Weighting for non-response was introduced on the GHS in 2000, and was described in detail in the GHS 2000 report. The effect of weighting on the smoking data is slight: it increases the overall prevalence of cigarette smoking by one percentage point. In 1998 and 2000, the upward revision was due solely to a change of two percentage points among men: there was no difference in the weighted and unweighted prevalence rates among women. In 2001, however, weighting increases prevalence for both men and women by one percentage point. The change occurs because weighting reduces the contribution to the overall figure of those aged 60 and over, among whom prevalence is relatively low.