Consistency of Assessment

There are three dimensions of consistency, which affect the quality and value of assessment:

Consistency over time

How many samples of one student's work need to be collected to know how well he or she writes? Emerging research on performance assessment, for example, indicates that somewhere between 8 and 16 trials of student performance may be necessary in order to obtain a reliable judgment.

Consistency from student to student

Does the assessment trigger the same dimensions of performance again and again with different students? If not, the assessment cannot be said to be reliable for any type of group testing.

Consistency of raters

Do judges looking for the same dimensions of performance see them consistently across the raters themselves? Would two people with the same criteria for judging rate the same student performance differently?