Group Decisions at the Checkout
Groups have a reputation for deciding badly, and it was earned on tasks that have a right answer. A purchase has not got one, so the strongest evidence we have about deciding together has never been pointed at the till.
Answer to the question
Do group decisions change what you buy?
Probably, though the evidence is thinner than most summaries admit. Tables order a more varied set of dishes than the same people would choose alone, a pair with one self-controlled partner chooses as indulgently as a pair with none, and influence in a couple tracks who prevailed in earlier decisions. The effect on spending reverses depending on whether your companion is a peer or a relative.
Key takeaways
- The pooled hidden-profile evidence covers 65 studies and 3,189 groups, and groups facing one were around eight times less likely to find the solution than groups handed the full picture.
- The second meta-analysis pools 72 studies, 4,795 groups and 17,279 individuals, and information sharing predicted team performance positively at every level of every moderator.
- Seven studies published in 2014 found that a pair with one self-controlled partner chose no less indulgently than a pair with none, so a joint choice is not the average of two individual ones.
- An experiment with 101 real couples, published in 2025, describes the literature on laboratory experiments with real couples as emerging, in its own footnotes.
Group decisions have a reputation for going badly, and it was earned in a laboratory that looks nothing like a shop. The studies that built that reputation used tasks with a right answer: a caucus with one objectively best candidate, a code that either cracks or does not, a work team whose output can be marked. Several thousand groups have now been watched attempting those tasks, often badly, sometimes well, and hardly any of them were buying anything. When two people stand in front of a sofa, or four people choose where to eat, there is no correct option waiting behind the choice to be recovered, and that absence cuts the whole body of work loose from the till.
The experiment that made groups look worse
In 1985 Garold Stasser and William Titus gave 228 introductory psychology students at Miami University a political caucus to run, in groups of four. Each member read a description of the candidates carrying only part of the information, and the part each person held was biased against the candidate who was in fact the strongest. Between them, the four people in a group held everything needed to reconstruct the true picture. That design is a hidden profile, meaning a decision where no individual has enough information to see the best option but the group collectively does.
The groups did not recover it. They settled on the candidate a plurality had already preferred before anyone spoke, and the researchers' before-and-after memory checks showed that discussion had carried the members' distorted pictures forward rather than correcting them. The account the authors offered is a biased sampling model: talk is dominated by what people already share and by what supports the view they walked in with. Four decades of group decisions research has been built on that one experimental move, and the move depends entirely on there being a strongest candidate to find.
Thousands of groups, one findable answer
The pooled version is large. A meta-analysis in 2012 led by Li Lu, which gathers many separate experiments into one statistical body, covered 65 studies, 101 independent effects and 3,189 groups. Groups mentioned the information their members already held in common far more often than the information only one person held. Groups facing a hidden profile were around eight times less likely to find the solution than groups handed the full picture. Two measures of how well a group covered and stayed on the unique information both predicted the quality of its decision, coverage the more strongly of the two.
One of that meta-analysis's results is a useful corrective to the usual commentary. Whether people talked face to face or through a screen made no difference the analysis could detect, to how much unique information came out or to how good the decision was. The things that did matter were group size, how much information was in play, how much of it was unique, how strong the hidden profile was, and whether a correct answer could be demonstrated once proposed.
A second meta-analysis, published in 2009 and led by Jessica Mesmer-Magnus, approached the same ground from team performance rather than from puzzles. It pooled 72 independent studies covering 4,795 groups and 17,279 individuals. Information sharing predicted performance, cohesion, satisfaction with the decision and how well knowledge was integrated, and the relationship stayed positive at every level of every moderator the authors identified. Three things raised sharing, and the first of them was again whether the task had a demonstrable answer.
The literature is also not the simple story of group incompetence it is often retailed as. In experiments led by Patrick Laughlin in 2006, groups of three, four and five people solved letters-to-numbers problems, where ten letters map onto ten digits and the mapping has to be worked out, faster and with more complex reasoning than the best of an equivalent number of individuals working alone. Groups of three were enough; adding a fourth and fifth member bought nothing measurable. The authors were precise about the scope of this, restricting the conclusion to highly intellective problems, which is their term for problems whose answers can be shown to be right.
A review of twenty-five years of experimental games, led by Tamar Kugler in 2012, points the same way. Across prisoner's dilemmas, ultimatum games, trust games and principal-agent games, group decisions came out consistently closer to the rational benchmark that game theory assumes than individual decisions did. It is a narrative review rather than a pooled analysis, so it carries no combined sample, and it is worth holding onto anyway, because it shows that the direction of a group effect depends on what you chose to measure.
What a sofa has not got
Task demonstrability is the name for whether a correct answer can be shown to be correct once somebody proposes it, and both of those meta-analyses list it among the things that decide whether a group gets there. It is also the one thing a purchase cannot supply. Every pooled result on group decisions above is scored against an answer somebody can be shown to have found, and there is no demonstrably right sofa. There is no objectively best holiday against which a couple's choice can be scored, and no trial count, no accuracy rate and no solution rate to put on the other side of the comparison.
| Where the answer can be shown | Where it cannot | |
|---|---|---|
| The task | a caucus with a best candidate, a letters-to-numbers puzzle, a team project | a sofa, a restaurant, a holiday, the weekly shop |
| What is measured | accuracy, trials to solution, decision quality | variety, indulgence, regret, spend, who prevailed |
| Pooled evidence | 3,189 groups on hidden profiles, 4,795 groups on information sharing | no meta-analysis of joint against individual purchases could be located |
| Strongest moderator | task demonstrability | nothing for it to attach to |
A purchase has no demonstrably correct option, so there is nothing for the organisational findings to attach to.
The framing most people meet this question through comes from organisational psychology. Colin M. Fisher, who teaches at the UCL School of Management, traces it in chapter two of The Collective Edge (2025) back to rope-pulling experiments and then to Ivan Steiner's subtraction formula, where what a group actually produces is its potential minus the losses from faulty process.
Richard Hackman later added process gains, the ones that can push a group above the sum of its parts, and Fisher's conclusion from the pair of them is that synergy is real but rare, with task complexity deciding whether a group beats its members.
His book is about task groups and says so. Every case that carries weight in it is a work team, a research collective, an orchestra, a field-service crew. The vault's full record of the book contains no consumer purchase, no household spending decision, no shopper and no retail setting, which is not a criticism of the argument but a statement of where it was aimed. The overreach would be in borrowing it.
> A purchase has no demonstrably correct option, so there is nothing for the organisational findings to attach to.
So here is the gap, stated as plainly as it can be. Even where the paradigm has been pointed at a domestic-sounding choice, a pair picking a flat, the design installs a demonstrably best option for them to find. An ordinary purchase will not allow that move, which is why the strongest and best-replicated evidence about group decisions does not reach the decision this article is about.
Tables order what the people would not
What does exist on the consumer side measures something else entirely. In 2000, Dan Ariely and Jonathan Levav tracked real orders of food and drink in three studies. Tables chose a more varied set of dishes than you would predict by drawing the same number of individual choices at random from everyone who ate there. In the beer study, that group-level variety came from the interaction between the people at the table, implicit or explicit, and it dissipated when the group was broken up and members chose strictly on their own.
House practice is that no single paper carries a finding on its own, and that applies here. The one published independent re-examination, by two authors, this pass could locate came in 2010, led by Pascale G. Quester, and it narrows rather than overturns. Quester and Andre Steyer's two studies find the relationship between an individual's choice and the group's opinion to be non-monotonic, meaning it does not simply rise or fall, and moderated by how close the group is to unanimous about an option. Their conclusion is that the earlier result describes specific sections of a curve rather than the curve.
The mechanism candidate comes from a line of work independent of either. Three experiments published in 2002 and led by Rebecca K. Ratner found that people build more variety into a consumption decision when it is made where other people can see it, because they expect to be judged more favourably for choosing variety than for repeating a favourite. The pressure lifted in a third study when a social cue signalled that sticking to a favourite was acceptable, and how much a person monitored their own self-presentation changed the size of the effect. A wine list in company is a public document in a way a kitchen cupboard is not.
Two people are not an average
The most useful verified finding about group decisions in a shop is that a joint choice is not the mean of the individual ones. Seven studies published in 2014, led by Hristina Dzhogleva, who later published as Dzhogleva Nikolova, put pairs through joint spending and diet decisions. Pairs where both people had high self-control chose less indulgently than pairs where both had low self-control. Mixed pairs, with one restrained person and one less restrained, were no different from the pairs with no restrained person in them at all. One careful partner bought no more restraint than none.
The authors' tested explanation is the part worth taking into a shop. The more restrained person deferred, prioritising the relationship over their own preference for restraint, and interventions that changed how much a person wanted to keep the peace changed the pattern. That is not process loss in the organisational sense, because nobody got the answer wrong. It is a joint decision producing a result neither person would have produced alone, which is a different and less tidy thing to say.
Influence inside a couple also turns out to have a memory. Experiments reported in 1987 by Kim P. Corfman and Donald R. Lehmann had couples make sequences of product choices, and tested what actually drove whose preference prevailed. The strongest contributions came from how intensely each person wanted the thing and from the outcomes of the preceding joint decisions.
A model of spousal interaction published in 2003 and led by Chenting Su points the same way across time, with spouses revising their influence tactics in later decisions rather than repaying coercion within one, although the paper's declared purpose is to help marketers aim messages at the more influential spouse.
The findings that did not move
On risk, the honest answer is that nothing much happens. A field experiment published in 2022 and led by Maria Bernedo Del Carpio elicited preferences from individuals, from married couples and from unrelated pairs. The risk preferences of both kinds of pair came out similar to those of the individuals who made them up, with no difference the experiment could detect. A separate elicitation published in 2013 and led by Mohammed Abdellaoui found couples' risk attitudes consistent with a mix of the two individual attitudes, with women more influential at low probabilities.
On patience the published record is thinner than the retelling suggests, and part of the gap is an artefact of what got into print. The 2022 field experiment found the joint decisions of both married couples and unrelated pairs more impatient than the decisions of the individuals in them, by a similar margin for both kinds of pair, which points at a group effect rather than anything about marriage. The widely repeated claim that couples discount the future less sits in a working paper from 2010 whose published version dropped the result, so it is not in the published record at all.
The newest work is the most careful about how early this all is. An experiment with 101 real couples, published in 2025 and led by Jiakun Zheng, varied household and individual income risks in a controlled task. About 79 per cent of couples pooled risk at the household level when the risk fell on both symmetrically, while about 36 per cent failed to pool when individual risks fell asymmetrically. The authors describe the literature on laboratory experiments with real couples as emerging, in their own footnotes, and record the evidence on whether spouses' risk preferences correlate as mixed.
Even the simplest question, whether shopping with others makes you buy more, splits in two. Two experiments published in 2005 by Xueming Luo found that the presence of peers raised the urge to purchase while the presence of family members lowered it, with both effects larger in cohesive groups and in people more susceptible to social influence. Unobtrusive observation of shoppers reported in 1992 found that groups spent longer in the store and left with larger loads, with the larger loads holding up once shopping time was taken into account.
What this evidence cannot carry
Four limits should sit beside everything above, and they are the reason this piece declines to give a verdict on whether group decisions buy better or worse than individual ones.
1. It is old. The consumer-domain studies here run from 1987 to 2014, and the work published since then is economics of the household rather than consumer psychology.
2. It is largely unreported here. Sample sizes were legitimately reachable for only a handful of these papers, and where a full text was paywalled this article prints no number rather than guessing one.
3. The settings have changed. Tables ordering beer, in-store observation of grocery loads and couples ranking products on paper all predate the phone in the hand, the shared basket, the group chat and delivery to the door.
4. The outcomes do not line up. Accuracy and solution rates on one side, variety and indulgence and regret and who prevailed on the other, are not the same dependent variable, and no amount of careful writing makes them one.
One more thing is unverified and deserves saying, because it is usually asserted without a source. There is no current representative figure for how many large purchases are made by more than one person. The nearest published material is a survey from 2008 in which about half of couples volunteered that they decide together on buying big things for the home, and self-reports of who decides are known to diverge between the two people giving them.
So what would settle it. A hidden profile needs a demonstrably right answer, and the thing that makes a purchase interesting is that it has not got one. That leaves two possibilities worth arguing over: either accuracy is the wrong outcome to measure, or group decisions in a shop are a separate phenomenon sharing a name.
Common questions
Do group decisions change what you buy?
Probably, though the evidence is thinner than most summaries admit. Tables order a more varied set of dishes than the same people would choose alone, pairs with one self-controlled partner choose as indulgently as pairs with none, and influence in a couple tracks who prevailed in earlier decisions. Direction of effect on spending reverses depending on whether your companion is a peer or a relative.
Why can't research on work teams explain a purchase?
Because nearly every result in it needs a demonstrably correct answer. Hidden-profile experiments score groups on whether they found the best candidate, and task demonstrability is among the conditions both large meta-analyses of that work find decisive. A sofa has no correct option, so there is no accuracy to measure and nothing for those findings to attach to.
Do couples take more risk together than alone?
There is no difference the experiments could detect. One field study found the risk preferences of married couples and of unrelated pairs similar to those of the individuals in them, and a separate elicitation found couples' attitudes consistent with a mix of the two individual attitudes. On patience the one published field result points towards pairs being more impatient, and the familiar claim that couples discount the future less sits in a working paper whose published version dropped it.
Does shopping with other people make you spend more?
Not as a general effect. Two experiments from 2005 found peers raised the urge to purchase and family members lowered it, so the companion's identity decides the direction. Observation from 1992 found group shoppers stayed longer and left with larger loads, but it is a single correlational study predating online baskets and group chats entirely.
Sources
Stasser, G., & Titus, W. (1985). 'Pooling of Unshared Information in Group Decision Making: Biased Information Sampling During Discussion.' Journal of Personality and Social Psychology, 48(6), 1467-1478. doi:10.1037/0022-3514.48.6.1467. [Full text read, from the copy hosted by the University of Muenster psychology faculty]: four-person groups of Miami University introductory psychology students fulfilling a course research requirement, randomly assigned, 72 subjects in each of the shared and unshared/conflict conditions and 84 in the unshared/consensus condition, 228 in total. Political caucus simulation; members individually read candidate descriptions carrying partial information biased against the most favourable candidate, then discussed as a group, the group collectively holding everything needed to reconstruct the true profiles. Groups chose the candidate a plurality had preferred before discussion rather than the objectively most favourable one, and pre- and post-discussion recall showed discussion perpetuating rather than correcting members' distorted pictures. Authors' stated scope: a biased sampling model in which discussion is dominated by information members already share and by information supporting their existing preferences. Author order confirmed at the Crossref item record: Stasser first, Titus second.
Lu, L., Yuan, Y. C., & McLeod, P. L. (2012). 'Twenty-Five Years of Hidden Profiles in Group Decision Making: A Meta-Analysis.' Personality and Social Psychology Review, 16(1), 54-75. doi:10.1177/1088868311417243. [Abstract read in full at journals.sagepub.com; full text paywalled and not read]: 65 studies, 101 independent effects, 3,189 groups. Groups mentioned two standard deviations more pieces of common information than unique information; hidden-profile groups were eight times less likely to find the solution than groups given full information; information coverage and discussion focus both predicted decision quality positively, coverage more strongly; communication medium affected neither unique-information pooling nor decision quality. Stated moderators: group size, total information load, proportion of unique information, task demonstrability, hidden-profile strength. Author order confirmed at the Crossref item record: Lu first, Yuan second, McLeod third.
Mesmer-Magnus, J. R., & DeChurch, L. A. (2009). 'Information sharing and team performance: A meta-analysis.' Journal of Applied Psychology, 94(2), 535-546. doi:10.1037/a0013773. [Abstract read in full via the OpenAlex item record; full text paywalled and not read]: 72 independent studies, 4,795 groups, 17,279 individuals. Information sharing predicted team performance, cohesion, decision satisfaction and knowledge integration, and the relationship held positively across all levels of every moderator identified. Raised sharing: task demonstrability, discussion structure, cooperation. Reduced it: information distribution, informational interdependence, member heterogeneity. Stated scope: teams, and information sharing as a team process. Author order at the item record: Mesmer-Magnus first, DeChurch second.
Laughlin, P. R., Hatch, E. C., Silver, J. S., & Boh, L. (2006). 'Groups perform better than the best individuals on letters-to-numbers problems: Effects of group size.' Journal of Personality and Social Psychology, 90(4), 644-651. doi:10.1037/0022-3514.90.4.644. [Abstract read in full via the OpenAlex item record; full text paywalled and not read]: individuals and groups of two, three, four and five solved two letters-to-numbers problems, identifying the mapping of ten letters to ten numbers by proposing equations and hypotheses and receiving feedback. Groups of three, four and five proposed more complex equations and needed fewer trials to solution than the best of an equivalent number of individuals; groups of three, four and five did not differ from each other and all beat two-person groups. The authors' own stated conclusion is the scope limit this article needs: three-person groups are necessary and sufficient to outperform the best individuals on highly intellective problems, meaning problems with demonstrably correct answers. Author order at the item record: Laughlin first.
Kugler, T., Kausel, E. E., & Kocher, M. G. (2012). 'Are groups more rational than individuals? A review of interactive decision making in groups.' Wiley Interdisciplinary Reviews: Cognitive Science, 3(4), 471-482. doi:10.1002/wcs.1184. [Abstract read in full via the OpenAlex item record; full text paywalled and not read]: a narrative review, not a meta-analysis, of twenty-five years of experiments comparing groups and individuals in prisoner's dilemma, dictator, ultimatum, trust, centipede and principal-agent games. The review's conclusion is that results are consistent in showing group decisions closer to the game-theoretic rationality assumption than individual decisions. No pooled sample, and it must be labelled a review. Stated scope: interactive strategic decisions with monetary payoffs. Used here as a counterweight: groups are not uniformly worse, and the direction of the group effect depends entirely on what is measured. Author order at the item record: Kugler first.
Ariely, D., & Levav, J. (2000). 'Sequential Choice in Group Settings: Taking the Road Less Traveled and Less Enjoyed.' Journal of Consumer Research, 27(3), 279-290. doi:10.1086/317585. [Abstract read in full via the OpenAlex item record; full text paywalled and NOT read, so no sample size is printed. A legitimate full text was sought at Duke Scholars, the MIT eRationality page, Columbia and RePEc and none was served]: three studies tracking real orders of dishes and drinks. The Lunch study found real groups, meaning tables, chose more varied dishes than would be expected from random sampling of the population of all individual choices across all tables. The Beer study showed this group-level variety seeking is attributable to the interaction, implicit or explicit, among members, and that it dissipates when the group is forced to disband and members choose strictly individually. The Wine study showed individual choices in a group context also serve information gathering and self-presentation as uniqueness. Authors' stated framing: choices balance strictly individual goals against goals triggered by the existence of the group, and the latter sometimes produce choices that undermine personal satisfaction and increase regret. Author order confirmed at the Crossref item record: Ariely first, Levav second. HANDLING RULE, not optional: the vault's note at People/Dan Ariely.md requires flagging the data-integrity problems in parts of Ariely's body of work before any citation and checking the specific paper. This pass found no retraction, correction or expression of concern attached to the 2000 paper at Crossref or OpenAlex, which is not the same as a clean bill of health and is not presented as one. Because this is the most load-bearing consumer source in the file, the article must name Jonathan Levav in the same breath, must state that one published independent re-examination exists, and must not rest any sentence on this paper alone.
Quester, P. G., & Steyer, A. (2010). 'Revisiting Individual Choices in Group Settings: The Long and Winding (Less Traveled) Road?' Journal of Consumer Research, 36(6), 1050-1057. doi:10.1086/644750. [Abstract read in full via the OpenAlex item record; full text paywalled and not read. Crossref gives published print and online April 2010; some listings date it 2009 and the vault prints 2010]: two empirical studies revisiting Ariely and Levav. The authors examine how group opinion and unanimity explain individual choice in a group setting, hypothesise that the relationship between individual choice and group opinion is non-monotonic and moderated by the degree of unanimity around an alternative, and conclude that choice patterns are curvilinear and that the previous findings accurately reflect only specific sections of the overall pattern. Stated scope: ordering food or beverages in a group. Author order confirmed at the Crossref item record: Quester first, Steyer second. This is the only published re-examination of the table-ordering finding this pass could locate, and it narrows rather than overturns it.
Ratner, R. K., & Kahn, B. E. (2002). 'The Impact of Private versus Public Consumption on Variety-Seeking Behavior.' Journal of Consumer Research, 29(2), 246-257. doi:10.1086/341574. [Abstract read in full via the OpenAlex item record; full text paywalled and not read, so the three studies' sample sizes are not printed]: three experiments. People built more variety into consumption decisions when their behaviour was subject to public scrutiny. Studies 1 and 2: consumers expect others to evaluate their decision more favourably if they choose variety, and this sometimes leads them to put more variety into public than private decisions, with self-monitoring moderating the effect. Study 3: the pressure to choose variety in public is eliminated when a social cue signals that consuming one's favourites is appropriate. Stated scope: variety seeking under expected evaluation. Author order at the item record: Ratner first, Kahn second. This is the mechanism candidate for the table-ordering result and an independent line of evidence, which matters given the handling rule on Ariely and Levav.
Dzhogleva Nikolova, H., & Lamberton, C. P. (2014). 'Should Birds of a Feather Flock Together? Understanding Self-Control Decisions in Dyads.' Journal of Consumer Research, 41(2), 361-380. doi:10.1086/676599. [Abstract read in full via the OpenAlex item record; full text paywalled and not read, so the seven studies' sample sizes are not printed]: seven studies on joint spending and diet decisions in pairs. Homogeneous high self-control pairs made less indulgent choices than both homogeneous low self-control pairs and mixed pairs, and the latter two did not differ, so having one high self-control partner produced no more restraint than having none. The authors' tested explanation: higher self-control individuals prioritise pro-relationship behaviour over their own preference for restraint and assent to the lower self-control partner's preferences, and interventions that change pro-relationship motivation alter the pattern. Stated scope: decisions consumers make in couples or pairs, with implications claimed for consumers, marketers and policy. Author order at the item record: Dzhogleva Nikolova first, Lamberton second. The best consumer-side demonstration in the file that a joint decision is not the average of two individual ones.
Corfman, K. P., & Lehmann, D. R. (1987). 'Models of Cooperative Group Decision-Making and Relative Influence: An Experimental Investigation of Family Purchase Decisions.' Journal of Consumer Research, 14(1), 1-13. doi:10.1086/209088. [Abstract read in full via the OpenAlex item record; full text paywalled and not read, so the number of couples is not printed]: a conceptual framework for group decision-making in conflict, tested with experimental data from couples making sequences of product choice decisions. Models tested power-related resources and power-use-related goals as determinants of relative influence. Result: relative preference intensity and the outcomes of preceding joint decisions consistently made the strongest contributions to relative influence. Stated scope: family purchase tasks. Author order confirmed at the Crossref item record: Corfman first, Lehmann second. The second half of that result is the most useful sentence in the consumer file: influence in a couple is partly a running tally of who won last time, which no single-decision study can see.
FURTHER VERIFIED, SUPPORTING RATHER THAN LOAD-BEARING. Su, C., Fern, E. F., & Ye, K. (2003). 'A Temporal Dynamic Model of Spousal Family Purchase-Decision Behavior.' Journal of Marketing Research, 40(3), 268-281. doi:10.1509/jmkr.40.3.268.19234. [Abstract read in full via the OpenAlex item record; full text not read; sample not established]: a dynamic simultaneous-equations model of spousal interaction within a single purchase decision and across decisions. Spouses tended not to reciprocate coercion within a discrete decision and adjusted influence strategies over time; the paper also models the effectiveness of influence strategies and spousal satisfaction as the mechanism by which spouses revise behaviour across later decisions. Stated scope is explicitly marketing-communication targeting, which is a limit worth printing: the paper's declared purpose is to help marketers aim messages at the influential spouse. Author order at the item record: Su first, Fern second, Ye third.
FURTHER VERIFIED. Sommer, R., Wynes, M., & Brinkley, G. (1992). 'Social Facilitation Effects in Shopping Behavior.' Environment and Behavior, 24(3), 285-297. doi:10.1177/0013916592243001. [Abstract read in full at the Crossref item record, which carries it in full; full text not read; the number of shoppers observed is not established]: unobtrusive observation comparing time in store and load size for lone against group shoppers. Groups spent longer in the store and bought larger loads, and the larger loads held up with shopping time controlled. Women and older people spent longer and bought more than men and younger people. Stated frame: social facilitation. The only direct observational evidence in the file that company changes what leaves the shop; it is from 1992 and it is correlational, and the article says both. Author order at the Crossref item record: Sommer first, Wynes second, Brinkley third.
FURTHER VERIFIED. Luo, X. (2005). 'How Does Shopping With Others Influence Impulsive Purchasing?' Journal of Consumer Psychology, 15(4), 288-294. doi:10.1207/s15327663jcp1504_3. [Abstract read in full via the OpenAlex item record; full text not read; the green open-access copy at papers.ssrn.com served a bot challenge and was not read, so sample sizes are not printed]: two experiments. The presence of peers increased the urge to purchase; the presence of family members decreased it. Both effects were larger when the group was cohesive and when participants were susceptible to social influence. Stated scope: impulsive purchasing, measured as urge to purchase, against the field's individual-level assumption. Single-authored. The direction reverses with the identity of the companion, which is the finding, and it is the cleanest refutation available of any claim that shopping with people has one effect.
FURTHER VERIFIED. Bernedo Del Carpio, M., Alpizar, F., & Ferraro, P. J. (2022). 'Time and risk preferences of individuals, married couples and unrelated pairs.' Journal of Behavioral and Experimental Economics, 97, 101794. doi:10.1016/j.socec.2021.101794. [Abstract read in full at the Wageningen University research portal, research.wur.nl; full text paywalled, sciencedirect.com refused the browser twice, so the sample is not established. Crossref gives the print issue as April 2022; some listings give 2021]: an incentivised field experiment eliciting preferences from individuals, married couples and unrelated pairs, the contrast between married couples and unrelated pairs separating a marriage effect from a plain group effect. Joint decisions of both married couples and unrelated pairs showed more impatience than the constituent individuals' decisions, and the difference was similar for both kinds of pair, implying a group effect on time preference rather than anything about marriage. Risk preferences of both kinds of pair were similar to those of their constituent individuals, which is a null. Author order confirmed at the Crossref item record: Bernedo Del Carpio first, Alpizar second, Ferraro third.
FURTHER VERIFIED. Abdellaoui, M., l'Haridon, O., & Paraschiv, C. (2013). 'Individual vs. couple behavior: an experimental investigation of risk preferences.' Theory and Decision, 75(2), 175-191. doi:10.1007/s11238-012-9322-7. [Abstract read in full at link.springer.com; full text paywalled and not read; sample not established]: elicits individuals' and couples' preferences under prospect theory using certainty equivalents. Couples' risk attitudes were compatible with prospect theory and showed deviations from expected utility similar to those found in individual decision making, and couples' attitudes were consistent with a mix of the two individual attitudes, with women more influential at low probability levels. Author order confirmed at the Crossref item record: Abdellaoui first, l'Haridon second, Paraschiv third. CORRECTION THE ARTICLE MUST CARRY: the circulating claim that couples discount the future less, meaning joint decisions reduce impatience, appears in the authors' 2010 and 2011 HEC Paris working paper, cahier de recherche number 944, and is NOT in the published 2013 article, whose abstract covers risk only. The published record therefore does not contain that time-preference result, and it points the opposite way to Bernedo Del Carpio and colleagues.
FURTHER VERIFIED. Zheng, J., Couprie, H., & Hopfensitz, A. (2025). 'Collective risk-taking by couples: individual vs household risk.' Theory and Decision, 98(4), 599-628. doi:10.1007/s11238-024-10021-z. [Abstract read in full at link.springer.com together with the article's first several footnotes; full text paywalled and not read]: 101 real couples in a controlled experimental risk-taking task varying household and individual income risks while controlling ex-ante income inequality. About 79 per cent of couples pooled risk at the household level when risks were borne symmetrically; about 36 per cent failed to pool when individual risks were borne asymmetrically. Intra-household risk inequality mattered more for unmarried than married couples, and the results held when restricted to couples where both spouses were risk-averse. Two things from the footnotes are quotable as the state of the field: the authors state that the literature on laboratory experiments with real couples is emerging, and they record that the evidence on whether spouses' risk preferences correlate is mixed, finding no correlation themselves where an earlier study found a positive one. Published January 2025, which also answers the charge that all of this work is old. Author order confirmed at the Crossref item record: Zheng first, Couprie second, Hopfensitz third.
FURTHER VERIFIED, WITH A SEVERE SCOPE LIMIT. D'Exelle, B., & Ringdal, C. (2025). 'Who decides in the household when risk preferences conflict?' Review of Economics of the Household, 24(2), 499-540. doi:10.1007/s11150-025-09771-8. [Abstract read in full at the Crossref item record, which carries it in full; full text not read. Crossref gives the publication date as 10 April 2025 and places the item in volume 24, issue 2, a later volume year than the publication date; the year printed is the Crossref publication year and the issue year should be re-checked at the publisher before citation]: survey data from both spouses of 675 couples in rural Tanzania. Spousal differences in risk preferences were associated with a lower likelihood that spouses decided jointly and a higher likelihood that the husband decided alone, and the relations were stronger in couples where women were more conflict-averse. The authors' stated frame is gender inequality in developing countries and women's access to resources. SCOPE CAUTION: rural Tanzanian survey data about household decisions in general, correlational, not about retail purchasing in a high-income market. It is in the file as evidence that who decides is itself an outcome of disagreement, and for nothing else. Author order confirmed at the Crossref item record: D'Exelle first, Ringdal second.