I commend the article by Weich et al (Reference Welch, Lewis and Jenkins2001) examining the effects of income inequality on mental health. Given the importance of psychosocial factors in Wilkinson's (Reference Wilkinson1996) thesis on inequality and health it is an important and long overdue contribution to this debate. Although this study was cross-sectional in design, the inferences that can be drawn from the findings in support or opposition to Wilkinson's hypothesis are contradictory. The finding that those on high incomes were more likely to report common mental disorders in areas of the greatest income inequality, fits comfortably within Wilkinson's model depicting declining social cohesion, combined with increasing crime and fear. However the second finding, that those on low incomes reported a lower prevalence of common mental disorders in areas of high, rather than low income inequality may appear contrary to this model. However, this finding may help redirect research along an emerging avenue exploring the importance of the appropriate size/scale of comparisons. Recent research has explored this hypothesis at a much smaller scale, examining effects at the level of localities in England and Wales (Reference Boyle, Gatrell and Duke-WilliamsBoyle et al, 1999). Wilkinson's work contains an inherent contradiction on this issue. He argues that the societal level is the most appropriate scale of comparison to explore inequality and health, while basing much of his support for his thesis on studies at a much lower scale.

Weich et al's second finding may reflect income differences in the scale of comparisons. Research exploring the mental maps of individuals, which may influence such comparisons, show dramatic differences by age, gender and particularly by income (Reference LynchLynch, 1960). Although it is dated and overplayed, Stacey's (Reference Stacey1960) comment that “The basis of the upper class social circle is national, with the West End of London as its town centre, The Times as its local paper, and certain national events, e.g. Ascott, as its focal points”, helps demonstrate this point.

Britain is no exception to the international trend of increasing socially homogenous housing estates. Therefore, it is likely that those on the lowest incomes are living in poor, socially segregated areas. They may also be using more local scales of comparison, and in their eyes may in fact be living in areas of low income inequality. Further work is urgently needed to explore the importance of scale in the inequality and health model, paying particular attention to the effects of income, gender and age on such comparisons.