The reliability of analytic and holistic methods in rating students' computer programs

The study examined analytic and holistic methods of scoring students's programs. The sample included 30 students in an introductory computer science class. The inter-rater reliabilities for both scoring methods was significant (analytic, r = .66; holistic, r = .73). The correlation between students' holistic and analytic scores on the programs was moderate (r = .66), indicating that the two scoring methods measure different attributes of a student's program.