Professional Documents
Culture Documents
DOI 10.1007/s11292-012-9144-1
Abstract
Objectives To empirically examine the absolute and relative impact of situational
characteristics and confidence in the criminal justice system on public support for
vigilantism.
Methods In an experimental study with a between-subjects design, members of a
Dutch household panel (n 0 1,930) responded to vignettes about vigilantism that were
varied across two experimental factors: (1) type of precipitating crime and (2) type of
formal sentence for the precipitating offender. In the measurement of support, we
distinguished between outrage at vigilantism, empathy with the vigilante, and desired
punishment for the vigilante. Confidence was assessed 1 month later.
Results Our findings show that situational characteristics have a substantial and
independent influence on support for vigilantism, in addition to the role of confidence. This means that when citizens express support for those who take the law into
their own hands, this is not necessarily rooted in a lack of confidence in the criminal
justice system. Furthermore, all three measures of support were affected more by the
situational characteristics than by confidence.
Conclusions Citizens are nuanced in their judgment of vigilantism and sensitive to
contextual information, which is in line with other recent findings regarding public
punitiveness. Future studies should assess whether the findings can be generalized to
other settings where citizens cannot rely (as much) on the state to deal with crime.
Keywords Confidence . Criminal justice system . Public support . Punishment .
Vigilantism
Introduction
Citizens who take the law into their own hands tend to spark heated debates.
Moreover, it is not uncommon for vigilantes to receive considerable public support
for their behavior, even when they go as far as killing an (alleged) offender. A famous
example is the case of Marianne Bachmeier, who took the law into her own hands
inside of a courtroom in Lbeck, West Germany, in 1981. She pulled out a gun and
fired eight times at the man who was on trial for the abduction, sexual abuse, and
murder of her 7-year-old daughter. He died on the courtroom floor (Kpcke 2006).
Many people felt sympathy for the vigilante, sending her support letters, money, and
flowers while she was in detention. Interestingly, public opinion shifted once more
negative information surfaced about Marianne Bachmeiers life (Justiz und Selbstjustiz
2008). It became known that she had an additional child at age 16 and another one at
age 18, who were both given up for adoption, and that her father was involved with
the Waffen-SS. All in all, Marianne Bachmeier no longer seemed to fit the role of the
innocent mother that was initially ascribed to her, which affected the publics
judgment of her act of vigilantism.
What does it mean when the public supports citizens who deal with crime outside
of the law? Does it signify that the legitimacy of the criminal justice system is at
stake?1 Public support for vigilantism is often seen as indicative of diminished
confidence in the criminal justice system (e.g., Harnischfeger 2003; Tankebe 2009).
However, little is known about actual determinants of support for those who take the
law into their own hands. Despite its relevance from both a social and legal perspective,
this phenomenon has been relatively overlooked in the criminological literature. We do
not know what public support for vigilantism constitutes, what causes it and what
implications it might have for society. This is unfortunate as such knowledge can
provide unique insights into how people view justice within and outside of the criminal
justice system. Why do citizens find certain acts of vigilantism justified even though
such acts are against the law? How do people view the vigilante and his victim? Why
does the public not always find it necessary for vigilantes to be punished for their
criminal behavior?
In this paper we seek to reach a better understanding of public support for
vigilantism. We will empirically test whether support for vigilantism can be explained
through a low level of confidence in the criminal justice system. We will also
introduce and test an alternative proposition, namely that support is affected by
characteristics related to the vigilantism situation itself. This will be studied by
conducting an experimental study. We conclude by discussing the implications of
our findings, and by offering suggestions for future research.
over the legitimate use of force in most developed countries, thereby harnessing the
emotions of citizens into a civilized justice system.2 The modern state in fact
characterizes itself by only permitting force that is used in preservation and enforcement of the law (Sarat and Kearns 1992). The presence of law is generally assumed to
result in a less violent society because it provides peaceful alternatives for handling
conflicts (Cooney 1997). The very existence of a criminal justice system can be
justified by contrasting it with the unjustifiability of citizens retaliating against
wrongdoers (Gardner 1998). One of the aims of restricting the prosecution and
punishment of offenders to an impartial third party is to prevent emotional excesses.
This function has been described in the literature as the displacement of retaliation:
The blood feud, the vendetta, the duel, the revenge, the lynching: for the
elimination of these modes of retaliation, more than anything else, the criminal
law as we know it today came into existenceThe displacement function of
criminal law always was and remains today one of the central pillars of its
justification (Gardner 1998, p.32).
The criminal law thus aims to channel the natural instinct for retaliation and turn
hot vengeance into cool, impartial justice (MacCormick and Garland 1998, p.26).
This channeling is deemed necessary as it is commonly believed that people cannot
be rational in the judgment of their own cases (see e.g., Locke 1689/1967).
Vigilantism is commonly said to occur when citizens, from whom authorities are
supposed to derive their legitimacy, believe that the criminal justice system is
inadequate (Abrahams 2002; Adinkrah 2005; Baker 2001; Benesh and Howell
2001; Goldstein 2003; Robinson and Darley 1995). Vigilantism can be seen as the
result of a so-called injustice gap: a discrepancy between the desired and actual
outcome (Exline et al. 2003). Taking the law into ones own hands is a way of
reducing this perceived injustice gap and restoring ones sense of justice. Vigilantism
may occur when people do not want the criminal justice authorities to get involved,
when the authorities fail to respond, or when the authorities were involved but not to
the satisfaction of the affected party. Situations like the latter include victims or their
relatives who physically attack defendants upon finding out that they are acquitted by
a judge or given too lenient a sentence (Weber 2000). The presented case of
Marianne Bachmeier provides an illustrative example. Even though an arrest had
been made, and a defendant was brought to trial, she apparently felt that none of the
possible judicial sentences would be adequate. She probably saw the death penalty as
the only satisfactory result, which was not an option in the German criminal justice
system, and effectuated this desired punishment by fatally shooting the defendant.
Similarly, in South Africa there are reports of citizens who, due to dissatisfaction with
the formal response to crime, collectively pay bail of jailed convicts and subsequently
kill them (Minnaar 2001).
When citizens resort to vigilantism, it can be a sign that the states capability
to displace retaliation is in jeopardy. However, what may matter most on a
societal level is the collective approval of citizens who take the law into their
own hands. Research suggests that support for vigilantism may increase the
likelihood of vigilante behavior in a community (Weisburd 1998). Even if
2
There are exceptions to this monopoly, such as self-defense and citizens arrest.
vigilantism itself is rare, public support for it can be widespread. Formal responses
like the arrest, prosecution and sentencing of vigilantes can trigger considerable
public controversy. Ideally, a legal system should represent the moral consensus of
the community, maximizing voluntary compliance. In reality, a justice system will
never be able to fully match the values and expectations of all citizens that it is
supposed to serve. Legal rules will always diverge to a certain extent from citizens
principles. However, when these discrepancies become too large, the danger exists
that public respect for the legal system will be lost (Carlsmith 2008; Darley 2001;
Roberts and Stalans 1997). Such a lack of respect for the law would be detrimental to
the functioning of the criminal justice system, as citizens often obey the law because
they find the norms to be legitimate and deserving of compliance (Coffee 1991; Tyler
1990). It has been argued in the literature that moral contempt for specific laws may
generalize to the entire criminal code (Greene and Darley 1998). Thus, if there is no
moral consensus in the community vis--vis the formal reactions to vigilantism, the
state should at least attempt to explain to its citizens why the existing practice is to be
preferred (Greene and Darley 1998). If not, people may lose confidence in the law
and the legal authorities, which may ironically bring about a higher frequency of
vigilantism itself.
self-defense and citizens arrest, which are related but quite distinct phenomena (see
e.g., Cohen 1989).
determine how people respond to vigilantism. Below we present two main propositions in this regard.
Confidence proposition
According to what we refer to as the confidence proposition, when a citizen supports
vigilantism, this is due to a low general level (or lack) of confidence in the criminal
justice system. In other words, a negative view of the criminal justice system is
thought to result in a rejection of the norm of the state monopoly on force, and in a
positive view of those who take the law into their own hands (Goldstein 2003;
Harnischfeger 2003; Lenz 1988; Tankebe 2009). This hypothesis appears to be
feasible at first glance. After all, why would people who do have confidence in
justice support those who defy the law? To date there has nonetheless been little
attempt to test this idea. Only one study has reliably evaluated this proposition, using
survey data to examine the effect of attitudes toward police on public support for
vigilantism in Ghana (Tankebe 2009). Findings confirm the expected negative relation between (perceived) police trustworthiness and support for vigilantism.
Situation proposition
In this section we will argue that public responses to an act of vigilantism are (at least in
part) affected by situational characteristics. In other words, the situation proposition
maintains that support for vigilantism is shaped by aspects of the vigilantism case
itself. This is in line with findings from the field of social justice research, which
suggest that peoples judgments of issues such as fairness, justice, and deservingness
at least partially depend on contextual factors (Tyler and Smith 1997). Research
shows public reactions to crime to be affected by situational aspects such as the
severity of the offense (Cullen et al. 1983; Darley et al. 2000; Jacoby and Cullen
1998; Stylianou 2003), crime type (Carlsmith et al. 2002; Rossi et al. 1985; Warr
1989), perceived responsibility of the offender (Heider 1958; Joseph and Tedeschi
1983; Shaw and Sulzer 1964) and characteristics of the offender (Applegate et al.
1996; Rossi et al. 1985; Scott et al. 2006; Warr 1989). When there are justifications,
excuses or mitigations, this can also affect reactions to harm (Darley and Pittman
2003). Provoked harm for instance leads people to be milder in their retributive
judgments (Vidmar 2001). Likewise, when people feel empathetic with an offender,
this is said to reduce punitiveness (Unnever and Cullen 2009) and increase forgiveness (Exline et al. 2003). It has also been argued in the literature that when someone
finds out that a victim has previously harmed the offender, this may result in reduced
empathy with the victim and increased empathy with the offender (Hoffman 1990;
Lerner 1980).
In line with the social justice literature, we predict that situational characteristics, such
as those related to the act of vigilantism or to the people involved, will affect public
support for vigilantism. In other words, rather than assuming that peoples response to an
act of vigilantism is exclusively a result of their overall level of confidence in the
criminal justice system, we expect them to be sensitive to case-specific information. If
the precipitating crime is for instance particularly heinous, this may result in a more
positive reaction to an act of vigilantism, regardless of someones view of the criminal
justice system. Importantly, this situational context can also concern the criminal
justice system: the kind of formal response or lack thereof to a specific (precipitating)
crime. If for instance the (alleged) perpetrator of a crime is not punished by the
authorities, this can result in a perception of injustice, which may manifest itself in
support for a subsequent act of vigilantism (cf. Goldberg et al. 1999). Perceptions of
the criminal justice system can thus play a role on two levels: on both a general and
specific (situational) one.
We found only one experimental study that provides empirical evidence for the role
of situational characteristics within the context of vigilantism. Based on an experiment
with vignettes and a between-subjects design, Skitka and Houston (2001) concluded
that whether respondents perceived vigilantism against a murder suspect as fair
depended on the murder suspects apparent guilt (guilty, innocent or ambiguous).
Methods
The study consists of two parts. In the first part we presented respondents with
vignettes in an experimental design, and measured their support for vigilantism. In
the second part, which was conducted 1 month later, we assessed confidence in the
criminal justice system. We created this time lapse to reduce possible effects of
reading and responding to vignettes on the confidence ratings, and to prevent
respondents from guessing the purpose of the study. Likewise, participants were
not informed that the measures were part of the same study.
Design
In a between-subjects experimental design, each respondent was presented with two
vignettes that together present information about all three events from the vigilantism
event sequence. Vignette 1 describes the precipitating crime, which was followed by
a first set of questions. Participants were subsequently presented with Vignette 2,
which describes the formal response to the precipitating crime and a subsequent act of
vigilantism (against the precipitating offender). This vignette was followed by a
second set of questions, measuring support for vigilantism.
The contents of the vignettes differ in correspondence with the experimental
manipulation of two situational characteristics. Experimental factor 1 is the type of
precipitating crime (three versions). Experimental factor 2 is the formal sentence for
the precipitating offender (four versions).3 Both factors are expected to affect how the
subsequent act of vigilantism is judged by respondents. Table 1 provides an overview
of the design; the manipulation, vignettes and measures are elucidated below.
Respondents were randomly allocated to one of the ten conditions.
Factor 1: type of precipitating crime
The first experimental factor is the type of precipitating crime, which was presented in Vignette 1. As described earlier, crime type can have a considerable impact
on how a crime situation and the people involved are judged by outsiders (e.g.,
Carlsmith et al. 2002). This experimental manipulation was operationalized by
constructing three versions of the precipitating crime: traffic aggression, a pedestrian
crash and a sex offense. The consequences for the victim of the precipitating crime
were purposely non-fatal in order to avoid reaching a ceiling effect of support.
In the traffic aggression version, a 39-year-old male cyclist is cut off in traffic by a
car driver. The cyclist reacts to this by raising his fists to the driver, after which the
latter purposely pushes him off the road. This causes the cyclist to fall off his bicycle,
resulting in a broken arm, a broken leg, bruised ribs and a concussion. He is taken to a
hospital. The driver is apprehended by the police and found to have a blood-alcohol
level of twice the legal limit.
The pedestrian crash version describes a 9-year-old girl who is hit by a speeding
car while walking her bicycle across a pedestrian crossing. The driver is arrested by
the police and is found to be intoxicated at the same level as the driver in the traffic
aggression version. The girls injuries are exactly the same as those suffered by the
male victim in the traffic aggression version, and she is also taken to a hospital.
3
In the sex offense condition, there are only two sentencing variations: a lenient and a severe one. The
reason for this is that we piloted this study using the two traffic offenses, and found relatively low levels of
support for vigilantism. As an extra check we therefore included a particularly heinous precipitating event
(child molestation) in the current study, but only operationalized two out of four sentences due to practical
limitations.
Traffic aggression
Pedestrian crash
Acquittal
Lenient
Normal
Severe
Sex offense
x
x
The third version of Vignette 1 concerns child molestation. The victim is a 9-year-old
girl who is pulled off her bicycle while on her way home from school. After she falls to
the ground, a man sexually assaults her. In the struggle that follows, she is able to get
away, and the man is later apprehended by the police.
The English translation of all three versions of Vignette 1 can be found in
Appendix 1. Apart from the type of precipitating crime, details were held constant
between the three vignettes as much as possible in order to avoid interference with the
experimental manipulation. However, it was challenging to vary crime type without
varying other situational aspects. We return to this in the discussion section.
Factor 2: sentence for the precipitating offender
The second experimental factor concerns the sentence for the precipitating offender,
and was manipulated in Vignette 2. The public tends to feel strongly about sentencing, and often perceives it to be too lenient (e.g., De Keijser et al. 2007, regarding the
Dutch). We expect that the type of sentence for the precipitating offender in the
vignette will affect to what extent the subsequent act of vigilantism is perceived as an
injustice. In our study, we distinguish four sentence levels: acquittal, a lenient, normal
and a severe sentence.4 The lenient sentence for the two traffic offenders (traffic
aggression and pedestrian crash) for instance consists of a 100 euro fine and a
suspended drivers license for 2 months. The severe version of the sentence is a
combination of 240h of community service, a 4-month suspended prison sentence
and a 2-year suspension of the drivers license. In case of acquittal, the precipitating
offender is acquitted by the judge due to a technicality (the wrong date in the
summons). Importantly, it is made clear that the precipitating offender is not released
because of a lack of evidence. The sentences for the two traffic offenders necessarily
differ from the sentences given to the sex offender due to the specific nature of the
offenses. The suspension of the drivers license is replaced by victim compensation in
the sex offense condition. Appendix 1 provides the English translation of selected
versions of Vignette 2.
Following the information about the sentence, the vignette reports that the sentenced precipitating offender was attacked by a vigilante. The vigilantism act is
identical in all conditions: a 39-year-old male vigilante goes to the house of the
4
These are our labels of the relative severity of the sentencing variations; respondents may have experienced the severity differently. In order to reinforce the suggestion of a lenient, normal or severe sentence,
the vignettes report what the Public Prosecution had demanded as a sentence for the precipitating offender.
precipitating offender and throws a brick through the front window. Following this,
the precipitating offender walks out his front door and gets beaten up by the vigilante.
The resulting injuries of the vigilantism victim (i.e., the precipitating offender) are
two broken teeth, a broken nose and contusions. This act of vigilantism matches our
definition. The identity of the vigilante differs slightly between the conditions. The
traffic aggression offender is attacked by the man he previously victimized (the
cyclist). In the other two conditions, the father of the victimized 9-year-old girl is
the one who takes the law into his own hands.
With the aim of inducing genuine reactions from our respondents, all vignettes
were made as realistic as possible by making them appear like articles from a popular
Dutch news website: www.nu.nl. A real crime scene picture from the website was
added to Vignette 1 to make it more salient.5 Each version contains the same picture
of a bicycle lying in an area between a road and a bicycle path, which matches all
three storylines.
Measures
Support for vigilantism
To measure support for vigilantism, we presented a wide range of items that pertain to
various sentiments. We based these on research in the field of social justice, which
has identified some of the most common emotional reactions to crime (e.g., Darley
and Pittman 2003). A 7-point Likert response format was used (10fully disagree; 70
fully agree). Items include I find it terrible that Ruben S. was beaten up and I feel
sympathy for Frank H.. Based on these items (k016), we constructed three distinct
measures of support: outrage at vigilantism (0.86), empathy with the vigilante
(0.84) and desired punishment for the vigilante (0.88).6 See Appendix 2 for the
items and summated scales.7
Confidence
In the literature on confidence, a distinction is commonly made between procedural
justice and effectiveness (e.g., Roberts and Hough 2005; Skogan 2009; Sunshine and
Tyler 2003a, 2003b). We used insights from both perspectives to create an elaborate
measure of confidence. It has additionally been shown that there is added value of
specifying confidence per criminal justice system agency, as it results in differential
confidence ratings (e.g., Allen et al. 2006; Roberts and Hough 2005). Our final
measure of confidence (k044) therefore combines items about a) procedural justice
5
The editors of www.nu.nl gave written permission to use the nu.nl format and picture, provided that
subjects would be informed about the fictitious nature of the articles. A disclaimer followed after the first
part of the study.
6
A factor analysis resulted in two rather than three factors, with an explained variance of 57 percent. We
nonetheless decided to construct three scales because the social justice literature commonly makes a
meaningful distinction between moral outrage, empathy and desire for punishment.
7
The items that were presented after Vignette 1 function as a manipulation check for the first experimental
factor (type of precipitating crime). These items are almost identical to those related to Vignette 2, except
that the former concern the precipitating crime instead of the vigilantism act. The responses to this first set
of items confirm that people were upset about the precipitating crime (analyses not reported here).
and effectiveness and b) police, public prosecution, judges and the criminal justice
system as a whole. Examples are Judges treat people fairly and The police are
effective in combating crime. A 7-point Likert response format was used. The items
were used to construct one overall summated scale: Confidence in the criminal justice
system (0.96).8 The items can be found in Appendix 2.
Hypotheses
Experimental factor 1: type of precipitating crime
Overall we expect that a more serious precipitating crime will lead to relatively more
positive reactions regarding the subsequent act of vigilantism; it will be seen as less of an
injustice. As child molesters tend to evoke very emotional reactions, we hypothesize that
when he becomes the victim of vigilantism, levels of outrage at vigilantism and
punishment desire will be lowest, and empathy with the vigilante highest, compared
to the other two precipitating crimes. The victimized cyclist who takes the law into his
own hands will probably induce the highest levels of outrage compared to the other two
vigilantes, as he can be seen as having partially provoked the traffic offense by raising
his fist to the driver. Empathy and punishment ratings will vary accordingly.
Experimental factor 2: sentence for the precipitating offender
We expect that the more adequate the formal reaction to the precipitating crime is
perceived to be, the more negative people will react to the subsequent act of
vigilantism. We predict that a severe sentence for the precipitating offender will lead
to relatively more outrage when he is also punished by a vigilante, as the latter event
will be perceived as more of an injustice than in the conditions where he is punished
more leniently (or even acquitted). Empathy and punishment desire will follow this
pattern. Vigilantism after acquittal will thus result in the lowest levels of outrage and
punishment desire, and highest level of empathy with the vigilante.
Confidence in the criminal justice system
In line with the confidence proposition, we expect a negative relation between
confidence in the justice system and support for vigilantism. People who have a
higher level of confidence are more likely to consider criminal justice procedures and
outcomes as legitimate and adequate, and will therefore react more negatively to
vigilantism. They will be more likely to see vigilantism as causing injustice rather
than justice. Confidence in the criminal justice system will thus be positively related
to outrage at vigilantism and desired punishment for the vigilante, and negatively to
empathy with the vigilante.
8
A factor analysis resulted in two factors, with one consisting of all items related to police. The results of
the regression analyses were highly similar when running them separately for these two confidence factors,
so we constructed one overall confidence scale for reasons of parsimony. We excluded five items that
loaded below .40 in the forced one-factor solution.
Findings
Table 2 shows the ratings on the three measures of support for vigilantism per type of
precipitating crime (experimental factor 1).
9
Funding for the establishment of the LISS panel was provided by the Netherlands Organization for Scientific
Research (NWO). For more information about the panel, see www.centerdata.nl and www.lissdata.nl.
10
The reference population is the Dutch speaking population that permanently resides in the Netherlands.
Children below 16 years of age are excluded. Households that did not have internet or computers available
were loaned equipment to provide access to the internet via a broadband connection. Panel members have
been presented with monthly online questionnaires on a variety of topics since October 2007. Completion
of a questionnaire takes about 20 to 30 minutes, and respondents are paid for each completed one.
11
Baruch (1999) compared the response rates of 175 different studies that were published in behavioral sciences
journals. He found the mean reported response rate to be 55.6 percent, with a median of 60. For conventional
populations, Baruch proposes response rates between 40 and 80 percent to be the acceptable norm.
12
Due to the random allocation of respondents, the non-response is divided equally over the experimental
conditions. Specific characteristics of the panel members who did not participate in the current study are not
available. Within the LISS panel in general, a considerable part of the monthly nonresponse is due to the
same panel members every month (Scherpenzeel and Vis 2010). Panel members who participated before
but have not completed a questionnaire for at least 3 months are defined as sleepers, which was the case for
13 percent of the LISS panel members in January 2010. The likelihood of becoming a sleeper is related
more to past response behavior than to demographic characteristics. Non-response in the LISS panel is
mainly due to technical reasons or personal circumstances; survey characteristics hardly play a role
(Scherpenzeel and Zandvliet 2010).
13
The educational levels of the sample are as follows: primary education (10 percent), pre-vocational
secondary education (27 percent), senior general secondary education, pre-university education or secondary vocational education (33 percent), and an associate or university degree (30 percent). With regards to
monthly net income, 11 percent of the sample had no income, 25 percent earned less than or equal to 1000
euros, 41 percent between 1001 and 2000 euros and 23 percent earned over 2000 euros. These statistics are
highly representative for the Dutch population, as are other sample characteristics such as ethnic background, household size and urbanity.
Precipitating crime
4.64 (1.25)
3.83 (1.37)
Mean (SD)
3.19 (1.29)
3.04 (1.20)
3.71 (1.35)
4.50 (1.27)
5.59 (1.36)
4.92 (1.55)
4.14 (1.68)
The levels of outrage at vigilantism are not very high, suggesting that respondents
were not particularly upset by the fact that the precipitating offender was assaulted.
Only in the traffic aggression condition does outrage exceed the midpoint of the scale.
It is also noticeable that, in the case of the two traffic offenses, desired punishment is
higher than outrage, and empathy is the lowest of all three measures of support. The
opposite is true for the sex offense. This implies that the specific nature of the sex
offense has a particular impact on peoples responses.
In Table 3, the mean ratings on the three support measures are presented per level
of the second experimental factor: the sentence level for the precipitating
offender.
We carried out a two-way analysis of variance (ANOVA) for each of the three
dependent variables to test for main effects and interactions (see Table 4). All main
Table 3 Support for vigilantism per sentence level (n01,930), scale 1-7
Scale (overall mean)
Outrage at vigilantism (4.01)
Sentence level
Mean (SD)
Acquittal (n0395)
3.81 (1.36)
Lenient (n0584)
3.88 (1.43)
Normal (n0360)
4.38 (1.39)
Severe (n 0591)
4.03 (1.43)
Acquittal (n0395)
3.74 (1.30)
Lenient (n0584)
3.78 (1.44)
Normal (n0360)
3.22 (1.30)
Severe (n 0591)
3.61 (1.40)
Acquittal (n0395)
4.84 (1.55)
Lenient (n0584)
4.85 (1.66)
Normal (n0360)
5.44 (1.46)
Severe (n 0591)
5.04 (1.63)
Sex offense was operationalized with only two sentencing levels (lenient and severe), so the acquittal and
normal conditions have relatively less respondents
SOURCE
df
Precipitating crime
181.35**
Sentence level
17.19**
.41
Precipitating crime
177.14**
Sentence level
14.22**
.63
Precipitating crime
126.21**
Sentence level
13.66**
.45
Outrage at vigilantism
* p<.05; ** p<.01
In order to examine the robustness
of our findings, we conducted
other analyses, but these did not
affect our conclusions
effects are significant. The two experimental factors (i.e., situational characteristics)
affect all three measures of support. This implies that public support for vigilantism
should not exclusively be attributed to a low level of confidence in the criminal
justice system. Interestingly, the impact of the type of precipitating crime is considerably larger than that of the sentence level. No interaction effects were found
between the experimental factors.
We will discuss these findings in more detail below, examining one measure
of support for vigilantism at a time. For each we first report outcomes of posthoc tests in which we more closely examine the differences between the
specific experimental conditions. We subsequently present the outcomes of
the OLS regression analyses in Tables 5, 6 and 7. These findings will show the
impact of confidence on support for vigilantism. We did not identify the two
experimental factors (type of precipitating crime and sentence level) separately in
these models, due to the fact that the sex offense only has two sentence levels and
thus cannot be compared directly to the other conditions. Instead, nine experimental
conditions were entered as one block of variables, and one condition (traffic aggression, acquittal) served as the reference category. The resulting beta values of the
experimental conditions thus need to be interpreted in relation to the reference
category.
We secondly will discuss the relative impact of situational characteristics and
confidence on support for vigilantism. The focus thereby is not on explained variance, but rather on the relative influence of situational characteristics and confidence.
For this reason we did not include demographic variables in the regression models.14
We calculated sheaf coefficients (Heise 1972) for both blocks of independent variables, which makes it possible to examine their relative impact on support. These
standardized measures range from 0 (no effect on the dependent variable) to 1
(a factor that explains all variance of the dependent variable). By comparing the
14
We did examine the influence of age, gender and educational level. The only noteworthy effect was that
of educational level on desired punishment (0.13, p<.01). This is in line with previous research (Tankebe
2009).
Constant
Traffic aggression acquittal
1.98 (.17)
Experimental conditions
Traffic aggression lenient
.47 (.13)
.10**
.53 (.13)
.11**
.56 (.13)
.12**
-.75 (.12)
-.17**
-.37 (.13)
-.08**
-.26 (.13)
-.05*
-.13 (.13)
-.03
1.12 (.13)
-.24**
-.99 (.12)
-.22**
.53 (.03)
.30**
Adj. R2
.26
Confidence
* p<.05; ** p<.01
Constant
Traffic aggression acquittal
5.25 (.17)
Experimental conditions
Traffic aggression lenient
-.30 (.13)
-.06*
-.43 (.13)
-.09**
-.43 (.13)
-.10**
.71 (.12)
.16**
.39 (.13)
.08**
.21 (.13)
.04
.09 (.13)
.02
1.22 (.13)
.27**
1.11 (.12)
.25**
-.45 (.03)
-.26**
Confidence
Confidence in the CJS
Adj. R2
* p<.05; ** p<.01
.24
Constant
Traffic aggression acquittal
2.80 (.20)
Experimental conditions
Traffic aggression lenient
.48 (.15)
.09**
.47 (.15)
.09**
.53 (.15)
.10**
-.66 (.14)
-.13**
-.33 (.15)
-.06*
-.02 (.15)
.00
-.05 (.15)
-.01
1.19 (.15)
-.22**
-.96 (.15)
-.19**
.56 (.04)
.29**
Adj. R2
.21
Confidence
* p<.05; ** p<.01
two sentence levels, t(405)01.19, p0.24. Outrage is low no matter at what level the
precipitating sex offender was sentenced.
In order to evaluate the confidence proposition, we next examined the outcome of
the regression analysis on outrage (see Table 5). We expected a positive relation
between confidence in the criminal justice system and outrage at vigilantism, which
was confirmed by our findings. We lastly calculated sheaf coefficients to examine the
relative impact of the experimental manipulation and confidence on outrage. The set
of experimental conditions has a larger sheaf coefficient (.42) than confidence (.30).
Outrage at vigilantism in the vignette is thus affected more by situational variation
than by confidence.
Empathy with vigilante
We now examine the second measure of support for vigilantism. We predicted
respondents to feel most empathy in the conditions where the vigilante attacks a
sex offender and least when a traffic aggression offender is assaulted. These expectations
were confirmed (see Table 2), and post-hoc analyses show all differences to be
significant at p<.001.
Concerning the effect of sentencing, the second experimental factor, we first examined the traffic aggression and pedestrian crash conditions. We expected people to feel
more empathy with the vigilante if his victim, the precipitating offender, was previously
acquitted than if he was sentenced. We found a main effect for sentencing (see Table 4).
Post-hoc analyses reveal that indeed the highest level of empathy, which was found in
the acquittal condition, differs significantly (p<.001) from each of the other three
sentencing conditions. A difference in empathy in the expected direction was also
found between the lenient and severe sentence conditions (p<.001), although it is not
particularly large. Differences between the other sentencing levels were not significant, so our expectations were only partially confirmed. In the sex offense conditions,
differences between the two sentence levels (lenient and severe) were not significant,
t(405) 0.99, p 0.33. The level of empathy with the vigilante was high in both
conditions, independent of the sentence level.
We next tested the confidence proposition by conducting a regression analysis on
empathy with the vigilante (see Table 6).
As expected, confidence in the criminal justice system had a substantive negative
impact on empathy. The sheaf coefficients reveal again that the situational factors
(.42) have a larger influence on empathy with the vigilante than confidence (.26).
Desired punishment for vigilante
The type of precipitating crime also had a substantial impact on our third support
measure: desired punishment. We expected that respondents would be least inclined
towards punishing the vigilante if his victim was a sex offender, and most likely to
find punishment necessary in the conditions where the vigilante assaulted the traffic
aggression offender. Our findings match this expectation (see Table 2), and post-hoc
analyses reveal all differences between the conditions to be significant at p<.001.
Concerning the second experimental factor (sentence level), we found a similar
pattern as before for the two traffic offense conditions. As expected, punishment desire
was lowest for the vigilante who victimized someone who had been acquitted by a
judge, in comparison to conditions where someone had been sentenced (see Table 3).
The difference in desired punishment between the traffic aggression and pedestrian
crash conditions was significant overall, F(3, 743)06.46, p<.001, but only between
acquittal and the other three sentencing types (all p<.01). Our expectations were
thereby partially confirmed, although it should be noted that the effects of sentencing
that we did find are not particularly large. Within the sex offense conditions, desired
punishment for the vigilante did not differ between the two sentence levels, t(405)0
1.56, p0.12.
Table 7 shows the outcome of our regression analysis on desired punishment. As
predicted, higher confidence is related to a stronger call for vigilante punishment.
Lastly, the sheaf coefficients reveal that the combined effect of the experimental
conditions (.36) on desired punishment is larger than that of confidence (.29).
Conclusions
Our findings reveal that public support for vigilantism is not necessarily rooted in a
lack of confidence in the criminal justice system: characteristics of the vigilantism
situation have a substantive and independent influence. When the public expresses
support for those who take the law into their own hands, this is not necessarily a sign
that the legitimacy of the criminal justice system is in jeopardy.
In line with the situation proposition, both experimental factors affected the three
measures of support for vigilantism (empathy, outrage and desired punishment). Situational characteristics thus affect how people judge an act of vigilantism. Especially the
type of precipitating crime had a strong impact. Empathy was for instance higher with the
vigilante who attacked a sex offender than with the two vigilantes who assaulted traffic
offenders. Punishment desire for the vigilante who attacked the sex offender was also
relatively low in comparison to the other vigilantes. Regarding the sentencing manipulation, (small) differences in support were mainly found when comparing the acquittal
condition to the other three sentence levels. Outrage at vigilantism was for instance
lowest when the precipitating offender had previously been acquitted by a judge.
However, in the conditions where he had been sentenced, the particular level of the
sentence did not affect the various measures of support. Respondents were thus mostly
influenced by whether the precipitating offender had been acquitted or sentenced
before becoming a victim of vigilantism, and not necessarily by the sentence level.
This adds an interesting dimension to current debates about public punitiveness;
perhaps citizens are more concerned with cases of acquittal than with overly lenient
sentencing. This may nonetheless be due to the fact that the acquittal in the current
study was the result of a mistake in the summons, as opposed to a normal type of
acquittal.
Within the sex offense conditions, the sentence did not have any impact at all on
the three measures of support. Whether the sex offender received a lenient or a severe
sentence did not affect peoples reactions to the subsequent act of vigilantism against
him. One explanation for this finding is the lack of an acquittal condition for the sex
offense: it was precisely in this condition that differences were found for the three
measures of support in comparison to the other sentence levels. Another possibility is
that the particular nature of the crime (child molestation) caused people to be more
supportive of vigilantism regardless of the level at which the sex offender was
sentenced. The average ratings of outrage and punishment desire were indeed lower
in the two sex offense conditions, and the level of empathy with the vigilante higher,
than in all of the traffic offense conditions.
We also found evidence for the confidence proposition. Confidence in the criminal
justice system affected all three measures of support for vigilantism. Respondents
with a higher level of confidence were more outraged at the act of vigilantism, felt
less empathy with the vigilante, and had a stronger desire for punishment of the
vigilante. Confidence in the criminal justice system therefore does provide a partial
explanation for support for vigilantism.
Using sheaf coefficients, we examined the relative impact of the experimental
factors (i.e., situational characteristics) and confidence. These analyses revealed that
all three measures of support for vigilantism were affected more by the situational
characteristics than by levels of confidence. This implies that when people judge an
act of vigilantism, they are reacting mainly to situational features.
Discussion
Before a further discussion of implications and suggestions for future research, there
are a few methodological matters to be addressed. Firstly, there were differences
between the vignettes that were not directly related to the experimental manipulation.
For instance, the victims of the pedestrian crash and sexual offense were 9-year-old
girls, while the traffic aggression victim was an adult male. Likewise, the vigilante in
the former two cases was the father of the girl victim, while in the latter case it was
the victim himself who assaulted the precipitating offender. These differences may
have affected peoples responses to the vignettes in addition to the impact of the two
experimental manipulations. Our main conclusion, namely that situational factors
have a considerable influence on support for vigilantism, is nonetheless not affected
by this.
Our respondents were not confronted with real crime victims. Nevertheless,
previous research has shown that the actual presence of a victim is not necessary to
induce emotional reactions; mental images of a victim are sufficient (cf. Hoffman
1990). Moreover, real-life situations of vigilantism are not that different from the one
in our study: people read newspaper articles about vigilantism and form their opinion
accordingly. The realistic appearance of our news articles adds to the external validity
of our results. Interestingly, in a space that was provided for extra remarks, some of
the respondents expressed having difficulty in deciding to what extent the offenders
deserved punishment because they had too little information. This suggests that it
may be easy for citizens to complain about the severity of a formal sentence as
described in a newspaper, but that coming up with ones own judgment is something
else (cf. De Keijser et al. 2007).
We found considerable differences between the three measures of support. The fact
that people empathize with a vigilante does not automatically imply that they believe
he should remain unpunished. Citizens seem much more nuanced in their judgments
than is often assumed. This is in line with recent findings regarding the impact of
criminal justice system is effective in combating crime (22) You can count on judges
to take decisions in the best interests of society (23) Judges know what is going on in
society (24) When a judge passes a lenient sentence, he will have a good reason for
doing so (25) You can count on the Public Prosecution to take decisions in the best
interests of society (26) The Public Prosecution does its job well (27) Citizens rights
are not adequately protected by the Public Prosecution (reverse coded) (28) The
police are trustworthy (29) The police are effective in combating crime (30) The
Dutch criminal justice system is trustworthy (31) I have respect for the Dutch
criminal justice system (32) The Dutch justice system succeeds at bringing criminals
to justice (33) Judges are trustworthy (34) Citizens rights are not adequately protected by judges (reverse coded) (35) The Public Prosecution treats people fairly (36)
I have respect for the Public Prosecution (37) The Dutch criminal justice system is
fair (38) You can count on the police to take decisions that are in the best interests of
society (39) The police do their job well.
References
Abrahams, R. (2002). Vigilantism, state jurisdiction and community morality: Control of crime and
'undesirable' behaviour when the state 'fails'. In I. Pardo (Ed.), Morals of legitimacy: Between agency
and system (pp. 107126). New York: Berghahn Books.
Adinkrah, M. (2005). Vigilante homicides in contemporary Ghana. Journal of Criminal Justice, 33, 413
427.
Alicke, M. D. (1992). Culpable causation. Journal of Personality and Social Psychology, 63(3), 368
378.
Allen, J., Edmonds, S., Patterson, A., & Smith, D. (2006). Policing and the criminal justice systempublic
confidence and perceptions: findings from the 2004/05 British Crime Survey (Online report 07/06).
London: Home Office.
Alvarez, A., & Bachman, R. (2007). Mob violence. Violence. The enduring problem (pp. 211236).
Newbury Park, California: Sage Publications, Inc.
Applegate, B. K., Cullen, F. T., Turner, M. G., & Sundt, J. L. (1996). Assessing public support for threestrikes-and-you're-out laws: global versus specific attitudes. Crime & Delinquency, 42(4), 517534.
Ayyildiz, E. (1995). When battered woman's syndrome does not go far enough: the battered woman as
vigilante. Journal of Gender and the Law, 4(141), 141166.
Baker, B. (2001). Taking the law into their own hands: Fighting crime in South Africa. Paper presented at
the ECPR, Grenoble.
Baruch, Y. (1999). Response rate in academic studies: a comparative analysis. Human Relations, 52(4),
421438.
Benesh, S. C., & Howell, S. E. (2001). Confidence in the courts: a comparison of users and non-users.
Behavioral Sciences and the Law, 19, 199214.
Bieneck, S. (2009). How adequate is the vignette technique as a research tool for psycho-legal research? In
M. E. Oswald, S. Bieneck, & J. Hupfeld-Heinemann (Eds.), Social psychology of punishment of crime
(pp. 255271). Chichester: Wiley.
Black, D. (1983). Crime as social control. American Sociological Review, 48(1), 3445.
Brown, R. M. (1975). Strain of violence. Historical studies of American violence and vigilantism. New
York: Oxford University Press.
Carlsmith, K. M. (2008). On justifying punishment: the discrepancy between words and actions. Social
Justice Research, 21, 119137.
Carlsmith, K. M., Darley, J. M., & Robinson, P. H. (2002). Why do we punish? Deterrence and just deserts
as motives for punishment. Journal of Personality and Social Psychology, 8(2), 284299.
Coffee, J. C. (1991). Does "unlawful" mean "criminal"? Reflections on the disappearing tort/crime
distinction in American law. Boston University Law Review, 71, 193246.
Cohen, R. L. (1989). The legitimacy of vigilanteism. Brigham Young University Law Review, 4, 1261
1276.