This experiment was one of a series of experiments aimed at investigating the role of social identity as an alternative explanation to 'automatic' contagion for of the spread of aggression, since as an account it fails to adequately explain social group boundaries of ‘passive’ social influence. A largely unsuccessful replication of Aggression 1. The shared identity manipulation did not work (means are in the wrong direction, non-sig). There was a significant main effect of crowd noise on shared identity (higher in neutral than aggression condition suggesting that non-normative aggression was detrimental to shared identity). Shared identity significantly moderated the relationship between perceived aggressiveness of crowd noise and explicit aggression. Explicit and implicit aggression do not correlate in any of the 6 conditions (or overall).How and why do behaviours spread from person to person? In particular, how does aggression and violent behaviour spread? When, as in 2011, riots began in London, why did they then occur in Birmingham, Manchester, and Liverpool? One of the most common ways of addressing such issues is through the notion of 'contagion'. The core idea is that, particularly in crowds, mere exposure to the behaviour of others leads observers to behave in the same way. 'Contagion' is now used to explain everything from 'basic' responses such as smiling and yawning (where the mere act of witnessing someone yawn or smile can invoke the same response in another) to complex phenomena like the behaviour of financial markets and, of course, rioting. What is more, laboratory experiments on the 'contagion' of simple responses (such as yawning) serve to underpin the plausibility of 'contagion' accounts as applied to complex phenomena (such as rioting). Despite this widespread acceptance, the 'contagion' account has major problems in explaining the spread of behaviours. In particular, there are boundaries to such spread. If men smile at a sexist joke, will feminists also smile in response to the men's smiles? If people riot in one town, why is it that they also riot in some towns but not others? For example, in 2011, disturbances spread from London to Birmingham, Manchester and Liverpool but they did not spread to Sheffield, Leeds or Glasgow. 'Contagion' explanations cannot answer such questions because they assume that transmission is automatic. They do not take account of the social relations between the transmitter and receiver. We propose a new account of behavioural transmission based on the social identity approach in social psychology. This suggests that influence processes are limited by group boundaries and group content: we are more influenced by ingroup members than by outgroup members, and we are more influenced by that which is consonant with rather than contradictory to group norms. The social identity approach is therefore ideally suited to explaining the social limits to influence, both for 'basic' phenomena and rioting. In order to advance both theoretical understanding and practical interventions, our research will develop a social identity analysis of transmission processes at multiple levels. Accordingly, the aims and objectives of this research project are as follows: First, we will conduct a series of experimental studies on 'basic' behaviours (yawning, itching) to examine whether the effects of being exposed to a behaviour depend on observers and actors being fellow ingroup members. We will also examine 'complex' behaviours (aggression and rioting) to see if (1) observers are more influenced when the actors are ingroup members; (2) observers are more influenced by the responses of other observers when these are also ingroup members; (3) willingness to copy others depends upon whether their behaviour is consonant with observer group norms. Second, we will examine the spread of urban disorder during the 2011 English riots. We have been granted special access to the full data-set from the Guardian/LSE 'Reading the Riots' study (270 interviews with participants carried out immediately following the events). This, along with other secondary sources (such as detailed crime figures), will allow us to examine the extent to which the spread of these riots was linked to a sense of shared identity with those who had rioted previously (that is, those who rioted 'saw themselves' in those who rioted before them, and those who lacked such a sense were less likely to riot). Third, we will use our findings to generate a wider debate about the nature of psychological transmission and the practicalities of addressing them. Activities will include workshops which will bring together researchers, practitioners (e.g., the police) and policy-makers in local and national government to address how we can mitigate against the spread of riots and violence.

A total of 152 Undergraduate students from the University of St Andrews participated in this research. The study was carried out online using Qualtrics. Qualtrics randomly assigned participants to one of three conditions; aggressive crowd noise identified as Dundee students at a demonstration vs non-aggressive crowd noise also identified as Dundee students at a demonstration vs unidentified.Participants were randomly allocated to one of six conditions by Qualtrics (3 x identity, 2 x clip type). Participants were presented with a consent form and information sheet to read and sign and were told that they were being invited to take part in a study investigating differences in ‘differences between different groups’ perceptions of crowd noise (detection and recall) in relation to cognitive style’. At the beginning of the questionnaire, participants answered 3 questions designed to make their identity as students salient. Following this, participants were instructed to put on headphones and listen to the sound clip of ‘Dundee students’. At the end of the sound clip participants then continued on the second section of the questionnaire involving questions to do with the emotions the soundtrack induced, shared identity with the people sound clip, and the explicit aggression measure. They then alerted the researcher and were presented with the IAT test before completing the remaining section of the questionnaire regarding demographics and the final manipulation check.