Estimating the Probability of Traditional Copying, Conditional on Answer-Copying Statistics

Area of Science:

• Educational statistics
• Psychometrics
• Academic integrity

Background:

• Traditional statistical methods for detecting test copying, such as p-values, are based on the null hypothesis of no copying.
• P-values represent the probability of observing data at least as extreme as the current data, assuming no copying occurred.
• The posterior probability of copying is more informative but often requires unknown prior probabilities.

Purpose of the Study:

• To develop an estimator for the posterior probability of copying that relies on estimable quantities.
• To create a flexible method applicable to any answer-copying statistic.
• To provide a practical tool for identifying potential academic dishonesty.

Main Methods:

• Development of a novel statistical estimator for the posterior probability of copying.
• Utilizing estimable quantities, making the method broadly applicable.
• Evaluation of the estimator's performance through simulation studies.

Main Results:

• The developed estimator provides a practical way to calculate the posterior probability of copying.
• Simulation results demonstrate the effectiveness and reliability of the proposed method.
• The approach was successfully applied to real-world data, showing its practical utility.

Conclusions:

• The new estimator offers a more relevant statistical measure for detecting copying than traditional p-values.
• This method enhances the ability to identify academic dishonesty in multiple-choice tests.
• The approach is generalizable and has potential applications for other forms of cheating.