Examining the Relationship between Type of Question and Score Earned in a Database Class

Abstract

In many modern tests, there are different types of questions assigned. Different tests, quizzes, or
assignments may have varying question types. In this study, we attempted to examine the impact
of the effect of the type of question (e.g. true/false and fill in the blank questions) on the
corresponding score a student receives for that type of question. Real student data from a George
Mason University undergraduate course was used in the study. The process included data
anonymization, data aggregation and data processing. We used an iterative process to
anonymize these files- we started by anonymizing the file name with a section and session key,
next we deleted and appended necessary columns to a new csv file, and finally we determined
the question type using a basic algorithm. After the files were anonymized, we wrote another
program to analyze the data, and we found that true/false questions were answered with 79.16%
correctness, while the fill in the blank questions were only answered with 72.52% correctness. We
plan to conduct a t-test to further analyze the significance of our results. From this research
project, we concluded that students do better on true/false questions compared to fill in the blank
questions on average, in database assignments assigned in the Database GMU class