OBJECTIVE: To develop and assess the feasibility and psychometric properties of multi-source feedback questionnaires to monitor international medical graduates practising in Canada under 'defined' licences. METHOD: Four questionnaires (patient, co-worker, colleague and self) were developed and administered in 2 phases through paper-based and telephone or Internet formats. Reliability was assessed with Cronbach's alpha and generalisability coefficient analyses. Validity was established through mean ratings, 'unable to respond' rates and factor analyses. RESULTS: A total of 37 doctors participated in the 2 phases. Overall response rates were 70% for patients, 86% for co-workers, 72% for medical colleagues and 92% for self, with response rates higher for the paper-based format than the Internet and phone formats. The instruments had high internal consistency reliability, with Cronbach's alphas of 0.83 for self-assessment and > 0.90 for the other instruments. The generalisability coefficients were Ep(2) = 0.71 for 25 patients on a 13-item survey, Ep(2) = 0.59 for 8 co-workers on a 13-item survey, and Ep(2) = 0.67 for 8 colleagues on a 21-item questionnaire. The range of mean scores was narrow (between 4 and 5) for all items and all surveys. The factor analyses identified that 2 factors accounted for 70% or more of the variance for the patient and colleague surveys and 60% of the variance for the co-worker survey. CONCLUSION: These data suggest that the instruments have reasonable psychometric properties. Traditional survey methods (i.e. paper-based) yielded better results than Internet or phone methods for this group of doctors.
OBJECTIVE: To develop and assess the feasibility and psychometric properties of multi-source feedback questionnaires to monitor international medical graduates practising in Canada under 'defined' licences. METHOD: Four questionnaires (patient, co-worker, colleague and self) were developed and administered in 2 phases through paper-based and telephone or Internet formats. Reliability was assessed with Cronbach's alpha and generalisability coefficient analyses. Validity was established through mean ratings, 'unable to respond' rates and factor analyses. RESULTS: A total of 37 doctors participated in the 2 phases. Overall response rates were 70% for patients, 86% for co-workers, 72% for medical colleagues and 92% for self, with response rates higher for the paper-based format than the Internet and phone formats. The instruments had high internal consistency reliability, with Cronbach's alphas of 0.83 for self-assessment and > 0.90 for the other instruments. The generalisability coefficients were Ep(2) = 0.71 for 25 patients on a 13-item survey, Ep(2) = 0.59 for 8 co-workers on a 13-item survey, and Ep(2) = 0.67 for 8 colleagues on a 21-item questionnaire. The range of mean scores was narrow (between 4 and 5) for all items and all surveys. The factor analyses identified that 2 factors accounted for 70% or more of the variance for the patient and colleague surveys and 60% of the variance for the co-worker survey. CONCLUSION: These data suggest that the instruments have reasonable psychometric properties. Traditional survey methods (i.e. paper-based) yielded better results than Internet or phone methods for this group of doctors.