How representative are volunteer physicians in collaborative research projects?

Few family physicians can generate large samples for statistically significant research, which has led to multicentre collaborative studies. Are the volunteers for these studies typical of family physicians as a whole? If they are not, results will be biased. This preliminary study examines the differences between a group of volunteers and non-volunteers, using several variables. The only statistically significant difference was in place of practice: volunteers were far more likely to come from small towns than from metropolitan or rural areas. Discriminate analysis showed it was possible to predict correctly to which group the physician belonged in only 62% of cases. Further study must be done on other potentially sensitive variables.