The effect of wine glass size on volume of wine sold: a mega-analysis of studies in bars and restaurants

Aim To estimate the effects of wine glass size on volume of wine sold in bars and restaurants. Design A mega-analysis combining raw (as opposed to aggregate-level) data from eight studies conducted in five establishments. A multiple treatment reversal design was used for each data set, with wine glass size changed fortnightly while serving sizes were unaffected, in studies lasting between 14 and 26 weeks. Setting and participants Five bars and restaurants in England participated in studies between 2015 and 2018, using wine glasses of five sizes: 250, 300, 370, 450 and 510 ml, with the largest size only used in bars. Measurements Daily volume of wine sold by the glass, bottle or carafe for non-sparkling wine were recorded at bars (594 days) and restaurants (427 days), averaging 4 months per study. Findings Mega-analysis combining data from bars did not find a significant effect of glass size on volume of wine sold compared with 300-ml glasses: the volume of wine sold using 370-ml glasses was 0.5% lower [95% confidence interval (CI) = -8.1% to 6.1%], using 450-ml glasses was 1.0% higher (95% CI = -9.1 to 12.2) and using 510-ml glasses was 0.4% lower (95% CI = -9.4 to 9.4). For restaurants, compared with 300-ml glasses, the volume of wine sold using 250-ml glasses did not show a significant difference: 9.6% lower (95% CI = -19.0 to 0.7). Using 370-ml glasses the volume of wine sold was 7.3% higher (95% CI = 1.5% to 13.5%); no significant effect was found using 450-ml glasses: 0.9% higher (95% CI = -5.5 to 7.7). Conclusions The volume of wine sold in restaurants in England may be greater when 370-ml glasses are used compared with 300-ml wine glasses, but may not be in bars. This might be related to restaurants compared with bars selling more wine in bottles and carafes, which require free-pouring.