أداء مؤشر مطابقة الفرد l_z^p في الكشف عن الاستجابات غير المطابقة للفقرات متعددة التدريج باختلاف عدد فئات التدريج ومستويات القدرة
The aim of study was to investigate the performance of l_z^p person fit statistic in detecting misfit responses according to Likert item scale points (3, 4, 5, and 7 points Likert type items) and ability level. To achieve the goals of this study, cognitive biases scale (DACOBS) was used to collect the data after creating four forms of the scale according to the number of item-scale points (3, 4, 5, and 7 points Likert type items). The population of the study consisted of all undergraduate students at Yarmouk University, four samples with a total of 1495 students were selected using multi-stage sample procedures. IRT graded response model was used to estimate item and ability parameters. Then l_z^p was calculated for every responder in each sample. The results showed that l_z^p distributions do not change over all Likert Item Scale Points (3, 4, 5, and 7), and still not normally distributed and negatively skewed. Moreover, the results showed that lowest detected rate of misfit response was when using three-point Likert type items (10.7%) and the highest detection rate was when using four-point Likert type items (No neutral option) (19.7%). (Author’s abstract)