The Effects Of The Reliability Levels And The Difference In Difficulties Between Tests On The Two CTT Equating Methods

Test equating, an important field of psychological and educational measurement,has been being concerned for more30years. The main reason is that people hasnoticed the unfairness when they compared scores among different test forms directly.Test forms tend to differ to some extent, even though constructors struggle to developparallel test forms. So subjects`scores on different test forms cannot be compareddirectly and we need test equating. For test equating could be influenced by manyfactors, it is a concerned aspect to detect how these factors influence test equating.Using Monte-Carlo method, this research detected effects of reliability levels anddifference in difficulty between test forms on linear-equating method and equalequipercentile-eqauting method. The conclusions showed as follows:a. The mean absolute error tends to increase as reliability level of test formsdecrease.b. The mean absolute error tends to increase as difference in difficulty betweentest forms increase.c. The effect of reliability level is not obvious, when difference in difficultybetween test forms is0or0.4; the mean absolute error tend to increase asreliability level decreases, when difference in difficulty between test forms is0.8.d. The effect of difference in difficulty is smaller for equipercentile equating thanit for linear equating.