Establishing the validity and reliability of course evaluation questionnaires

Abstract

This article uses the case of designing a new course questionnaire to discuss the issues of validity, reliability and diagnostic power in good questionnaire design. Validity is often not well addressed in course questionnaire design as there are no straightforward tests that can be applied to an individual instrument. The authors propose the technique of establishing validity by deriving constructs from naturalistic qualitative research—in this case by interviewing award‐winning teachers about their principles and practices. Analysis of the interview transcripts led to nine principles of good teaching, which were developed into nine questionnaire scales. Reliability was tested with Cronbach’s alpha and with confirmatory factor analysis, as the use of Cronbach’s alpha alone can mask issues of multi‐dimensionality in scales. The concept of diagnostic power as the ability of an instrument to distinguish between related constructs is introduced. This is important in course evaluation questionnaires, as it enables relative strengths and weaknesses to be identified, which makes it possible to advise on remedial action.