Design
The study used a between-subjects design, with participants randomised to one of two groups to receive 175 ml of wine served in one of two wine glass sizes: (a) smaller (250 ml), (b) larger (370 ml).
Participants
Participants comprised of 166 female students (age M = 22.93; SD = 3.52, range 18–42) who drank red wine, were at least 18 years of age, were not currently pregnant or taking any medication that interacts with alcohol, and who had not consumed alcohol in the 12 h prior to the study. The study included only women to minimise gender differences in average sip duration [10]. The study was powered to test the first hypothesis assessing drinking rate, based on effect sizes from a previous study [5]. Power analysis indicated that 160 participants were needed to detect a medium sized effect (d = 0.5) in a two-tailed test with α = 0.05 and power of 0.85.
Materials and measures
Wine glasses
The larger wine glass was 370 ml in volume and the smaller wine glass was 250 ml in volume. The wine glasses were Royal Leerdam Fortius glasses differing only in their capacity. They were the same as those used in a previous field study documenting higher sales when wine was served in the larger of the two glasses, compared to a 300 ml glass of the same design [4]. The glasses used in the study are shown in Fig. 1.
Measures
Micro-drinking behaviours
The experimental sessions were recorded using a Raspberry-Pi camera module. The video recordings were coded using a custom-written program in Python (v.2.7), with a researcher pressing a button when the wine touched participants’ lips – indicating sip initiation, and pressing the button again when the wine left participants’ lips – indicating sip end. A second coder, blind to the study hypotheses, independently coded 20% of the videos selected at random to assess coding reliability (presented in the Results section). Variables derived from the video recordings included total time taken to consume the wine, number of sips, and average sip duration.
Satisfaction with perceived amount of wine
This was assessed in two parts: firstly, by exploring perceptions of the amount of wine served, and secondly, examining participants’ satisfaction with the perceived amount of wine.
Perceived amount of wine
Two questions, rated on a seven-point rating scale ranging from 1 (Much less) to 7 (Much more) asked: “How does the amount of wine in the glass you just drank compare to a typical glass of wine you would drink at home?” and “How does the amount of wine in the glass you just drank compare to a typical glass of wine you would drink at a pub or restaurant?”. The baseline of the typical wine portions participants consumed was established by the following question: “What size would your typical glass in a pub or restaurant be?”. Participants could answer by indicating small (=1), medium (=2), or large (=3).
Satisfaction with perceived amount of wine
Five attributes of the given amount of wine (Plentiful, Generous, Inadequate, Unsatisfactory, Disappointing) were each rated using seven-point scales, ranging from 1 (Strongly agree) to 7 (Strongly disagree). The latter three attributes were reverse-coded prior to the analysis. A composite score (‘Satisfaction’) combining these attributes was formed (Cronbach’s α = 0.71).
Pleasure
The pleasure of the drinking experience was assessed by rating the experience of drinking the wine on five attributes using seven-point rating scales: Pleasurable, Enjoyable, Disagreeable, Unpleasant, Distasteful. The latter three dimensions were reverse-coded prior to the analysis. A composite score was developed to reflect this variable (Cronbach’s α = 0.95).
Desire to drink more
Desire to drink more was assessed by indicating agreement on a seven point scale ranging from 1 (Strongly agree) to 7 (Strongly disagree) for the following statements: “I wish I had another glass of wine right now”, “I don’t want any more wine right now”, “If I were offered another glass of wine right now, I would drink it”, “If I were in a pub or bar, I would buy another glass of wine right now”, “ If I had the chance, I would not have any more wine right now”. Appropriate items were reverse-coded prior to the analysis. Scale reliability assessed by Cronbach’s α was 0.78.
Perceived intoxication
Perceived intoxication was assessed with a single item: “I feel drunk at the moment”. Participants responded by indicating agreement with the statement on a scale ranging from 1 (Strongly agree) to 7 (Strongly disagree).
Alcohol use
The Alcohol Use Disorders Identification Test (AUDIT [19]), a 10-item measure, was used to assess the quantity and frequency of alcohol use and harmful drinking behaviour. Scores of 0–7 are considered low-risk, scores of 8–14 are considered hazardous and scores of 15 or over are considered harmful.
Subjective craving
The Alcohol Urge Questionnaire (AUQ [20]), an 8-item measure, was used to assess current craving for alcohol. The AUQ scores were used as a baseline measure, to ensure that any differences in the outcome measures were not due to urges to consume alcohol.
Filler task
The nature of the study was disguised by administering a computerised version of the Trail Making Test as a filler task [21]. The Trail Making Test assesses cognitive processing, visual attention and executive functioning [22,23,24]. It consists of connecting 25 circles distributed over a computer screen according to set rules (sequentially connecting numbers or alternating between letters and numbers). The dependent variable is the total time required to complete the task. There were two practice trials and two search trials. The results of the filler task were not analysed.
Procedure
Participants were recruited from the University of Cambridge and from Anglia Ruskin University via mailing lists, poster advertisements and word-of-mouth. They received £10 payment for participation. Participants completed the sessions individually in a quiet laboratory between 12:00 and 21:00 h during weekdays. To avoid participants’ consumption being influenced by awareness of the study hypotheses, the study was presented to them as investigating the effects of limited amounts of alcohol on cognitive performance. After giving consent, eligibility to participate in the study was confirmed by a breathalyser check, to ensure that participants had refrained from consuming alcohol in the preceding 12 h. Participants then completed the AUDIT and AUQ measures. Participants were randomised to either the smaller (250 ml) or larger (370 ml) wine glass condition using a computer generated randomisation schedule with the constraint of having an equal number of participants per group. A portion of 175 ml Cuvée des Vignons Beaujolais red wine (12% alcohol by volume) was measured out by filling a 175 ml pub thimble to the brim, and then poured into the wine glass allocated by randomisation, immediately prior to each experimental session. Wine bottles were secured with a vacuum-pump to minimise oxidation of the wine between experimental sessions. When participants had completed the baseline measures, the experimenter switched on the hidden camera from a remote laptop, and returned to the lab with a glass of red wine. Participants were told to drink the wine at their own pace while watching a nature documentary (“The Story of Earth” National Geographic 2011). The experimenter then left the room and returned when the participants had indicated by ringing a bell that they had finished their glass of wine. If participants did not finish drinking the wine after 30 min, the experimenter returned to ask if everything was alright. If participants were still drinking, the experimenter left and returned to end the session either when the participant indicated that they had the wine finished or after an additional 15 min, whichever was sooner. After the drinking session, participants were given the questionnaires to complete, followed by the filler task. Finally, participants were asked what they thought the aim of the study was. Participants were blind to the study aims and were fully debriefed about the purpose of the study via email at the end of the study, i.e., when the last participant had completed the study.
Data analysis
Preliminary analyses included examining differences between groups between-group mean differences for effects of glass size on outcome variables using non-parametric bootstraps in R: (i) Total drinking time (ii) Satisfaction, and (iii) Pleasure, and (iv) Desire to drink more, as well as other aspects of drinking behaviour that may contribute to drinking time, i.e.: a. number of sips and b. sip duration. We also tested between-group mean differences for effects of glass size on perceived intoxication as a proxy for how much people believe they had consumed. Regression analyses were conducted to test for “proof of concept”; namely that desire to drink more is predicted by the measures listed in (i) to (iii) above.
Sensitivity analyses were conducted by removing any participants who had not drunk all the wine they had been served or who indicated they were aware of the study aims.