Meta-analysis:a study on reliability generalization of Multidimensional Students' Life Satisfaction Scale

Many systematic review techniques are used for literature reviews. Meta-analysis is a type of literature review and evaluation method which can achieve results that are quantitative and more comprehensive than the results obtained by these methods. Reliability of the meta-analysis technique has incresed with the advancement in the statistical methods, possibility of including majority of the work related to the research in the analysis, and development of the internet.The objective of this thesis includes, providing detailed information on meta-analysis and describing how reliability generalization of a measure utilizing meta-analysis is performed over the application on the Multidimensional Students? Life Satisfaction Scale (MSLSS) of Huebner (1994). In the research, the reliability generalization approach which is defined by Vacha-Haase (1998) has been utilized for explaining the reliaibility of the MSLSS scores of the published studies.265 studies which use MSLSS have been determined as a result of the literature review. However, only 14 of them have been included based on the determined inclusion criteria. 23 different samples have been obtained from these 14 studies. Relevant data from the literature pool have been input to the SPSS 19.0 software. As a result of the statistical calculations like descriptive statistics, Q homogeneity test, and Fisher's z-transformation general reliabilities of MSLSS scores and their 95% confidence intervals have been obtained. Also, regression analysis has been performed in order to determine the factors influencing reliability generalization.It has been found that the general reliability coefficients and 95% confidence intervals of the scores of five dimensions (identity, family, friend, life environment and school) which constitute MSLSS are at a sufficient level. It has been concluded that none of the hierarchical regression models that are established for the five dimensions of MSLSS are statistically significant; hence, gender, average age, originality code, sample size, year code, and country code variables do not have an effect on the reliability coefficient.