ArticlePDF Available


Half a century of research on bystander behavior concludes that individuals are less likely to intervene during an emergency when in the presence of others than when alone. By contrast, little is known regarding the aggregated likelihood that at least someone present at an emergency will do something to help. The importance of establishing this aggregated intervention baseline is not only of scholarly interest, but is also the most pressing question for actual public victims—will I receive help if needed? The current paper describes the largest systematic study of real-life bystander intervention in actual public conflicts captured by surveillance cameras. Using a unique cross-national video dataset from the United Kingdom, Netherlands, and South Africa (N = 219), we show that in nine-out-of-ten public conflicts, at least one bystander, but typically several, will do something to help. We record similar likelihoods of intervention across the three national contexts, which differ greatly in levels of recorded violent crime. Finally, we find that increased bystander presence is related to a greater likelihood that someone will intervene. Taken together these findings allay the widespread fear that bystanders rarely intervene to help. We argue that it is time for psychology to change the narrative away from an absence of help and towards a new understanding of what makes intervention successful or unsuccessful.
Would I be Helped? Cross-National CCTV Footage Shows That Intervention Is the Norm in
Public Conflicts
Richard Philpot
Lancaster University
University of Copenhagen
Lasse S. Liebst
University of Copenhagen
Mark Levine
Lancaster University
University of Exeter
Wim Bernasco
Netherlands Institute for the Study of Crime and Law Enforcement (NSCR)
Vrije Universiteit Amsterdam
Marie R. Lindegaard
Netherlands Institute for the Study of Crime and Law Enforcement (NSCR)
University of Copenhagen
Author Note
Replication data, statistical scripts, and video coding procedures are made available
on the Open Science Framework -
Correspondence concerning this article should be addressed to Richard Philpot,
Department of Psychology, Lancaster University, Lancaster, LA1 4YF, United Kingdom
Accepted manuscript (published in American Psychologist)*
Half a century of research on bystander behavior concludes that individuals are less likely to
intervene during an emergency when in the presence of others than when alone. By contrast,
little is known regarding the aggregated likelihood that at least someone present at an
emergency will do something to help. The importance of establishing this aggregated
intervention baseline is not only of scholarly interest, but is also the most pressing question
for actual public victims—will I receive help if needed? The current paper describes the
largest systematic study of real-life bystander intervention in actual public conflicts captured
by surveillance cameras. Using a unique cross-national video dataset from the United
Kingdom, Netherlands, and South Africa (N = 219), we show that in nine-out-of-ten public
conflicts, at least one bystander, but typically several, will do something to help. We record
similar likelihoods of intervention across the three national contexts, which differ greatly in
levels of recorded violent crime. Finally, we find that increased bystander presence is related
to a greater likelihood that someone will intervene. Taken together these findings allay the
widespread fear that bystanders rarely intervene to help. We argue that it is time for
psychology to change the narrative away from an absence of help and towards a new
understanding of what makes intervention successful or unsuccessful.
Keywords: Bystander effect; bystander intervention; aggression and violence; dangerous
emergencies; helping and prosocial behavior
* © 2019, American Psychological Association. This paper is not the
copy of record and may not exactly replicate the nal, authoritative
version of the article. Please do not copy or cite without authors'
permission. The nal article will be available, upon publication, via its
DOI: 10.1037/amp0000469
How likely is it that a victim of aggression in urban public space receives
bystander help? Perspectives from across the social sciences long suggest that the urban
experience of living in a mass of strangers dulls our sensitivity to the needs of others, thus
giving rise to ‘norms of non-involvement’ (Milgram, 1970, see also Goffman, 1963). In
resonance with this, many can recall media reports in which available bystanders did not
assist a victim of a public assault. A particularly renowned case is the 1964-rape and murder
of Kitty Genovese, about which the media reported that 38 bystanders watched for half-an-
hour without providing help (though note that recent work questions this account, see
Manning, Levine, & Collins, 2007). Prompted by this tragic incident, and the subsequent
public furor, social psychologists Darley and Latané (1968) set out to explain why those
individuals present did not intervene. In a series of highly influential experiments simulating
a number of different emergency scenarios, researchers found that additional bystander
presence prompted a diffusion of responsibility that constrained individual intervention
motivations (Latané & Darley, 1970; Latané & Nida, 1981). This phenomenon, which came
to be known as the ‘bystander effect,’ is well-replicated and appears in the majority of
introductory psychology textbooks (Ferguson, Brown, & Torres, 2018; Fischer et al., 2011).
The knowledge that an individual’s likelihood to intervene reduces in the
presence of others does not establish, however, the aggregated likelihood that at least
someone will help. From the perspective of the victim this remains the most important
question (Latané & Nida, 1981)—will I receive help if needed? The originators of the
research tradition were aware of the significance of the distinction (Darley & Latané, 1968)
highlighting that: “It is meaningless to compare directly individual with group responses,
since … there is a purely mechanical potential for getting more help with more people”
(Latané, 1981, p. 350). This distinction between responsibility diffusion and mechanical
helping potential is an important one, but is usually sidestepped in public and scientific
discourse (Stalder, 2008). For example, by default, experimental studies and student texts
foreground the individual bystander effect with the Kitty Genovese case, which, essentially, is
a story of aggregated non-intervention. What remains is a lasting impression that in
populated public settings, because of the high number of individuals present, victims will
rarely be helped when in need (Lurigio, 2015; Manning et al., 2007).
It is important therefore to recognize a key distinction between the likelihood of
individual intervention and the aggregate that at least someone provides help. Yet, in
comparison to the vast number of studies that examine intervention from the perspective of
the individual bystander, we know surprisingly little about the situational intervention
likelihood – i.e., the probability that at least one bystander at the emergency event intervenes.
One possible route-to-knowledge would be to attempt to establish this baseline through an
assessment of the previous experimental work. However, this would say little about public
safety in numbers, given that the group conditions of these experiments typically consist of
one, and at most a handful, of other bystanders, who are often confederates instructed not to
intervene. Further, if interested in intervention rates during aggressive encounters, such as the
Kitty Genovese incident, then the experimental literature offers limited insights, given that it
is ethically and practically difficult to simulate violent emergencies (Cherry, 1995; Liebst,
Heinskou, & Ejbye-Ernst, 2018). Finally, the use of sparsely populated simulation conditions
to understand violence intervention raises issues of ecological validity, given that public
violent crimes tend to be concentrated in high-density pedestrian areas. For example, a recent
bystander victimization study notes that around an average of sixteen bystanders are present
during public assaults (Liebst et al., 2018).
Outside of experimental work, a few studies have assessed the rate of bystander
intervention in real-life violent settings, but with mixed findings. Using official case files of
reported assaults, Felson and Steadman (1983) found that a bystander attempted conflict
mediation in 10.8% of incidents. A more recent assessment of police-reported case files of
public assaults showed that at least one bystander intervened in 73.8% of cases (Heinskou &
Liebst, 2017). Alternatively, in-situ observations of bystander actions during barroom assaults
recorded that between 26.2% and 39.5% of incidents contained de-escalatory bystander
intervention (Parks, Osgood, Felson, Wells, & Graham, 2013; Wells & Graham, 1999).
However, official case files and in-situ observations are unreliable methods for capturing
multiple and intricate bystander interactions occurring during chaotic violent events
(Morrison, Lee, Gruenewald, & Mair, 2016; Nassauer & Legewie, 2018). This limitation may
explain the large variation between studies, and further suggests an underestimation of the
intervention rate in previous work.
Video footage is increasingly recognized as the most complete data source for
assessing real-life human interactions (Gilmore & Adolph, 2017; Lindegaard & Bernasco,
2018). The recent proliferation of closed circuit television (CCTV) provides an opportunity to
systematically observe real-life emergencies as they occur in their naturalistic settings, yet its
use within the bystander field remains surprisingly rare. Levine and colleagues (2011)
pioneered the first systematic behavioral analysis of actual bystander behavior, captured by
public surveillance cameras. Coding the frame by frame actions of 228 bystanders across 42
CCTV clips, this work mapped the patterns of bystander actions associated with the
escalation and de-escalation of public violence. More recently, Liebst, Heinskou and Ejbye-
Ernst (2018) utilized surveillance footage of police-reported street assaults to establish the
likelihood and related risk factors of being victimized when intervening as a third-party.
Although novel in their methodological approach and findings, these rare video-based studies
did not establish the actual base rate of a victim receiving some form of bystander help.
Therefore, in the present study, we examine bystander intervention rates in 219
naturally occurring aggressive public incidents, captured by surveillance cameras across three
national contexts—the Netherlands, the United Kingdom and South Africa. This sample,
which is the largest and first cross-national video corpus of public conflicts assembled, offers
a unique opportunity to observe how common bystander intervention into real-life public
aggression actually is. With the inclusion of data from the city center of Cape Town, South
Africa, we provide an interesting cross-cultural reference category for intervention from a
place where perceptions of public safety are significantly lower (Lemanski, 2004; United
Nations, 2015). The purpose of the current study is as follows: First, we aim to determine the
percentage of real-life conflicts captured by public cameras where at least one bystander
intervenes. Second, in an explorative data analysis, we examine whether this intervention
likelihood varies across the national contexts that differ in public perceptions of safety.
Finally, we assess whether the situational intervention likelihood increases with additional
bystander presence. In other words, we aim to answer the often neglected question of whether
the increased number of potential helpers offsets the reduced willingness to help in each of
them, a question raised by Darley and Latané (1968; Latané, 1981) but most frequently
neglected in the subsequent experimental literature (Stalder, 2008).
Data and Sample
The data comprised of video surveillance footage of real-life public space
conflicts captured by actively monitored surveillance cameras in urban areas of the
Amsterdam, Netherlands, Cape Town, South Africa, and Lancaster, the United Kingdom. All
public surveillance cameras were located within the inner entertainment areas and central
business districts of the cities and typically captured public streets with shop fronts and
drinking venues, parks, plazas, pedestrian walk ways and transport station exteriors.
Following other surveillance camera studies (Levine et al., 2011; Liebst et al., 2018;
Lindegaard et al., 2017) video access was provided under the conditions that data would be
securely stored, shared only for legitimate research purposes and not with the wider public,
and that the identity of the individuals visible in the footage would be protected (see Philpot
et al., 2019). The research was authorized by the Dutch Ministry of Justice, the municipality
City of Cape Town, Lancaster City Council and was approved by the Ethics Committee for
Legal and Criminological Research (CERCO) of the Faculty of Law, Free University
Amsterdam and the Ethics Committee of the Department of Psychology, University of Exeter,
UK (ref: 2016/1295).
All data were recorded by municipality employed camera operatives, who
according to identical guidelines were instructed to record all incidents of public space
aggression that contained any level of conflict—from the mildest animated disagreements to
grave physical violence. This inclusive sampling strategy avoided skewing data towards
severe police reported incidents—a known issue in prior video-based crime analyses
(Lindegaard & Bernasco, 2018)—and provided a more representative picture of everyday
public space conflicts. Recordings were typically taken from the point at which the operative
noticed a potential for violence (e.g., postural and gestural displays of anger and aggression,
see Dael, Mortillaro, & Scherer, 2012; Levine et al., 2011) to the point when the incident had
either naturally dissipated or the police had separated the conflict parties.
The raw sample of video data contained 1,225 clips. From this raw sample, we
selected clips that conformed to the following inclusion criteria: The clip was taken from an
inner city, urbanized setting. The clip contained a conflict between at least two individuals
and did not show another type of incident (e.g., traffic accident, robbery or theft, begging,
drug dealing). The police or paramedics were not present when the conflict commenced (see
Levine et al., 2011). The clip was not a duplicate and had a technical quality (e.g., resolution,
brightness) that allowed for behavioral coding with none, or negligible, breaks in the
recorded interaction sequence (see Nassauer & Legewie, 2018) (for flowchart of video
exclusion criteria and sampling considerations, see Supplementary Materials at Open Science
The final video corpus comprised of 219 unique video clips of aggressive
incidents captured by actively monitored surveillance cameras in urban areas of the
Netherlands (n = 63), South Africa (n = 61), and the United Kingdom (n = 95). After the
application of exclusion criteria to the wider corpus, and prior to commencing analysis, we
conducted a statistical power analysis to assess whether the sample size of 219 videos would
be adequate to detect a small-medium effect size in line with meta-analytical bystander effect
evidence (Fischer et al., 2011) and of practical significance (Kirk, 1996). An insufficient
sample power could have indicated the need lower the stringent data inclusion criteria and
admit additional videos of lower quality and sequence capture. Several power analysis
scenarios indicated that a sample of around 210 videos would detect a small-medium effect
size (OR ~ 1.75), with a power of 95%, and α = .05. We therefore maintained the initial video
inclusion criteria and sample size.
Coding Procedure
The 219 videos were coded by four trained research assistants in accordance with a
systematic and interrater reliability tested codebook (for the full Observational Codebook
containing detailed descriptions of all study variables, see Supplementary Materials at Coding began by identifying the emergency conflict parties—the two
individuals between whom the encounter initially manifested itself. As we were interested
solely in baseline levels of intervention (irrespective of personal or group properties of the
bystander), all other individuals present in the situation were defined as bystanders.
In experimental bystander literature, intervention acts are defined from the
specifics of the simulation—for example, in a setup where a handful of pencils or coins are
‘accidentally’ dropped, intervention is operationalized accordingly as helping to pick them up
(Latané, 1981). In the current study, intervention is similarly defined from the affordance of
the emergency at hand—in this case, real-life conflicts that require the bystander to act
towards a perpetrator or victim in a manner that may soothe the conflict. Therefore, to code
bystander helping acts we applied standardized behavioral definitions of intervention, as
recently developed through video observational bystander research (Levine et al., 2011;
Liebst et al., 2018; Lindegaard et al., 2017). Specifically, a bystander was determined as an
intervener if they attempted to placate the conflict with any of the following acts: pacifying
gesturing; calming touches; blocking contact between conflict parties (see Figure 1c);
holding, pushing, or pulling an aggressor away from the conflict (see Figure 1c); consoling
victims of aggression; providing practical help to a physically harmed victim (for the full
Observational Codebook containing detailed descriptions of all study variables, see
Supplementary Materials at
For each clip, coders recorded the total number of interveners. This provided
descriptive detail regarding the sum of interveners per context and allowed for the assessment
of situational bystander intervention. In addition, coders noted the total number of bystanders
at an event (M = 16.29 minutes, SD = 13.16), the duration of the conflict (M = 3.27 minutes,
SD = 4.20), and the national context of each clip. To test the reliability of the coded variables,
we randomly selected 24 videos (11% of the total sample) for double coding (see Riffe, Lacy
& Fico, 2014). Both rater-observed variables included in the analyses (the number of
attentive bystanders present and the number of interveners) reached Krippendorff’s alpha
values of α = .85 and .87, respectively, indicating high levels of inter-rater reliability (De
Swert, 2012; Krippendorff, 2004) (for further information, see Supplementary Materials at Disagreements between coders were resolved by randomly selecting one
coder’s response.
We first descriptively examined the situational baseline of bystander
intervention across all 219 CCTV clips. We found that at least one bystander intervened in
90.9% of the situations, with an average of 3.76 interveners per video (SD = 3.01). Next,
using a Firth penalized likelihood logit analysis, we assessed whether these intervention
likelihoods varied across national contexts while controlling for the duration of the conflict.
With this model, we further assessed whether the situational intervention likelihood increased
with additional bystander presence.
We did not find evidence of a significant difference in the likelihood that at least
one bystander intervenes when comparing the Netherlands (OR = 0.61, 95% CI [0.19, 1.94],
p = .41) and the United Kingdom (OR = 1.59, 95% CI [0.51, 4.91], p = .42) to the South
African reference category (see Figure 2) (for full regression outputs, see Supplementary
Materials at Open Science Framework To further assess whether data
offers evidence in favor of a non-association, we used Bayesian information criteria (BIC) to
calculate Bayes factors (see Dienes, 2014) to compare the relative evidence between models
excluding and including national contexts—note, a benefit of BIC approximated Bayes
factors is that they do not require the specification of priors (see Wagenmakers, 2007). An
estimated Bayes factor (BF01 = 17.27) indicated evidence in favor of a non-association
between national context and intervention.
The number of bystanders at an event was positively associated with the
situational likelihood of intervention (OR = 1.10, 95% CI [1.03, 1.18], p = .008, BF01 = 0.03),
with each additional bystander increasing the odds that a victim receives help by factor 1.1
(see Figure 2). When assessed in a standardized effect size metric (see Gelman, 2008), the
number of bystanders is a very strong (Rosenthal, 1996) predictor of situational intervention
(OR = 12.72, 95% CI [1.96, 82.50], p = .008, BF01 = 0.03), though the broad confidence
interval leaves uncertainty regarding the precision of this estimate.
Decades of social scientific research has been concerned with the issue of
bystander apathy and lack of safety in numbers during public emergency events, with the
bystander effect hypothesis the epicenter of these discussions. This hypothesis is an example
of the way in which a scholarly theory can come to be adopted by the public as a ‘lay
concept’ (Giddens, 1984) to understand incidents where no-one intervened when required
(Lurigio, 2015; Manning et al., 2007). Without careful interpretation of the findings, the
individual likelihood of intervention and the aggregate likelihood that at least someone
intervenes become conflated, and may provide an easily misleading answer to the public’s
key question: Will I be helped if victimized in public? In fifty years of bystander research,
little has been done to answer this question (Stalder, 2008), meaning we still do not know
what the aggregate likelihood is that a victim of public aggression will receive some help.
In an effort to establish a veridical and ecologically valid baseline of
intervention in actual public conflicts, we systematically observed real-life bystander
behavior captured by public surveillance footage from three national contexts. We found that
in nine-out-of-ten conflicts at least one person—but typically several—did something to help.
These findings challenge established assumptions that urban living (Milgram, 1970), and
certain neighborhood contexts (Sampson, Raudenbush, & Earls, 1997), undermine social
responsibility and the related willingness to intervene into public violence. Therefore, in
contrast to the notion that non-involvement is the norm in urban environments, the high
levels of intervention found in this study across different national and urban contexts suggests
that involvement is the norm in real-life inner-city public conflicts.
Initially, the high rate of intervention found in this study may be surprising, as the
bystanders are entering real-life conflicts with some risk of victimization (see Liebst et al.,
2018) that may deter involvement. On the other hand, this willingness to intervene is in line
with behavioral economic evidence from public-good experiments demonstrating that third-
parties frequently engage in personally costly behaviors to punish norm-violating individuals
and to preserve social norms (Fehr & Fischbacher, 2004; Fehr & Gächter, 2002; Gächter,
2014). The overall high level of intervention is also consistent with experimental bystander
research showing that individuals are likely to respond to dangerous (although not actually
violent) emergencies, even at an individual cost (Fischer, Greitemeyer, Pollozek, & Frey,
2006; Harari, Harari, & White, 1985; for review, see Fischer et al., 2011). Here, perceptions
of victim distress may motivate bystanders to help as a direct means to relieve the victim’s
suffering (Batson, 2011) or in order to alleviate their own unpleasant arousal from witnessing
the emergency (Cialdini et al., 1987; Piliavin, Dovidio, Gaertner, & Clark III, 1981). More
broadly, this underlying propensity for empathy with others and third-party peacekeeping
behavior is argued to be central for the maintenance of co-operative societal living (de Waal,
2007; Verbeek & Peters, 2018).
It is further surprising that the magnitude of the intervention rates does not vary
between the three national-city contexts, given that inner-city Cape Town, South Africa has
comparatively lower perceptions of public safety (Lemanski, 2004; United Nations, 2015)
that may elicit significantly less intervention (because of personal risk aversion) or more
intervention (due to greater perceived victim need). The non-association evidenced in this
study would indicate that it is not the level of perceived danger that sets the overall rate of
intervention. Instead it appears that any sign of danger is perceived as a signal which prompts
the importance or imperative of intervention. This is assumption is in line with meta-
analytical evidence that bystander apathy decreases when the situation is perceived as
dangerous (Fischer et al., 2011). The consistent intervention rate found across the disparate
national cities also supports anthropological work suggesting that third-party conflict
resolution is a human universal (Brown, 1991), with a plausible evolutionary basis, as
reflected by the abundance of peacekeeping practices among our closest non-human primate
kin (Boehm, 2000; de Waal, 2000).
An often neglected question in bystander research is whether a low individual
likelihood to intervene can go hand-in-hand with a high aggregate level of intervention
(Stalder, 2008). Importantly, by examining intervention on the situational rather than the
individual level our research does not evaluate whether bystanders are less likely to provide
help when in the presence of other bystanders compared to when they are alone (i.e., the
bystander effect). Our results do suggest, however, that with increasing numbers of
bystanders, the greater the likelihood it is that at least someone will intervene to help. It is
important to recognize that a positive association between bystander numbers and the
likelihood of receiving help has been establish before. Specifically, Stalder (2008) re-
analyzed previous meta-analytical work (Latané & Nida, 1981) and found a positive
correlation between group size and help provision under conditions in which bystanders were
seperated, yet knew of one another’s presence. The current study adds to this account with
evidence from systematic obsevations of real-life public space emergencies.
The immediate theoretical implication of these accumulating bodies of evidence
is that bystander research should distinguish between responsibility diffusion and mechanical
helping potential that may operate inversely on the individual and situational levels,
respectively—in other words, while additional numbers may reduce the individual helping
propensity, it can provide a larger pool from which help-givers may be sourced. The finding
that help is more likely given in populated settings also has several societal implications.
First, it informs current debates within criminology regarding whether populated settings
provide a safety in numbers, or alternatively, facilitates violent events (Hillier, 2004;
Summers & Johnson, 2017). Our data supports a rapprochement between these views: While
the co-presence of more people may increase the opportunities for conflictual interactions,
additional numbers conjointly offer more opportunities for bystander intervention. Second,
this latter point offers an alternative picture to the common public association between the
‘unknown other’ and ‘stranger danger’ (Hale, 1996; Jackson & Gouseti, 2016). Finally, for
policing strategies, our results indicate that groups are willing to ‘self-police’ (Reicher et al.,
2007) conflicts, and as such, bystanders are an unharnessed crime preventive resource for
informally regulating violence prior to the arrival of the police.
The bystander effect paradigm, like other research pillars of the social sciences
(Mortensen & Cialdini, 2010), was prompted by a naturally occurring atrocity (e.g., the Kitty
Genovese case). Subsequent bystander work, however, has almost exclusively been
conducted in sparsely populated experimental setups, simulating trivial events with little
potential danger or negative costs for the intervener (for critique, see Fischer et al., 2006).
The incongruence between real-life aggressive public assaults and simulations risks creating
a body of knowledge ill-equipped to explain the real-life bystander phenomenon that initially
prompted such research (see Cialdini’s (1980; Mortensen & Cialdini, 2010) ‘Full Cycle
Psychology’—see also, Swann & Jetten, 2017). The current paper offers a methodological
corrective by systematically observing bystander behavior in dangerous, populated,
naturalistic settings captured by video footage. By observing the real-world phenomenon, the
low ecological validity of the experimental bystander work becomes more apparent. First,
emergencies in public spaces can be dangerous. Second, people in urban spaces are rarely
alone or solely in the company of one-or-two others. Today, with the increased availability of
high-quality video recordings of actual emergencies there are new opportunities for the study
of human violence and pro-social behavior in real-life settings (Gilmore & Adolph, 2017;
Lindegaard & Bernasco, 2018). While the current study assessed intervention as a situational
aggregate, future video studies should prioritize the inclusion of individual-level predictors of
helping, known to influence intervention. For example, evidence shows that gender roles may
shape expressions of helping behavior (Eagly, 2009), while other studies report a robust
association between shared group membership and the likelihood of help provision (Levine &
Manning, 2013; Lindegaard et al., 2017; Slater et al., 2013).
The current study has limitations that warrant discussion. In light of the very
high intervention rate reported, also in comparison to prior research (e.g., Felson &
Steadman, 1983), we consider factors that may have inflated our intervention figure. First, it
could be argued that we deployed an over-inclusive behavioral definition of bystander
intervention, e.g., at a minimum, a calming open-handed gesture towards a perpetrator or
consoling a victim of aggression1 (for further discussion around the theoretical justification of
including consolation in the intervention measure, see Supplementary Materials at However, this may be balanced by the fact that all videos lacked sound,
and thus did not allow for the coding of verbal interventions, such as spoken pleas to ‘calm
down’ or warnings that ‘I’m calling the police.’ Furthermore, the current paper did not code
aggressive bystander interventions, which may at times be used as a means to de-escalate
conflict (Stott, Hutchison, & Drury, 2001).
A second limitation of the current data is the sampling skewness towards inner-city
areas with a high density of restaurants, bars, and discotheques. As such, it is plausible that a
proportion of bystanders consumed alcohol that may have affected their risk assessment
1 On the request of an anonymous reviewer, we assessed the extent to which consolatory acts may have
overinflated the intervention figure. To this end, we examined whether consolation was the ONLY bystander
intervention performed in a random subsample of 85 videos (38.9% of all data). We found that 1 out of 85 cases
contained consolation without any prior bystander intervention act. As such, the inclusion of consolation in the
intervention measurement is likely to have negligible impact on the overall findings.
(Weafer & Fillmore, 2016). This inner-city restriction on the data selection strategy meant
that we did not capture violence intervention rates in the encompassing areas, which
particularly in the South African case might provide different results because surrounding
disadvantaged areas (referred to as townships) have disproportionally higher levels of public
crime (South African Police Service, 2018). Furthermore, our inner-city data bias means that
we do not have comparative data of intervention in other public spaces, including conflicts at
music and sporting events, or sexual aggression on campuses (Banyard, 2015; Stott, Adang,
Livingstone, & Schreiber, 2007). As a result, we are not able to say very much about the
extent to which the bystander intervention likelihood varies across wider social contexts.
A final methodological limitation of our analytical strategy is that it is non-
experimental, and therefore lacks the counterfactual evidence for drawing the strongest
conclusions about the relations between bystander intervention, group size, and national
contexts. Although our correlational design cannot provide the most rigorous evidence of
causality, its external validity is unparalleled.
According to conventional wisdom, there is an epidemic of bystander non-
involvement during public emergencies. Challenging this view, the current cross-national
study of video data shows that intervention is the norm in actual aggressive conflicts, with
more populated settings providing a greater likelihood that someone helps. This is reassuring
for potential victims of violence, the public as a whole, and may inform crime preventive
efforts to make use of the already very active bystanders. Given these societal implications,
we believe it is critical that future bystander research and its public dissemination flag the
often-neglected distinction between the bystander effect and the simple mechanic of getting
more help with more people. In shifting the perspective from an absence of help to an almost
ever-presence, we leave behind the question of ‘why don’t individuals help?’ and explore a
new avenue of enquiry asking: ‘what makes intervention successful?’
Banyard, V. L. (2015). Toward the next generation of bystander prevention of sexual and
relationship violence: Action coils to engage communities. London: Springer.
Batson, C. D. (2011). Altruism in humans. Oxford: Oxford University Press.
Boehm, C. (2000). Conflict and the evolution of social control. Journal of Consciousness
Studies, 7(1-2), 79-101.
Brown, D. E. (1991). Human universals. New York: McGraw-Hill.
Cherry, F. (1995). The 'stubborn particulars' of social psychology: Essays on the research
process. London: Routledge.
Cialdini, R. B. (1980). Full-cycle social psychology. Applied Social Psychology Annual, 1,
Cialdini, R. B., Schaller, M., Houlihan, D., Arps, K., Fultz, J., & Beaman, A. L. (1987).
Empathy-based helping: Is it selflessly or selfishly motivated? Journal of Personality
and Social Psychology, 52(4), 749.
Dael, N., Mortillaro, M., & Scherer, K. R. (2012). Emotion expression in body action and
posture. Emotion, 12(5), 1085-1101.
Darley, J. M., & Latané, B. (1968). Bystander intervention in emergencies: Diffusion of
responsibility. Journal of Personality and Social Psychology, 8(4), 377-383.
De Swert, K. (2012). Calculating inter-coder reliability in media content analysis using
Krippendorff's Alpha. Retrieved from
de Waal, F. B. M. (2000). Primates - a natural heritage of conflict resolution. Science,
289(5479), 586-590. doi:10.1126/science.289.5479.586
de Waal, F. B. M. (2007). The "Russian doll" model of empathy and imitation. In S. Bråten
(Ed.), On being moved: From mirror neurons to empathy (pp. 49–69). Amsterdam:
Dienes, Z. (2014). Using Bayes to get the most out of non-significant results. Frontiers in
Psychology, 5(781), 1-17. doi:10.3389/fpsyg.2014.00781
Eagly, A. H. (2009). The his and hers of prosocial behavior: An examination of the social
psychology of gender. American Psychologist, 64(8), 644-658. doi:10.1037/0003-
Fehr, E., & Fischbacher, U. (2004). Third-party punishment and social norms. Evolution and
Human Behavior, 25(2), 63-87.
Fehr, E., & Gächter, S. (2002). Altruistic punishment in humans. Nature, 415(6868), 137-140.
Felson, R. B., & Steadman, H. J. (1983). Situational factors in disputes leading to criminal
violence. Criminology, 21(1), 59-74. doi:10.1111/j.1745-9125.1983.tb00251.x
Ferguson, C. J., Brown, J. M., & Torres, A. V. (2018). Education or indoctrination? The
accuracy of introductory psychology textbooks in covering controversial topics and
urban legends about psychology. Current Psychology, 37, 574-582.
Fischer, P., Greitemeyer, T., Pollozek, F., & Frey, D. (2006). The unresponsive bystander: Are
bystanders more responsive in dangerous emergencies? European Journal of Social
Psychology, 36(2), 267-278. doi:10.1002/ejsp.297
Fischer, P., Krueger, J. I., Greitemeyer, T., Vogrincic, C., Kastenmüller, A., Frey, D., . . .
Kainbacher, M. (2011). The bystander-effect: A meta-analytic review on bystander
intervention in dangerous and non-dangerous emergencies. Psychological Bulletin,
137(4), 517-537. doi:10.1080/00224545.2012.697931
Gächter, S. (2014). Human pro-social motivation and the maintenance of social order. In E.
Zamir & D. Teichman (Eds.), The Oxford Handbook of Behavioral Economics and
the Law (pp. 28-60). Oxford: Oxford University Press.
Gelman, A. (2008). Scaling regression inputs by dividing by two standard deviations.
Statistics in Medicine, 27(15), 2865-2873. doi:10.1002/sim.3107
Giddens, A. (1984). The constitution of society: Outline of the theory of structuration: CA:
University of California Press.
Gilmore, R. O., & Adolph, K. E. (2017). Video can make behavioural science more
reproducible. Nature Human Behavior, 1(7). doi:10.1038/s41562-017-0128
Goffman, E. (1963). Behavior in public places: Notes on the social organization of
gatherings. New York: Free Press of Glencoe.
Hale, C. (1996). Fear of crime: A review of the literature. International Review of
Victimology, 4(2), 79-150.
Harari, H., Harari, O., & White, R. V. (1985). The reaction to rape by American male
bystanders. The Journal of Social Psychology, 125(5), 653-658.
Heinskou, M. B., & Liebst, L. S. (2017). Gadevold. En sociologisk kortlægning af vold i
byen. København: Djøf Forlag.
Hillier, B. (2004). Can streets be made safe? Urban Design International, 9(1), 31-45.
Jackson, J., & Gouseti, I. (2016). Threatened by violence: Affective and cognitive reactions to
violent victimization. Journal of Interpersonal Violence, 31(18), 2987-3016.
Kirk, R. E. (1996). Practical significance: A concept whose time has come. Educational and
Psychological Measurement, 56(5), 746-759.
Krippendorff, K. (2004). Reliability in Content Analysis. Human Communication Research,
30(3), 411-433. doi:10.1111/j.1468-2958.2004.tb00738.x
Latané, B. (1981). The psychology of social impact. American Psychologist, 36(4), 343-356.
Latané, B., & Darley, J. M. (1970). The unresponsive bystander: Why doesn't he help? New
York: Appleton-Century-Crofts.
Latané, B., & Nida, S. (1981). Ten years of research on group size and helping.
Psychological Bulletin, 89(2), 308-324. doi:10.1037/0033-2909.89.2.308
Levine, M., & Manning, R. (2013). Social identity, group processes, and helping in
emergencies. European Review of Social Psychology, 24(1), 225-251.
Lemanski, C. (2004). A new apartheid? The spatial implications of fear of crime in Cape
Town, South Africa. Environment and Urbanization, 16(2), 101-112.
Levine, M., Taylor, P. J., & Best, R. (2011). Third parties, violence, and conflict resolution:
The role of group size and collective action in the microregulation of violence.
Psychological Science, 22(3), 406-412. doi:10.1177/0956797611398495
Liebst, L. S., Heinskou, M. B., & Ejbye-Ernst, P. (2018). On the actual risk of bystander
intervention: A statistical study based on naturally occurring violent emergencies.
Journal of Research in Crime and Delinquency, 55(1), 27-50.
Lindegaard, M. R., & Bernasco, W. (2018). Lessons learned from crime caught on camera.
Journal of Research in Crime and Delinquency, 55(1), 155-186.
Lindegaard, M. R., Liebst, L. S., Bernasco, W., Heinskou, M. B., Philpot, R., Levine, M., &
Verbeek, P. (2017). Consolation in the aftermath of robberies resembles post-
aggression consolation in chimpanzees. PloS one, 12(5), e0177725.
Lurigio, A. J. (2015). Crime narratives, dramatizations, and the legacy of the Kitty Genovese
murder: A half century of half truths. Criminal Justice and Behavior, 42(7), 782-789.
Manning, R., Levine, M., & Collins, A. (2007). The Kitty Genovese murder and the social
psychology of helping: The parable of the 38 witnesses. American Psychologist,
62(6), 555-562. doi:10.1037/0003-066X.62.6.555
Milgram, S. (1970). The experience of living in cities. Science, 167(3924), 1461-1468.
Morrison, C., Lee, J. P., Gruenewald, P. J., & Mair, C. (2016). The reliability of naturalistic
observations of social, physical and economic environments of bars. Addiction
Research & Theory, 24(4), 330-340. doi:10.3109/16066359.2016.1145674
Mortensen, C. R., & Cialdini, R. B. (2010). Fullcycle social psychology for theory and
application. Social and Personality Psychology Compass, 4(1), 53-63.
Nassauer, A., & Legewie, N. (2018). Video data analysis: A methodological frame for a novel
research trend. Sociological Methods & Research, 1-40.
Parks, M. J., Osgood, D. W., Felson, R. B., Wells, S., & Graham, K. (2013). Third party
involvement in barroom conflicts. Aggressive Behavior, 39(4), 257-268.
Philpot, R., Liebst, L. S., Møller, K. K., Lindegaard, M. R., & Levine, M. (2019). Capturing
violence in the night-time economy: A review of established and emerging
methodologies. Aggression and Violent Behavior. 46, 56-65.
Piliavin, J. A., Dovidio, J. F., Gaertner, S. L., & Clark III, R. D. (1981). Emergency
intervention. New York: Academic Press.
Reicher, S., Stott, C., Drury, J., Adang, O., Cronin, P., & Livingstone, A. (2007). Knowledge-
based public order policing: Principles and practice. Policing: A Journal of Policy
and Practice, 1(4), 403-415.
Riffe, D., Lacy, S., & Fico, F. (2014). Analyzing media message: Using quantiative content
analysis in research. London: Routledge.
Rosenthal, J. A. (1996). Qualitative descriptors of strength of association and effect size.
Journal of Social Service Research, 21(4), 37-59. doi:10.1300/J079v21n04_02
Sampson, R. J., Raudenbush, S. W., & Earls, F. (1997). Neighborhoods and violent crime: A
multilevel study of collective efficacy. Science, 277(5328), 918-924.
Slater, M., Rovira, A., Southern, R., Swapp, D., Zhang, J. J., Campbell, C., & Levine, M.
(2013). Bystander responses to a violent incident in an immersive virtual
environment. PLOS ONE, 8, e52766. doi:10.1371/journal.pone.0052766
South Africa Police Service (2018). Crime Statistics 2017/2018. Pretoria: SAPS Head Office.
Retrieved from
Stalder, D. R. (2008). Revisiting the issue of safety in numbers: The likelihood of receiving
help from a group. Social Influence, 3(1), 24-33. doi:10.1080/15534510701766181
Stott, C., Adang, O., Livingstone, A., & Schreiber, M. (2007). Variability in the collective
behaviour of England fans at Euro 2004: ‘Hooliganism’, public order policing and
social change. European Journal of Social Psychology, 37, 75–100.
Stott, C., Hutchison, P., & Drury, J. (2001). ‘Hooligans’ abroad? Intergroup dynamics, social
identity and participation in collective ‘disorder’at the 1998 World Cup Finals. British
Journal of Social Psychology, 40(3), 359-384.
Summers, L., & Johnson, S. D. (2017). Does the configuration of the street network influence
where outdoor serious violence takes place? Using space syntax to test crime pattern
theory. Journal of Quantitative Criminology, 33(2), 397-420. doi:10.1007/s10940-
Swann, W. B. J., & Jetten, J. (2017). Restoring agency to the human actor. Perspectives on
Psychological Science, 12(3), 382-399. doi:10.1177/1745691616679464
United Nations (2015). Human development report: Work for human development. New
York: United Nations Development Programme. Retrieved from
Verbeek, P., & Peters, B. A. (2018). The nature of peace. In P. Verbeek & B. A. Peters (Eds.),
Peace ethology: Behavioral processes and systems of peace (pp. 1-13). Hoboken,
New Jersey: Wiley & Sons Ltd.
Wagenmakers, E.-J. (2007). A practical solution to the pervasive problems of p values.
Psychonomic Bulletin & Review, 14(5), 779-804. doi:10.3758/BF03194105
Weafer, J., & Fillmore, M. T. (2016). Low-dose alcohol effects on measures of inhibitory
control, delay discounting, and risk-taking. Current Addiction Reports, 3(1), 75-84.
Wells, S., & Graham, K. (1999). The frequency of third-party involvement in incidents of
barroom aggression. Contemporary Drug Problems, 26(3), 457-480.
Figure 1a. On the bottom right-hand side, a man dressed in a white shirt
assaults another man who is on the ground. Some bystanders observe.
Figure 1b. To the bottom left-hand side, two bystanders leave their
standing positions and approach the conflict parties.
Figure 1c. The two bystanders are joined by others. A male bystander in
a dark shirt and jeans pulls the main aggressor from his target, while a
female bystander steps between the conflict parties and extends both
arms out in a blocking motion.
Figure 2. Estimates (odds ratio) coefficients and 95% standard errors. Firth logistic regression model,
N = 219, including control for conflict duration (estimate not shown in figure).
* © 2019, American Psychological Association. This paper is not the
copy of record and may not exactly replicate the nal, authoritative
version of the article. Please do not copy or cite without authors'
permission. The nal article will be available, upon publication, via its
DOI: 10.1037/amp0000469
... These pioneering works demonstrated that when individual contributions are ambiguous and group members are not assigned particular roles (such as 'leader'), attributions of responsibility follow a self-serving rule: individuals tend to claim more credit for successes (Fig. 1d) and avoid responsibility for failures [32][33][34][35] . Research has shown that the decreased sense of responsibility in groups can promote adverse and undesirable behaviors 36 , such as free-riding 1,37-39 , groupthink 40 , social loafing 41 , abstaining in elections 42 and inaction in emergency situations (known as the bystander effect [43][44][45] ; but see new findings 46 ). Such behaviours may emerge because individuals in groups act as if they have delegated responsibility and agency to others 1,37 . ...
Full-text available
Research investigating collective decision-making has focused primarily on the improvement of accuracy in collective decisions and less on the motives that drive individuals to make these decisions. We argue that a strong but neglected motive for making collective decisions is minimizing the material and psychological burden of an individual’s responsibility. Making difficult decisions with others shields individuals from the consequences of negative outcomes by reducing regret, punishment and stress. Considering shared responsibility as a another key motivation to join groups helps understand behaviours with societal implications such as political voting, committing norm violations, predicting natural disasters and making health-related decisions.
ResearchGate has not been able to resolve any references for this publication.