Masoumeh - University of Tehran
Seyyed Mohammad - University of Tehran
Hossein - University of Tehran, Aborz Campus

This study investigates some problems of PhD applicants in their entrance exams in the case of answering the reading comprehension questions. To this end, the researchers considered the item types and text types as main effects and their interaction effect using generalizability theory for examining the variability; the answer sheets of a mock-test from 321 applicants, from all parts of Iran, enrolled in an institute were randomly selected. Using a partially nested design of G-theory in the GENOVA program, the researchers identified five variance components in the two different passages with distinct items and investigated various sources of error that are involved in the measurement process. The results of the study showed that the main effect for items cannot be separated from the interaction between items and texts, and clarified that the items facet had a noticeable amount of variance, and therefore, they impacted the applicants' performance. However, the results of D-studies showed that the main effect for text types was zero, and both texts were at the same level of difficulty. Also, the persons had effects on the texts in their interaction. This study can motivate the researchers, test developers, and test designers to consider their work more carefully.

Key Words: D-study, G- study, Generalizability theory, PhD Entrance Exam, reading comprehension