This study describes the influence of social desirability on Rokeach Dogmatism scores and demonstrates how misleading self-report inventories can be. Dogmatism instruments with instructions suggesting either that commitment or control of personal beliefs is desirable were randomly distributed to 20 anonymous elementary teachers before a workshop on values clarification. After the workshop, the instrument was readministered without set-inducing instructions. Differences in instructions produced the expected differences in preworkshop dogmatism scores. There was not change from pre- to posttest within treatments. Posttest scores still differed between groups. Clearly, teachers adapted responses to conform to perceived desirability. (Author)