Reducing youth violence related to student parties: Findings from a community intervention project in Stockholm
Centre for Psychiatry Research, Karolinska Institutet/Stockholm County Council Health Care Provision, Stockholm, Sweden
Correspondence to Dr Mats Ramstedt, Centre for Psychiatry Research, Karolinska Institutet/Stockholm County Council Health Care Provision, Teknologgatan 8E, 113 60 Stockholm, Sweden. Tel: +46 8 12345502; Fax +46 8 12345501; E-mail: firstname.lastname@example.org
During the spring of 2007, the police reported a marked increase in violence and binge drinking related to high school student graduation parties on weekday nights at restaurants in Stockholm city. This spurred a multi-component community intervention project to reduce these problems.
This study aims to evaluate the impact of the intervention on youth-related violence on weekday nights in 2008–2010.
Design and Method
The outcome measure entailed the number of violence-related emergency room visits on weekday nights (10:00 pm–6:00 am) by adolescents aged 18–20 years. The study period was 1 April–31 May, which is when most student graduation parties took place. The data covered the years 2005–2010, with three data points before the intervention, and three after the intervention was introduced. Because the intervention was expected to apply to weekdays only, the control series involved a corresponding indicator pertaining to weekend nights (10:00 pm–6:00 am). The intervention effect was assessed by means of difference-in-differences estimation.
The estimated intervention effect according to the difference-in-differences estimation models was a statistically significant 23% reduction of violence among young people.
Discussion and Conclusion
This type of intervention is a promising measure of preventing youth violence and deserves to be continued. Such continuation would also provide additional data required for a more conclusive assessment.[Ramstedt M, Leifman H, Müller D, Sundin E, Norström T. Reducing youth violence related to student parties: Findings from a community intervention project in Stockholm. Drug Alcohol Rev 2013;32:561–565]
In recent years, it has been increasingly popular among high school students in Stockholm to throw graduation parties in inner city restaurants, often with help from designated ‘events’ companies. Most of these parties—approximately 90% of them—are held on weekdays, when most restaurants have fewer patrons and some would otherwise keep their doors closed. Many restaurants clearly see weekday student parties as an opportunity to raise their revenues.
During the spring of 2007, the police in Stockholm reported a marked increase in binge drinking and violence related to these graduation parties. In response, a community intervention project, ‘Student08’, was set up by Stockholm Prevents Alcohol and Drug Problems (STAD) and the police. The project was inspired by the community intervention approach developed by Holder  and included several parts: co-operation, control/enforcement and information/education. Since the first year, the project has continued annually under ‘Student09’, ‘Student10’, etc.
The link between alcohol and violence is well established . There are also indications that the link is particularly strong in younger age groups, where both drinking and rates of violence are high. In Sweden, studies suggest that 9% of 16- to 24-year-old men had been victims of violence during the last 12 months compared with 3% in the age group 25–44 years. Corresponding figures for women were 6% and 2%, respectively . This age pattern is also confirmed in hospitalisation data . What is more, drinking and binge drinking in particular peak among the 16- to 24-year-olds in Sweden .
That a multi-component intervention on drinking environments may be effective in reducing binge drinking and alcohol-related violence is supported by several studies. One example is the STAD program in Stockholm, Sweden, which focused on community mobilisation, training in responsible beverage service as well as on stricter enforcement of existing alcohol laws. Evaluations of this program have found significant reductions in night-time violent crime [6-8]. Another example is the safety action program in Queensland, Australia, which targeted the environment of licensed venues, including drinking regulations as well as the social and physical environments. Major reductions in aggression and violence have been observed in the assessments of this program . Furthermore, studies by Holder et al.  from the USA (California) have similarly demonstrated a significant reduction in drinking and violence as a result from similar multi-component interventions.
While the Swedish ‘Student08’ intervention program was influenced by previous community interventions, it differed from former interventions in significant respects. This arena or, more precisely, student parties at licensed premises, and this target group of high school graduates have not been subject to community interventions before. Our study is the first to assess the effects of a community intervention on this type of arena and for this particular target group.
Drawing on the experiences of the first three years of the project in 2008–2010, this study seeks to evaluate the overall impact of these interventions in Stockholm. More specifically, we will test whether the intervention was followed by a reduction in youth violence during the time period when student graduation parties were taking place on weekday nights in April and May.
The main goal of the intervention was to decrease the number of violent crimes related to alcohol consumption in connection with student parties at licensed premises in Stockholm. Furthermore, a long-term goal was to gain knowledge about restaurant regulations and sustainable norms among future restaurant visitors, the students themselves. The intervention consisted of three key components: (i) co-operation; (ii) information/education; and (iii) increased enforcement. Each component is presented in more detail in Table 1.
Table 1. Overview of the student community intervention project in Stockholm
Direct phone number and e-mail address to the police working at student parties
Restaurant owners and guards/doormen
Feedback to restaurant after inspections
Police, the Licensing board and The Swedish Tax Agency
Mapping of student parties
Restaurant owners, event companies
Meetings with restaurant owners and event companies
Restaurant owners, event companies
STAD, police, the Licensing board and The Swedish Tax Agency
Information cards with restaurant regulations
Brochures with information to students
Brochures with information to parents
Poster to high schools
Posters to restaurants
Guidebook; adapting the party for high school students
Webpage with information
Two visits/inspections from the police at each party
Students and restaurants owners
Licensing board and The Swedish Tax Agency
Control of areas close to the parties; public transportation, fast food restaurants and parks
The following authorities in Stockholm participated in the intervention; STAD (the county council), the city police, the licensing board (the municipality) and the Swedish tax agency. In addition to these authorities, restaurants and events companies also took part. The authorities, restaurant owners and events companies would meet before the start of the season of student parties. A direct telephone number and an email address to a police unit working solely on student parties were distributed to restaurant owners and events companies, which could then call the police at any time for support in case of violence or other crime.
Information about restaurant regulations was spread to students through training evenings, information cards, brochures and posters. At the training evenings, the police and a representative from the restaurant business gave lectures about the alcohol law—what rights and obligations the students had –and some practical tips to consider before, during and after the party. Moreover, written information about restaurant regulations was further distributed to students through events companies and schools. The police also had brochures delivered to parents. A guidebook on how to prepare and adapt the party to high school students was distributed to restaurant owners. The guidelines were further discussed at action group meetings.
The events companies and restaurant owners reported all parties to the police, documenting the date, the name of the restaurant and the name/orientation of the school. As a result, the police could plan the enforcement/visits that took place at least twice for each student party. And finally, the licence board and the Swedish tax agency increased their inspections at the student parties.
Data and methods
As an indicator of youth violence, we used violence-related emergency room visits by young people aged 18–20 years during weekday nights (when these student graduation parties predominantly take place). The data were obtained directly from the five major hospitals in Stockholm, those of Danderyd, Huddinge, S:t Göran, Söder and Karolinska. The data included the date and exact time of the violence-related visit, which allowed for dividing the data into weekday versus weekends and also to select cases that had occurred at night (10:00 pm–6:00 am). The outcome measure thus comprised the number of violence-related emergency room visits on weekday nights during the period 1 April–31 May, which is when most of the student graduation parties take place. The data covered the years 2005–2010, which yields three data points before the intervention, and three after the intervention was introduced. As the intervention was confined to a specific time, we applied a quasi-experimental design. More specifically, because the intervention was expected to have effect on weekdays only, the control series comprised the corresponding indicator pertaining to weekend nights. The intervention effect was assessed by means of difference-in-differences estimation (DiD) . This technique is often used by economists evaluating policy changes that have been implemented in a quasi-experimental fashion . The DiD model was specified as follows:
where V is the violence indicator with the observations for weekdays and weekends pooled into one single series. E is a dummy variable that captures possible differences between the experiment group (weekdays) and the control group (weekends) before the intervention. It takes the value 1 for the experiment group and 0 for the control group. T is a time period dummy that captures factors that would affect violence also in the absence of the intervention. It takes the value 1 in the intervention period and 0 in the pre-intervention period. E*T is the intervention variable; it thus takes the value 1 only in the experiment group during the intervention period, and 0 otherwise. The parameter of interest is thus b3, which captures the intervention effect. The percentage intervention effect is obtained from the expression: (exp [b3]–1) * 100. We used robust standard errors, which yield more conservative levels of significance.
As a test of robustness, we estimated two models for pseudo-experiment groups, that is, for two periods when no intervention effect should be expected. The first one was weekday nights during the period 2 January–31 March, and the second one weekday nights during the period 1 June–30 September. As above, the control series comprised the corresponding indicator pertaining to weekend nights.
Violence-related emergency room visits
Figure 1 indicates that the violence level was markedly lower in the experiment group during the intervention period compared with the pre-intervention period. No corresponding pattern is discerned in the control group (Figure 1) or in the pseudo-experiment groups (Figures 2 and 3).
The estimated DiD models confirm the impression of this graphical evidence. The estimated intervention effect (Table 2), corresponding to a reduction of 23%, is thus statistically significant (P = 0.027). In contrast, the estimated intervention effects in the pseudo-experiment groups are non-significant.
Table 2. Difference-in-differences estimates of intervention effect on violence in experiment group
This paper has sought to evaluate the multi-component community intervention program that was launched in 2008 in Stockholm to reduce problems with binge drinking and violence at inner-city high school student graduation parties. In particular, we wanted to test whether violence was reduced among students as a result of the intervention. We used violence-related emergency room visits as outcome measure. The findings suggest that the intervention was associated with a 23% decrease in violence.
We have not identified any other evaluation of a similar intervention program addressing violence related to student parties. However, evaluations of other kinds of multi-component programs have found evidence of reductions both in violence  and in heavy drinking among young people . For instance, depending on which statistical model was applied, night-time violent crime was found to have been reduced by between 21% and 32% in the evaluations of another multi-component Responsible Beverage Service (RBS) program launched in Stockholm between 1998 and 2001 [6, 7]. Thus, although the interventions and the target groups were different, our result is within the range of effects found for similar programs in Stockholm.
The results should be viewed in light of certain limitations. First, we must acknowledge that violence-related emergency room visits by youth may not be alcohol-related or by graduation party attendees. However, given that approximately 70% of boys and practically all girls among those ending up in the emergency rooms were intoxicated according to reports from one of the included hospitals (Södersjukhuset) , it is likely that a clear majority of the cases are in fact alcohol-related. Furthermore, it is not common that young people in Sweden drink alcohol on weekdays, which also suggests that a substantial proportion of alcohol-related violence in this age group is connected to these parties, although the exact numbers are missing. Still, a crucial assumption of the analysis is that the number of emergency room visits is a valid indicator of changes in violence related to the graduation parties. Furthermore, the number of data points in the statistical analysis is on the low side, suggesting that caution is needed in the interpretation of the results.
On the other hand, the results are strengthened by the outcome of the robustness tests, which suggest that the apparent intervention effect is not artificially produced by a general downward trend in violence during the intervention period. It should also be noted that there was an increasing number of young people celebrating their graduation at parties in the inner city during the study period, which rather would have pushed the number of violent incidents up during the study period and not, as our analysis found, down to a significant degree. The number of young people in the age group 18–20 years increased by 25% in Stockholm during the study period, from roughly 63 000 in 2005 to 79 000 in 2010. In addition, the number of student parties increased during follow-up—from 630 parties in 2008 to 650 and 710 parties in 2009 and 2010, respectively. This fact indicates that the effect may have been even stronger than is suggested by our results.
In conclusion, the evaluation suggests that this type of intervention is a promising measure for preventing violence and deserves to be continued. A continuation of the program would also provide additional data required for a more conclusive assessment of the effects.
This study was funded by grants from the Swedish Council for Working Life and Social Research (FAS, grant number 2009-170) and from the Swedish National Institute of Public Health.