A test equating study concerning to ALES (Academic Personnel and Postgraduate Education Entrance Exam) scores obtained at different times in a year

The purpose of this study is to equate ALES (academic personnel and postgraduate education entrance exam) scores with linear equating and equipercentile equating methods in the periods of 2011 spring and autumn conducted by SPSS, to determine the most suitable one of these two methods for research and to propose the most suitable one thanks to the findings in case of a similar study. The study population was 21860 people participating in ALES made in the periods of both 2011 spring and 2011 autumn. The sample of the study were 2186 individuals, respectively selected 10 % of population randomly via computer. The data used in this study was raw scores concerning to the fields of ALES's verbal, quantitative, equally weighted. The data analysis was performed at four stages. At the first stage, descriptive statistics were calculated for all sub-tests; at the second stage, it was checked whether equation conditions were fulfilled; at the third stage, the equated points were found by using linear and equipercentile equating methods; at the fourth stage, it was found out that which equation method was more suitable by obtaining error quantities of equation methods WMSE coefficient. It was calculated whether there was a difference between the mean, the variance, the internal consistency coefficients, the average difficulty about one-dimensionality in order to ensure the equation requirements. In the study, the internal consistency was calculated without eliminating items by using KR-21 formula because datas were raw scores. Analyzes were carried out because of the fact that internal consistencies were high without eliminating agents of heights. When the periods of 2011 spring and autum were analyzed it seemed that all of the prerequisite features were provided. Linear and equipercentile methods applied seperately to the raw scores of sizes. It was found that there was a positive high level corelation between raw scores and equated scores in the first sub-problem. Deviations were observed towards extreme values in the quantitive and equally weighted sub-tests when the difference values are examined but extreme values were negative in the verbal subtest. It was found that there was a positive high level corelation between raw scores and equated scores in the second sub-problem. Also fluctuations were observed between raw scores and graph which is made upof equated scores corresponding to this scores.Error quantities were measured by WMSE formula to determine which equation method was more appropriate for research and when the data of research findings were analysed it was found that WMSE coefficient of equipercentile method was lower. As a result of this study equal percentile equating was found more appropriate on the equation study concerning to the field of quantitative, verbal, equally-weighted. The keywords:Test equating, single group design, linear equating, equipercentile equating.