Racial stereotyping: survey of psychiatrists in the United Kingdom

摘要

Ten years ago, psychiatrists rated black male patients as potentially more violent than white patients. We aimed to establish whether such racial stereotyping still occurs. We sent a postal questionnaire concerning the first presentation of a young man at casualty-which included a photograph, brief history, and findings on the patient's mental state-to a random sample (generated by SPSS statistical software) of 1000 British psychiatrists obtained from the Royal College of Psychiatrists' database. The sample was randomised so that half received a picture of a black man and half received a picture of a white man. (Photographs were of one of four healthy volunteers, whom we had not seen previously; they were matched for age and occupation, and photographed under identical conditions.) To exclude the possibility that results stemmed from differences between individual photographs, such as facial expression and mode of dress, we photographed two men from each race; one was a footballer and the other an academic (the photographs can be seen on BMJ's website). We used recommended terminology for ethnicity. Respondents were asked to rank five questions, in order of importance, to supplement the assessment χ~2 tests compared "black" with "white" questionnaires after questions were grouped into "important" (ranking 1-2) and "less important" (ranking 3-5). Respondents rated questions on management issues by putting a cross on a 10 cm continuous line. For each question, mean scores for "black" and "white" questionnaires were compared using the Mann-Whitney U test (table). Of the 823 psychiatrists who could be contacted (18% had changed address or retired), 59% (n = 485)-equivalent to 10% of British psychiatrists-returned completed questionnaires. Forty eight per cent (232) had received a "black" questionnaire. Fourteen respondents, who had all received a questionnaire with a photograph of a black man, guessed the hypothesis; six completed the questionnaire and were included in the analyses. Five others returned questionnaires uncompleted. Prior power calculations, based on expected mean (SD) risks of violence of 2.41 (1.76) v 2.87 (1.53), gave the study 85% power at the 5% level.