Purpose: Surveys of medical students are widely used to evaluate course content and faculty teaching within the medical school. Gathering information that accurately reflects student perceptions requires that students buy into the evaluation process and be willing to provide thoughtful responses to the teaching evaluation. To maintain student commitment, it is important that medical students are not overburdened with poorly planned evaluations. Sampling might decrease the number of evaluations required of students and might also reduce the proportion of non-responses and other forms of inattentive response biases.
Methods: A sampling technique employed within a large medical lecture is described and evaluated. A generalisability study of the teacher evaluations is conducted.
Results: A high response rate and high levels of reliability were obtained by sampling a small proportion of the total class. The largest source of error was related to rater and utilising sufficient numbers of student-raters is critical to achieving reliable results.
Conclusion: Sampling can reduce evaluation demands placed on students, and preserve reliability and increase the validity of mean evaluation scores. With computer presentation, efficient sampling techniques become practical and should be part of software packages used to present teacher evaluations.