Assessing population-wide behaviour change: concordance of 10-year trends in self-reported and observed sun protection

Increases in socially desirable responses in self-reports might occur in the context of ongoing public education. We examine concordance of trends in two long-term studies monitoring population impact for SunSmart. One study employed telephone interviews of Melbourne residents; the other entailed observations at public recreation venues across Melbourne. The studies assessed people's sun protection on identical weekend dates (Nw = 33 dates). Data from five summers between 1992 and 2001 (n similar to 23,000 individuals) were analysed. A body cover index score was calculated for participants on each date. Outcomes were aggregated separately for Saturdays and Sundays by date and year. Regression analyses tested whether these trends differed by survey method. The pattern of change in body cover over time was similar for both surveys. Self-reported body cover was consistently higher than observed body cover, suggesting that social desirability bias may be present. Regression analyses showed no divergence between self-reported and observed trends in mean body cover, suggesting no evidence of significant increased social desirability bias in self-reporting over time. Findings suggest that self-report offers a valid means of assessing change in a population's sun protection compliance over time, at least when self-reports are precisely focussed for time and activity context.