-alpha internal consistency coefficient for all sub-scales of the scale. The Cronbach-alpha coefficients obtained from the rank and the sub-scales were ,87, ,75, ,78, and ,88 for the whole scale. In addition, considering the differentiation of factor load values and error variances, the compound reliability coefficient is calculated because it is considered that the compound reliability coefficient calculated for multidimensional scales has a stronger reliability value than the Cronbach-alpha coefficient. Confirmatory factor analysis on the validity level of the scale was conducted to test the three-factor structure of the scale. These findings were handled as evidence of the validity of the scale.As the test stress model obtained from the SSÖ was considered to be interpreted differently from the sub-groups, the measurement invariance of the model according to the variables of gender, school type and class level was examined in the research. In the first stage, the SSÖ is composed of 27 items and three dimensions confirmed by confirmatory factor analysis . After it was determined that all the data fit well with the model, the model data consistency was tested separately for each group in the sub-groups. lt has been seen that modeling in each of sub-groups is well adapted. Subsequently, multi-group confirmatory factor analysis technique was used to examine the measurement invariance in the sub-scales of gender, school type and class level of the examination stress model. The invariance tests were carried out in a gradual manner and ΔCFI values are interpreted based on the difference between the unrestricted model and the more limited model. lt was seen that the examination stress model provided all the stages of invariance in the gender, school type and class level sub-groups. Based on findings, it can be concluded that all comparisons made in sub-groups of gender, school type and class level are significant. This suggested that SSÖ provides valid and reliable measures at different levels in sub-groups of gender, school type, and class type. Keywords: Multi-group confirmatory factor analysis, Measurement invariance,Examination stress, Structural equation model