Unit Bias is Not a Silver Bullet for Nudging Alcohol Consumption
Our recent experimental paper in Journal of Behavioral Economics for Policy [1] explore why reducing the serving size of alcohol may not have much effect on consumption in the wild.
For years, research has suggested that reducing portion sizes can nudge people towards consuming less. The idea, often tied to the concept of unit bias, has been applied successfully to food consumption. For example, smaller plates have been shown to reduce the amount people eat, and in another Insights Post, we demonstrated how reducing the portion size of cake in buffets led to people eating less cake. See our previous Insights Post on nudging towards healthier eating.
Previous studies have suggested that reducing serving sizes could also be an effective way to curb alcohol consumption, with experiments showing that smaller serving sizes lead to lower intake [2-4]. However, these studies were mostly conducted in controlled lab settings or “lab-in-the-field” environments, where participants were aware they were part of an experiment—raising concerns about whether the effects would hold in a natural drinking environment. Having spent considerable parts of our youth in bars, we were somewhat sceptical of these findings and wanted to test whether the effect would replicate in a real-world setting.
A Cleverly Designed Experiment
At iNudgeyou, we pride ourselves on designing rigorous field experiments that test real-world behavioural interventions. This time, we conducted a randomised controlled field experiment at a student bar in Copenhagen to examine whether reducing the serving size of beer would lead to lower overall consumption.
Figur 1. Design of the glasses used in the experiment (filled to 40cl and 50cl respectively).
Here’s how it worked:
• Upon arrival, students were given a cover story and randomly assigned a beer glass—either 50 cl (standard size) or 40 cl (reduced size), see figure 1.
• The glasses were pre-randomised before the event, meaning participants were assigned to their condition without knowing they were part of an experiment.
• To ensure they kept the same glass throughout the night, their names were written on a label attached to it.
• Beer was free all evening, removing any potential pricing effects.
• We tracked total consumption by recording how many times each participant refilled their glass, see figure 2.
Take your Organisation
26 – 28 may, 2025 COPENHAGEN
JOIN OUR NEXT MASTERCLASS
Figur 2. Before glasses were handed to students, their name was asked for and written on a label on the glass. Go-pro cameras placed under the bar-desk out of sight from the students then video-taped names on glasses and the pouring of the beer.
With previous research suggesting that smaller portion sizes typically lead to lower consumption, the expectation was clear: smaller glasses should result in lower beer consumption.
Yet, as is often the case with real-world behaviour, things didn’t go quite as expected.
The Surprising Results
So, what happened?
Contrary to expectations, students with smaller beer glasses drank just as much beer as those with standard-sized glasses, see figure 3.
Figure 3. The average amount of beer in cl consumed per person in each of the two groups. Note: Error bars indicate 95% confidence intervals.
Figure 4. The average number of servings consumed per person in each of the two groups. Note: Error bars indicate 95% confidence intervals.
Instead of drinking less, those with smaller glasses compensated by refilling more often—2.55 refills per person versus 2.03 for those with larger glasses, see figure 4. This suggests that in social drinking settings, people don’t necessarily conform to a predefined portion size. Instead, they adjust their behaviour dynamically to maintain their expected drinking patterns.
Fact box: The Importance of Null Results
It’s tempting to focus only on experiments that produce significant findings, but null results are just as important. They tell us what doesn’t work, helping researchers and policymakers avoid ineffective interventions.
For decades, academic publishing favoured experiments that found statistically significant results, leading to publication bias. This bias created an overrepresentation of positive findings and even incentivised questionable research practices [5-7]. Fortunately, this trend is changing, and journals are increasingly recognising the value of publishing well-conducted experiments, regardless of whether they produce a significant effect.
Our study is a prime example of why null findings matter. Just because an idea makes intuitive sense—like smaller portions leading to lower consumption—doesn’t mean it works in all contexts and for all behaviours. By publishing studies like this one, we help refine our understanding of what truly influences behaviour and ensure that policies and interventions are based on real evidence rather than assumptions.
Implications for Alcohol Policy and Nudging
These findings have important policy implications. In the wake of earlier findings public health strategies have been suggested that propose reducing default serving sizes as a way to curb excessive alcohol consumption. However, our results suggest that in settings where people can freely refill their drinks, reducing serving size may have little to no effect.
This doesn’t mean that nudging can’t help reduce alcohol consumption—it just means that serving size alone isn’t the right lever in this context. Instead, besides the obvious price-setting mechanisms, interventions should focus on:
• Reducing the alcohol percentage: Instead of reducing serving sizes, bars and events could lower the alcohol content in drinks. For example, serving beers that are 10% weaker while maintaining taste and volume could be an effective way to reduce alcohol intake without altering behaviour.
• Introducing alternative activities in drinking environments: Many bar settings focus exclusively on drinking. Including more structured activities—such as pub quizzes, games, or live music—could shift attention away from continuous alcohol consumption. This aligns with findings discussed in one of our favourite books Queueing for Beginners (Moran, 2007) [8], which highlights how structured activities can change consumption patterns.
• Considering earlier closing hours: Research suggests that longer opening hours and later closing times can contribute to increased alcohol consumption. When pubs in the UK reduced structured activities such as billiards and darts but extended opening hours, this unintentionally led to increased drinking. Factors such as fatigue from staying out late, dry throats from shouting over loud music, and the lack of engaging activities created an environment where drinking became the main activity. This suggests that adjusting bar opening hours may be a more effective policy lever than simply reducing serving sizes.
This study highlights the importance of understanding specific behaviours and the contexts in which they unfold if you want to develop effective and generic nudges. This, so called ‘diagnostic approach’ to behaviour change is at the very heart of our BASIC framework; a framework we use in all of our projects and which have served us well in identifying quite effective and simple, but not so ‘magical’ interventions for behaviour change.
Final Thoughts
While reducing serving sizes may work for food, alcohol consumption is driven by more than just portion sizes. Our findings highlight the need for a diagnostic approach to behaviour change — what works for one specific behaviour in certain context does not necessarily translate to what may look on the surface like similar behaviours. When it comes to alcohol consumption, we believe that future interventions should especially consider the social and environmental factors at play, rather than relying on a copy-paste-strategy.
This experiment is just one step in the broader effort to understand the complexities of alcohol consumption behaviour. We encourage policymakers, researchers, and practitioners to continue exploring ways to foster healthier drinking habits.
Want to learn more? You can read our paper here at the Journal of Behavioral Economics for Policy.
Read the references:
[1] Hansen, P. G., Larsen, E. G., Jacobsen, A.-M., Malthesen, M. S., & Schilling, M. (2024). Does reducing the serving size of beer influence how much students drink on a night out? A randomised controlled field experiment. Journal of Behavioral Economics for Policy, 8(1), 7-13.
[2] Pechey, R., Couturier, D., Hollands, G.J., Mantzari, E., Mu naf` o, M.R. and Marteau, T.M. (2016). Does wine glass size influence sales for on-site consumption? A multiple treatment reversal design. BMC Public Health, 16(390) https://doi.org/10.1186/s12889-016-3068-z
[3] Clarke, N., Pechey, R., Pilling, M., Hollands, G. J., Mantzari, E., Marteau, T. M. (2019). Wine glass size and wine sales: Four replication studies in one restaurant and two bars. BMC Research Notes, 12(1), 426. https://doi.org/10.1186/s13104-019-4477-8
[4] Kersbergen, I., Oldham, M., Jones, A., Field, M., Angus, C., & Robinson, E. (2018). Reducing the standard serving size of alcoholic beverages prompts reductions in alcohol consumption: A systematic review and meta-analysis. Addiction, 113(7), 1244–1253. https://doi.org/10.1111/add.14119
[5] Carey, B. (2011, November 2). Fraud case seen as a red flag for psychology research. The New York Times. Dutch psychologist, Stapel, accused of research fraud – The New York Times
[6] Head, M. L., Holman, L., Lanfear, R., Kahn, A. T., & Jennions, M. D. (2015). The extent and consequences of p-hacking in science. PloS Biology, 13(3), e1002106. https://doi.org/10.1371/journal.pbio.1002106
[7] Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22(11), 1359-1366. https://doi.org/10.1177/0956797611417632
[8] Moran, J. (2007). Queuing for beginners: The story of daily life from breakfast to bedtime. Profile Books. ISBN: 9781861978417
WISH TO STAY IN THE LOOP?
Subscribe to join our Insights
Community