Psychology › Social psychology
Group behavior: in-groups, crowds and bystanders
· Concept
Well replicatedStrong, repeated evidence.
How people act differently in groups: favoring their own side, slacking off, drifting to extremes or failing to help.
See it on the Psychology map →What it says
Groups change individuals. People favor their in-group even when groups are arbitrary. Max Ringelmann, publishing in 1913, found people pulled less hard on a rope as the team grew, now called social loafing. Discussion tends to push members toward a more extreme version of their initial lean, called group polarization. Irving Janis's 'groupthink' (1972) described cohesive teams silencing doubts, as before the Bay of Pigs. Crowds, once seen as mindless mobs, often cooperate in emergencies.
Why it matters
Boards, committees, investment clubs and juries all run on group dynamics; knowing the failure modes helps, for example by gathering independent opinions before discussion.
Think of it like
Like a condo board meeting where no one objects aloud, then everyone grumbles in the parking lot afterward.
The evidence
In-group bias, social loafing and group polarization are well replicated in lab and field studies. Groupthink rests mostly on historical case studies and has weaker experimental support.
Open questions
Whether online echo chambers polarize people as much as feared, and when groups decide more wisely than individuals rather than worse, remain open.
Inside this (2)
Connections (3)
▸In-group bias and social identity (Tajfel) is a basic force in Group behavior: in-groups, crowds and bystandersWell replicated
The connection
In-group favoritism is the starting point for much group behavior: team loyalty, rivalries, office factions and partisan politics. Once people identify with a group, they trust, help and forgive members more readily and judge outsiders more harshly, which can escalate when groups compete for resources.
Who showed it and when
Tajfel's minimal group studies in Bristol, 1970–71; Muzafer Sherif's 1954 Robbers Cave summer camp study in Oklahoma, where rival boys' groups clashed.
The evidence
Minimal group results replicate widely, and a 2014 meta-analysis of economic games confirmed in-group cooperation. Robbers Cave has been criticized for staging.
Open questions
Favoring 'us' does not automatically mean hostility toward 'them'; what tips favoritism into conflict is still studied.
▸The bystander effect is a classic finding about Group behavior: in-groups, crowds and bystandersMixed evidence
The connection
The bystander effect is a classic case of groups changing individual behavior: the more people present, the less each feels personally responsible, so any one person is less likely to act. It parallels social loafing, where individuals work less hard on a team.
Who showed it and when
John Darley at New York University and Bibb Latané at Columbia, 1968; their 1970 book The Unresponsive Bystander.
How solid is it?
A 1981 review of about 50 studies and a 2011 meta-analysis of more than 7,000 participants support the lab effect.
Open questions
Real-world video shows larger groups make help from someone more likely, so the effect may describe individuals rather than whether victims get help.
▸Group behavior: in-groups, crowds and bystanders amplifies Social influence: conformity and obedienceWell replicated
The connection
Groups intensify influence. Discussion pushes like-minded members toward more extreme positions (group polarization), cohesive teams suppress dissent (groupthink), and a unanimous majority exerts far more pressure than a split one. In markets the same dynamics fuel herding, bubbles and panics.
Who showed it and when
James Stoner's 1961 'risky shift' thesis at MIT; Serge Moscovici's 1969 polarization studies in Paris; Irving Janis's Groupthink (1972).
The evidence
Group polarization is well replicated, including in a 1986 meta-analysis; groupthink rests largely on case studies such as the Bay of Pigs.
Open questions
How much online echo chambers polarize people is disputed, as is when group deliberation corrects errors rather than amplifying them.