Measuring attitude change in medical students: lessons from a short course on global health

Educators seek to influence attitudes as well as enhance knowledge and skills. We wanted to know how a one week course on global health had altered the attitudes of the learners. We were aware that analysing group responses to Likert items using an interval approach has methodological flaws. All 126 third-year students at Trinity College Medical School were invited to complete an attitude survey before and after the course, with 18 Likert items. Changes in attitudes to each item were analysed by interval and ordinal methods. A total of 82 (65%) students completed both surveys and gave consent. There was a nine-fold difference in attitude change by individual students. Attitudes changed markedly in relation to some items and little to others. Analysing responses using an ordinal method revealed significant change regardless of whether responses to each item before the course were clustered at one end of the scale or were more neutral. The most sensitive and reliable analysis was an ordinal approach, presenting degrees of attitude change by individuals to each item. When presented by the ‘change step’ method in a histogram important attitude change by individuals within the group is revealed, that the interval method conceals.