Criminology In Europe

Newsletter of the European Society of Criminology

The International Crime Victims Survey

Back to 2012/3| Issue Download this issue in .PDF
  • Jan van Dijk

Latest Results and Prospects

Introductory remarks

The International Crime Victims Survey (ICVS), initiated in 1988, has developed into one of the largest projects in international criminology now covering 80 different countries. Its results are acknowledged as an important source of comparative information on levels of crime (Lynch 1996). In 2005 the European Commission (EC) co-sponsored the work of the ICVS in the 15 older member states of the European Union (van Dijk, Van Kesteren, and Manchin 2007). The results were combined with those of the fifth round of the ICVS, covering 31 nations altogether (van Dijk et al. 2007). In 2008–2009 the EC started preparatory work for a full-fledged EU-wide victimization survey to be executed by Eurostat. Consultancy work was first contracted out to European Institute for Crime Prevention and Control (HEUNI) (Aromaa et al. 2007) and subsequently to Tilburg University and Lausanne University respectively (van Dijk et al. 2010). Fieldwork was scheduled to take place in 2013 but has been postponed. In conjunction, the EC sponsored a round of national surveys in 2009 using a reduced version of the ICVS 2005, with a view of testing new methods of data collection—notably web-based interviewing. These pilots were duly executed in 2010, under the auspices of the International Government Research Directors, in Canada, Denmark, England/Wales, Germany, The Netherlands, and Sweden. Preliminary results were published by the Netherland`s Institute for Urban Research and Practice (NICIS) which was supervising the data collection (Ghauharali, Meuldijk, and Smit 2010).

Independent of these six EU-sponsored surveys in 2010, ICVS-based surveys were also conducted in Azerbaijan (2011), Estonia (2009), Georgia (2010 and 2011), Moldova (2011), Switzerland (2010), and Tajikistan (2011). Findings from a new round of ICSV-based surveys are available from 12 nations1. Results from 10 countries can be compared with results from one or more earlier surveys, allowing trend analyses. In this article we will briefly describe the methodology of these surveys, present some key results and provide a brief interpretation of emerging international trends in crime. In the final section we will discuss evolving plans for repeating ICVS, especially in Europe.

Methodology

The most important follow-up to the 2005 round of the ICVS was a series of pilot studies in six Western nations using a reduced version of the questionnaire. A first analysis of the results revealed that most computer-assisted web interviewing (CAWI)-based surveys had achieved very low response rates. Also, internet-based surveys showed much higher victimization prevalence rates than surveys using computer-assisted telephone interviewing (CATI). These challenging results were discussed at two expert seminars in 2010, convened by the two polling companies involved and the Max Planck Institute for Foreign and International Criminal Law respectively.

Parallel to these six surveys, Estonia, one of the countries which adopted the ICVS as its national survey, repeated the survey in 2008–2009 using face-to-face interviewing, as had been done in previous years (this time with an enlarged sample of 4,000). In Georgia the ICVS, which was previously conducted in 1992 and 1996, was repeated in both 2010 and 2011 with samples of 3,000, using face-to-face interviewing, with funding from the European Commission. In Switzerland, the ICVS was conducted in 2010 with a multi-mode approach, combining a postal questionnaire with internet-based interviewing and, to a lesser extent, telephone interviewing. Finally, an abridged version of the ICVS, designed by the author and his Georgian partners, was conducted face-to-face as part of omnibus surveys among national samples in Azerbaijan, Tajikistan, and Moldova in 2011.

The results of the six 2010 pilots, funded by the EC, were, in some respects disappointing. The first complication was that response rates for the CAWI-mode samples were exceedingly low, especially when the sample was drawn from a register (e.g. 3 percent in Canada). Better response rates were obtained when respondents were selected from panels comprising of persons who regularly agreed to be interviewed over the internet. However, whether such panels are fully representative of the population is far from assured. The second complication was, as mentioned, that the findings of the CAWI studies differed systematically from those of the studies using CATI. Victimization prevalence rates for 10 crimes together were approximately 5 percentage points higher in the CAWI studies than in the CATI studies in five of the six countries. The CAWI studies also showed somewhat higher rates on two “fear of crime” items. These findings are in line with other crime surveys testing mixed mode designs. Victimization rates were found to be significantly higher when using CAWI in Dutch, Finnish, and Belgian victimization surveys (van Dijk et al. 2010). Whether, or to what extent, the higher rates of CAWI studies are caused by undercoverage and non-response bias, by pure mode effects, or both, remained an open question.

To address this issue a large-scale experiment was conducted in The Netherlands with the purpose of disentangling mode-specific selection effects (caused by undercoverage and response bias) and pure measurement bias in victimization surveys. In their experiment Buelens et al. (2012) used four modes, computer-assisted personal interviewing (CAPI), CATI, CAWI, and paper. The results show that after reweighting for demographics, victimization prevalence rates are indeed higher in CAWI interviews relative to those of CAPI interviews. In contrast, prevalence rates were found to be lower in CATI interviews and paper-based questionnaires relative to those of CAPI interviews. These differences appeared to be almost completely caused by pure mode or measurement effects (different responses by the respondents), and only marginally by selection (undercoverage and/or non-response). As possible explanation of the measurement bias in CAWI interviews, the authors mention that respondents may tend to rush the completion of web-based interviews. In the 2010 EC funded pilots, CAWI-based interviews were indeed shorter than the CATI-interviews (Ghauharali, Meuldijk, and Smit 2010).

In earlier publications on the ICVS the authors reviewed existing evidence of mode-related biases in crime surveying. Their conclusion was that CAPI and CATI interviews produce roughly similar results on the condition that the CATI interviews are executed by competent interviewers (van Dijk et al. 2007; see also Killias, Kuhn, and Aebi 2011). The new Dutch experiment suggests that rates from high quality CATI-based interviews might be somewhat deflated compared to those of CAPI. Comparisons between CATI and CAWI-based results seem to be more problematic, since these tend to be distorted in opposite directions, the first downwards and the second upwards. The Dutch Central Bureau of Statistics has therefore decided to use in its 2012 national victimization survey a mixed mode consisting of CAWI and paper, excluding CATI.

Considering the above mentioned experimental findings, we will in the present overview disregard the results of the CAWI-based studies in the six countries participating in the EU pilots and include the CATI-based results only. These results are based on telephone-interviewing of representative samples of 2,000 per country. Response rates were reasonably high (Ghauharali et al. 2010). Results are therefore comparable with the results from previous ICVS studies in these countries. In Switzerland the 2011 survey was carried out with a mixed mode of CAWI and CATI, with the interviews equally split between these two modes (Killias et al. 2011). The response rate was 54 percent. In relatively less-developed countries, face-to-face interviewing remains the preferred mode since telephone penetration is often still insufficient and costs of interviewing are low. In Estonia the survey was carried out in 2008–2009 under the aegis of the Estonian Statistical Office, using CAPI, with a sample size of 4,000 (Ahven et al. 2009). The response rate was 58 percent. The survey in Georgia was conducted using face-to-face interviewing in 2010 and in 2011, with samples of 3,000 each (van Dijk and Chanturia 2011). Response rates were above 80 percent. Fieldwork was conducted by a local polling company, Georgian Opinion Research Business International (GORBI), which was also responsible for previous rounds of the Georgian ICVS surveys (1992 and 1996 [Pachulia 1996]). As mentioned, an abridged version of the ICVS, consisting of no more than 10 questions, was conducted in 2011 in Azerbaijan, Moldova, and Tajikistan by regional consortium partners of GORBI. The samples were 1,000 per country and interviews were conducted face-to-face. These findings have not previously been published.

Considering the outcome of the Dutch experiment, the results of the 12 national studies must be compared with due caution. Results from the six CATI-based surveys on victimization prevalence might be somewhat deflated compared to the results of the other surveys using either face-to-face interviews or, as in the case of Switzerland, a mixed-mode combining CATI and CAWI. On the positive side, all 12 surveys were supervised by experts who have been involved in the ICVS program for many years (Martin Killias, Andri Ahven, Merab Pachulia, Paul Smit, and Jan van Dijk).

Results on victimization

In this section we will first present the one-year victimization prevalence rates for five types of crime from 2009 or 2010. Next we will present trend data on the same rates. The selection of the five crime types is determined by their inclusion in the abridged version of the ICVS used in Azerbaijan, Moldova, and Tajikistan. For more background information on the ICVS methodology and calculation of rates, consult van Dijk et al. (2007). In the subsequent section some survey-based data on police performance will be presented.

51

The results presented in Table 1 show that victimization by threats or assaults is the most common form of victimization in most countries, more so even than victimization by personal theft (not included here). Victimization by car thefts is relatively rare. As shown below, these features indicate a distinct shift in the nature of victimizations. In terms of perceived seriousness, the crime situation seems to have deteriorated. Intercountry differences among the 12 countries are fairly large. For victimization by common crimes Switzerland, Estonia, Denmark, and Moldova stand at the top, and Azerbaijan, Georgia,2 and Tajikistan at the bottom. The relationship between levels of victimization is inversely correlated with GDP. Low prevalence in Azerbaijan and Tajikistan seem related to the largely rural nature of these countries.

For victimization by bribe-seeking, Moldova stands at the top, with Azerbaijan in second place. In Moldova bribes were often sought not just by public officials but also by medical staff and teachers. Corruption is, as known from previous ICVS surveys, most prevalent in countries in transition. Georgia and Tajikistan are exceptions to this rule. In Georgia the rate was very high in the recent past. The 1996 rate was the highest rate ever observed in the ICVS, with the exception of Albania (van Dijk 2008). The rate of experiences with bribe-seeking officials in Georgia went down from 22 percent in 1996 to less than 1 percent in 2010.

Car theft prevalence is highest in the most developed countries, notably Canada and the UK. Rates are medium–high on the European continent and lowest in Eastern Europe. This distribution is partly explained by differential car ownership rates, and the extent of car security (mandatory inmobilisers were first introduced in Germany). Regarding household burglary, Denmark and Estonia stand out with the highest rates. Medium−high rates are found in Switzerland, Moldova, and the UK. Rates in Azerbaijan and Tajikistan are lowest. Previous analyses have shown that burglary victimization is a good predictor of over all levels of common crime. Rates of robbery victimization are the highest in Estonia.

The rates of threats–assaults show that, among these 12 countries, victimization is higher in the more affluent countries. The highest rates were measured in The Netherlands, Switzerland, Sweden, UK, Denmark, and Germany. In previous analyses of ICVS results, a positive relationship was established between levels of common violence and the consumption of beer per capita (van Dijk 2008).

In Moldova and Tajikistan the survey was conducted for the first time in 2011. In Azerbaijan the ICVS was conducted once before, in 2000 in the capital city, Baku. The 2000 victimization rates in Baku, Azerbaijan were exceptionally low. In fact, the overall victimization rate for 10 crimes of Azerbaijan was the lowest of all 78 countries that participated in the ICVS up to 2005 (van Dijk 2008). The results of the 2011 survey show that in spite of an oil-related economic boom, levels of crime have not risen. In Estonia and Georgia levels of crime have boomed in the early stages of their transition to market economies and democracy. In Azerbaijan as in Tajikistan levels of crime seem to have remained at the level of former soviet times.

Many countries seem to have experienced booms in common crime which peaked around 2000. The boom has been particularly pronounced in countries in transition, here represented by Estonia and Georgia. After 2000, levels of crime have declined almost everywhere. There are two important exceptions to this international trend in crime. In Georgia levels of crime continued to rise up to 2005 or even later. The crime drop observed internationally seems to have arrived in Georgia with a delay of several years. Since then, crime in Georgia has dropped with a vengeance, transforming the country from one of the most crime-ridden to one of the safest in the region. The explanation for this special trajectory can be found in its prolonged transition which was interrupted by strife and civil war, instigated by Russia. From 2006 onwards the new administration under President Michael Saakashvili has effectively combated crime, including organized crime (the Thieves in Law) and petty corruption.

The trend in crime in Switzerland is the second exception to the general pattern. Between 2005 and 2010 levels of common crime, such as burglary and threats–assault, have gone up. This recent trend has transformed Switzerland from a low-crime into a high-crime country. We will discuss some possible explanations below.

In tables 2 and 4 available trend data on victimization by burglary and threats–assault respectively, are presented.

52

Only Canada and The Netherlands have participated in all six ICVS rounds. Their trends in burglary victimization are curvilinear with a peak in 1996 and a very steep decline thereafter. The UK and Sweden show a similar curvilinear trend. As discussed above, Georgia experienced a delayed curvilinear trend due to prolonged civil strife around 2000. The trends in Denmark, Estonia, and Switzerland seem out-of-tune with those elsewhere. In Switzerland burglaries have trended upwards since 1989, especially in recent years. In Denmark the level also increased between 2005 and 2010. In Estonia the level has remained more-or-less stable at a relatively high level.

Elsewhere we have demonstrated that trends in burglary in the six pilot countries between 2005 and 2010 are related to the level of anti-burglary measures (special locks and burglar alarms) in 2005 (van Dijk and Vollaard 2012). When data from Estonia and Switzerland are added to the analysis, the relationship between home security and trends in burglary is even more evident. Levels of home security are by far the highest in The Netherlands and the UK, where burglary rates have dropped, and the lowest in Estonia, Switzerland, and Denmark where burglary rates have risen3 (see table 3 for results). The fact that the UK and The Netherlands resemble Denmark and Switzerland on most economic and social indicators makes the relationship the more telling.

57

The relationship between the two security variables and the trends in burglary is statistically strong. When the two measures of home security are added up, the correlation between level of home security in 2004 and the trends in burglary victimization between 2005 and 2010 is near perfect (r= +.88; n=8; p< 0.002)4. In other words, trends in burglary at the national-level between t1 and t2 appear to be co-determined by levels of home security at t1. Figure 1 visualizes this relationship.

56

In the UK home security has been promoted by the police through its Secured by Design certification program. Similar initiatives have been taken in The Netherlands culminating in the incorporation of mandatory home security for new houses in the Building Regulations of 1999. Research has shown that burglary rates dropped significantly in neighborhoods with newly built houses without spatial displacement (Vollaard and van Ours 2010). The promotion of home security by central and local government in countries such as The Netherlands and the UK seems in the longer term to have paid off in terms of reductions in crime.

In Denmark, Estonia, and Switzerland the public is aware of their high burglary risks. In Denmark the percentage of the public thinking it likely or very likely to be burgled in the coming 12 months jumped from 14 percent in 2005 to 32 percent in 2010. In Switzerland this percentage was 25 percent in 2010. The percentages of people worried about burglary are now much higher in Denmark and Switzerland than elsewhere in Western Europe (the mean in Germany, The Netherlands, UK, and Sweden was below 15 percent in 2010). This increased risk awareness will by itself stimulate the use of home security. The economic case for the governments of Denmark, Estonia, and Switzerland to promote elementary home security in their countries is nevertheless strong. In these countries elementary home security is clearly underused. Investments in home security would be economically beneficial5.

Trends in victimization by threats–assaults are different from those of burglary. In none of the 12 countries except Georgia and Estonia have levels of common violence gone down over the full period. Especially noteworthy is the upward swing between 2005 and 2010 in several countries (see table 4 for results).

58

Elsewhere we have argued that the significant increases in police-recorded crimes of violence in North America and Western Europe over the past 15 years result from greater detection efforts and improved recording, rather than from an increase in real crime. There is evidence from several countries that this has indeed been the case. In my view the drop in the volume of crime has freed-up resources allowing police forces to more actively respond to less serious crimes of violence which would have remained undetected and unrecorded in the past (van Dijk 2012). Having said this, the results of the latest round of ICVS surveys point to a surge in violent crime over the past 5 years in at least some countries, most notably Switzerland (Killias et al. 2011; Killias and Lanfranconi 2012; Aebi and Linde 2012). I agree with my Swiss colleagues that the most likely explanation for the recent boom in violent crime is the increased participation of younger people in nightlife and the related increase of alcohol abuse. A reduction of the emerging problems of violent crime seems within reach for governments committing themselves to a more stringent regulation of these industries.

Attitudes towards the Police

An important indicator of police performance is the percentage of criminal victimizations reported to the police by the victims. Low reporting rates point to a lack of trust in the skills and integrity of the police. Reporting rates tend, for example, to be very low in many Latin American countries (van Dijk 2008). Since reporting rates vary by type of crime, we present reporting rates of victims of burglary only. A second ICVS-based indicator of police performance is the level of satisfaction of reporting victims with their treatment by the police. In Table 5 the satisfaction rates of victims who did report to the police are presented. Finally, all respondents are asked to rate the performance of the police in their neighborhood on a four-point scale (“Taking everything into account, how good of a job do you think the police do in your area controlling crime?”). Table 5 shows the scores of these three police performance indicators in the 12 countries. The results reveal a gap in police performance between Eastern European and South Caucasian countries on the one hand and Western European countries and Canada on the other. In the first group of countries victims are less inclined to report to the police, and those who do are less satisfied6. This is also true for other types of crime. In spite of various EU and Council of Europe directives and recommendations, police forces in these former communist countries have apparently not managed to make their victim reception practices more victim-friendly. Further police reforms in Eastern Europe and the Southern Caucasus focusing on victim reception seem to be called for. The scores on victim satisfaction for Azerbaijan and Tajikistan are missing because of the small numbers of victims identified.

59

The public’s assessment of the overall performance of the police tends also to be less favorable in ex-communist countries, with Georgia as a notable exception. The lowest rating is given by the public in Moldova. This is in line with the relatively high levels of victimization. The positive evaluation by the public of the Swiss police is somewhat surprising, considering the steep rises in crime in the country.

In Western countries victims are more satisfied with their treatment by the police but their satisfaction has not improved since 2000 (see also van Dijk et al. 2007). Apparently the EU Framework Decision of 2002 on the position of crime victims in criminal procedure, recently updated and upgraded into a Directive, has so far failed to improve police practices as perceived by reporting victims. It is possible police forces have relegated victim reception to specialized support agencies and/or have streamlined reporting processes, making them less instead of more victim-friendly. Another possibility is that the new victim policies have raised expectations which are not fulfilled. In the ICVS 2010 surveys conducted in Denmark, Germany, The Netherlands, Sweden, and the UK reporting victims were asked whether the police had kept them informed about their case, as they should according to EU legislation. Of the victims of household burglary only 43 percent had been kept informed (n= 373). Of those who had not, 59 percent said they would have appreciated follow-up information. The provision of follow-up information was lowest in Denmark and The Netherlands. The implementation of relevant EU legislation and of national legislation or guidelines on victims’ rights clearly still leaves much to be desired.

Conclusions and Outlook

The sixth round of the ICVS was partly coordinated by the International Government Research Directors Network (IGRD). Surveys in six countries were independently supervised. No comprehensive report on the results of these surveys has so far been published. Of the 12 surveys, 11 were conducted with either CATI (in the West) or face-to-face interviewing (in the East). The Swiss survey combined CATI with CAWI. Results should be compared with some caution. The relatively high rates of Estonia might partly be the result of face-to-face interviewing. This is not the case with the high rates of Denmark, though. The relatively high Swiss rates might also have been somewhat inflated, in this case by the use of CAWI with half of the sample.

With the exception of Switzerland, the modes of data collection have largely been the same as in previous rounds, CATI in the most developed countries and face-to-face everywhere else. In this respect the ICVS compares favorably with some of the national victimization surveys such as the NCVS of the USA, the British Crime Survey, and the ever-changing Dutch national surveys. These national surveys have all been repeatedly and fundamentally redesigned, compromising comparability of results over time. The ICVS-based crime trends are based on a methodology which remained the same for over 20 years. Although the sample sizes of the ICVS are relatively small—typically 2,000 per country—the results are indeed remarkably consistent over time.

Funding of the ICVS on a global scale remains a huge challenge. On the positive side, web-based modes of data collection hold the promise of conducting international surveys in the future with much lower budgets. In this respects the results of the surveys in Azerbaijan, Moldova, and Tajikistan are also encouraging. They demonstrate the potential of assessing the crime situation in a country with an abridged version of the ICVS, taking no more than three minutes of interview time per respondent. For analyses of secondary characteristics of victimization experiences, however, a sample size of 1,000 per country seems too small in many countries, especially so when victimization prevalence is low.

As mentioned in the introduction, the European Commission, and more specifically Eurostat, has done preparatory work on a future European Safety Survey (SASU) for some years. In most member states pilots have been carried out with a draft questionnaire (Aromaa et al. 2009). Building on these experiences, a new, slimmed-down questionnaire has subsequently been drafted (van Dijk et al. 2010). This questionnaire also covered internet-based crimes and a set of questions designed to monitor the new EU directive on victims’ rights. At the insistence of Italy, it also included a separate set of questions on violence between intimate partners. A budget had been earmarked at 12 million euro to conduct the SASU in all member States with sample sizes of 3,000 up to 10,000 per country in 2013. The conduct of such surveys required legislation and a proposal was submitted to the European Parliament in 2011 (2011/0146 [COD]). In September 2012 the European Parliament notified the Commission that the proposal had been rejected and that new proposals were to be made. This negative decision is a major setback for European criminology, especially for some of the new member states in which no victimization surveys have ever been conducted. To my knowledge neither the European criminological community, nor in fact any criminological expert, has been consulted about this negative decision by the European Parliament. The appointed rapporteur on the proposal, a British MEP, Timothy Kirkhope (Conservative), submitted a highly critical report. The rapporteur reportedly raised doubts about the added value of the SASU project, considering results from national surveys were available from so many member states. He also noted that the surveys would not be fully standardized because some countries, notably France and Ireland, had refused to include questions on domestic violence and/or sexual abuse.

Although some of these criticisms are not without merit—standardization was indeed not fully assured—the rejection of the SASU by the European Parliament seems to be a grave political mistake. With the growing interest in issues of crime, crime prevention, victim assistance, and criminal justice within the European Union and the expanded mandates of the Commission in this domain, a set of comparable statistics on crime seems indispensable for the planning, conducting, and monitoring of fact-based EU-policies. The absolute numbers of police-recorded crimes per country, published annually by Eurostat, are nothing but a source of misinformation about the distribution of crime across the Union. When expressed in rates per 100,000 inhabitants, crime is supposed to be the most prevalent in the Scandinavian countries and by far the least in Rumania and Bulgaria. Is there any MEP truly believing that this is information on which policies can be built? As criminologists know, the only possible way to collect credible and comparable statistics on the prevalence and trends in crime is a standardized victimization survey. This is why the federal government of the USA has set aside and sustained a considerable budget for its National Crime Victimization Survey. The EU SASU promised to do even more than just estimating levels of crime across the member states. It would also, as shown above, have provided policy relevant information on, inter alia, the use of crime prevention measures, police performance, and the treatment of victims by the police. In the absence of a European crime survey, criminal policies in the Union will be prepared, carried out, and evaluated “in the fog”.

I sincerely hope that the European Commission, including Eurostat with its national partners, will soon go back to the drawing table to put together a new proposal. One option worth exploring, considering the current stalemate, is to issue a call for public tender on the conduct of a standardized victimization survey using the SASU questionnaire. Building on the experience of the European parts of the fifth and sixth rounds of the ICVS, also duly funded by the Commission, such survey could be implemented by a consortium of polling agencies and crime survey experts. The project could use more modest sample sizes than the ones envisaged by Eurostat, for instance 3,000 per country. This would still produce robust indicators and surely cut down costs considerably. The content of the questionnaire could be revisited once more, especially the controversial set of detailed questions on violence and sexual abuse between intimate partners. As important as this topic is, the inclusion of a special set of questions about such sensitive issues complicates the execution of the SASU, especially among some of the new member states and Ireland. It seems to be a better option to design and execute the latter surveys with a tailor-made technique of interviewing, assuring optimal confidentiality. In fact, this is what the EU Fundamental Rights Agency (FRA) is currently doing in all member states. Some broad questions on victimization by violence and sexual incidents, taken from the ICVS, can of course be retained in the general crime survey. A challenging issue is the choice of the mode of data collection. The future of crime surveying seems clearly aligned with web-based interviewing. However, in the new member states internet penetration might not prove to be high enough. In my opinion, the CATI mode is still the most attractive option for an EU-wide survey in 2013 or 2014, on the condition that the samples include exclusive owners of mobile phones.

A CATI-based SASU with an abridged questionnaire and reduced sample sizes will be much less costly than the proposed SASU. It could be executed and reported upon by a competent consortium within a time span of months after the contract is signed. It will provide the Union with the necessary statistics on crime during the governing period of the current Commission. And it will buy time for Eurostat to prepare for a full-fledged SASU, conducted within the EU Statistical System, in 2015. In the meantime arrangements should also be made for continued collaboration with countries outside the EU.

Jan van Dijk is Professor of Criminology at Tilburg University.

1 ICVS-based surveys have also been carried out after 2005 in Argentina (Buenos Aires), Brazil, and Japan.
2 The survey in Georgia was, as said, repeated in 2011. The prevalence rates were equally low or lower than in the 2010 survey (van Dijk and Chanturia 2011).
3 No data are available on home security for Georgia but this country might be an outlier, with steeply declining trends in burglary in spite of low levels of home security. Data on home security for Switzerland are only available for 1996 (van Dijk et al. 2007). It is unlikely that home security has improved since 1996 to such an extent that the country no longer ranks among the countries with the least secured homes. The fact that the Swiss authorities funding the survey have replaced the questions on home security with a set of questions on social crime prevention suggests a lack of interest in situational crime prevention.
4 The correlation between special locks and burglary trends is less strong because the UK is an outlier with a rate for special locks of 60 percent, almost 20 points below the Netherlands and Germany. This is compensated by a high rate of burglar alarm ownership.
5 The costs of elementary home security have been estimated at 433 euro per house by Vollaard & van Ours (2010). The benefits are estimated at 780 euro over a 30 year period per house.
6 Reporting of crimes like burglary is likely to increase when more households are covered by insurance.

Aebi, M.F., and A. Linde. 2012. “Crime Trends in Western Europe according to Official Statistics from 1990 to 2007”. In The International Drop in Crime; New Directions in Research, edited by J. van Dijk, A. Tseloni, and G. Farrell. Hampshire, UK: Palgrave/Macmillan.
Ahven. A., K. Hommel, A. Markina, T. Rannama, J. Salla, L. Surva, M-L. Soer, K. Tamm, K. Tuisk, and K. Vakakas. 2010. Crime Victim Survey 2009, Criminal Policy Series 14. Tallinn: Ministry of Justice.
Aromaa, K., M. Heiskanen, S. Laaksonen, and M. Viuhko. 2007. Final Report for Action Entitled “Crime and Victimisation Survey”. Helsinki: HEUNI.
Beulens, B., J. van der Laan, B. Schouten, J. van der Brakel, J. Burger, and T.G. Klausch 2012. Disentangling Mode-Specific Selection and Measurement Bias in Social Surveys, discussion paper. The Hague: Statistics Netherlands.
Dijk, J.J.M. van. 2008. The World of Crime: Breaking the Silence on Problems of Security, Justice, and Development across the World. Los Angeles: SAGE Publications.
Dijk, J.J.M. van, R. Manchin, J. Van Kesteren, and G. Hideg. 2007. The Burden of Crime in the EU; A Comparative Analysis of the European Crime and Safety Survey (EU ICS). Brussels: Gallup Europe.
Dijk, J.J.M. van, J. van Kesteren, and P. Smit. 2007. Criminal Victimization in International Perspective: Key Findings from the 2004–2005 ICVS and EUICS. Den Haag: Boom Juridische Uitgevers.
Dijk, J.J.M. van, P. Mayhew, J. van Kesteren, M. Aebi, and A. Linde. 2010. Final Report on the Study on Crime Victimisation. Tilburg: Intervict report/Eurostat.
Dijk, J.J.M. van, and T. Chanturia. 2011. The Remarkable Case of Georgia; Georgian Crime Trends in an International Perspective; Secondary Analysis of the 2010/2011 Crime and Security Survey. Tblisi: Ministry of Justice.
Dijk, J.J.M. van, and B. Vollaard. 2012. “Self-limiting Crime Waves”. In The International Drop in Crime; New Directions in Research edited by J. van Dijk, A. Tseloni, and G. Farrell. Hampshire, UK: Palgrave/Macmillan. Dijk, J.J.M. van, and T. Tseloni. 2012. “International Trends in Victimization and Recorded Crime”. In The International Drop in Crime; New Directions in research, edited by J. van Dijk, A. Tseloni, and G. Farrell. Hampshire, UK: Palgrave/Macmillan.
Ghauharali, R., D. Meuldijk, and P. Smit. 2011. ICVS 2010 pilot; project report. The Hague: NICIS.
Glonti, G. 2011. Crime and Security Survey Georgia 2011. Tbilisi: GORBI.
Killias, M., A. Kuhn, and M. F. Aebi. 2011. Grundriss der Kriminologie; Eine Europaische Perspektive. Bern: Staempfli.
Killias, M., S. Staubli, L. Biberstein, M. Nanzinger, and S. Iadan­za. 2011. Studie zur Kriminalität and Opfererfahrungen der Schweizer Bevölkerung; Analysen im Rahmen der schweizerische Opferbefragung 2011. Zürich: Universitat Zürich/Kriminologisches Institut.
Killias, M., and B. Lanfranconi. 2012. “The Crime Drop Discourse- or the Illusion of Uniform Continental Trends; Switzerland as a Contrasting Case”. In The International Drop in Crime; New Directions in research, edited by J. van Dijk, A. Tseloni, and G. Farrell. Hampshire, UK: Palgrave/Mac­millan.
Lynch, J. P. 1996. “Problems and Promise of Victimization Surveys for Cross-National Research”. In Crime and Justice: A Review of Research, vol. 34, edited by Michael Tonry. Chicago: University of Chicago Press.
Pachulia, M. 1993. “Report on Georgia”. In Understanding Crime, edited by A. Alvazzi del Frate, U. Zvekics, and J. van Dijk. Rome: UNICRI.
Vollaard, B.A., and van Ours J.C. 2011. “Does Regulation of Built-In Security Reduce Crime? Evidence from a Natural Experiment”. The Economic Journal 121:485–504.