The mystery of 'b := (b = false)'

摘要

This paper describes some unusual patterns that emerged from a statistical analysis of the 1988 Advanced Placement Exam in Computer Science. Most multiple-choice questions on the exam had few significant correlations with other parts of the exam. But a small set of five questions had a nontrivial correlation with many parts of the test. One question in particular demonstrated such correlations. It asked about the effect of the assignment statement "b := (b = false)" for a boolean variable b. One interpretation of this data is that these questions are testing general programming aptitude. The paper presents the analysis along with a discussion of the possible implications.