Final Report Summary - CPSI (Changing perceptions of security and interventions)
CPSI aimed to create a methodology to collect, quantify, organise, query, analyse, interpret and monitor data on actual and perceived security, determinants and mediators.
The project's four main objectives were to:
- develop a conceptual model of actual and perceived security and their determinants;
- design a methodology to register and process security-related data;
- develop a data warehouse to store amassed data; and
- carry out an empirical proof-of-principle study to test the model, methodology and data warehouse.
In CPSI we focused on security related to 'everyday' crime, such as theft, assault and vandalism. The CPSI methodology, however, can be applied to other areas of security as well, such as terrorism or financial security. During the life of the project, we carried out many activities towards realising these objectives.
Project context and objectives:
The project has the following objectives:
1. develop a conceptual model to describe the relationship between actual and perceived security and the determinants that influence this relationship;
2. develop a methodology to collect, quantify, organise, analyse and interpret data on the factors described in the conceptual model;
3. develop a data warehouse to aggregate and store data from different sources;
4. conduct a validation study to test the model, methodology and data warehouse in the field.
The degree to which we have realised these objects will now be addressed in more detail:
1. Conceptual model
The development of the conceptual model has been completed with the completion of work package (WP) 5. The first steps in this process were undertaken in WPs 2.2 (Morphological analysis) and 2.4 (Background data). Together, these work packages yielded a preliminary conceptual model which formed the basis for:
1) the perceived security survey;
2) the information extracted from existing databases on actual security;
3) the media analyses and ultimately;
4) the structure of the data warehouse. In sum, this objective has been reached as planned.
2. Methodology
The CPSI methodology is an integration of techniques for data collection, data storage, and data analysis and interpretation. It is specifically focused on improving understanding of the relationship between perceived security and:
1) actual security;
2) security reporting in the media; and
3) interventions intended to improve security.
Originally, we had anticipated being able to develop a method that could be used off-the-shelf by end users to monitor and better understand perceived security in their area. However, during the project, we discovered that much of the work needed to implement the CPSI methodology cannot be structured in a one-size-fits-all manner. As a result, we have developed a standardized methodological concept, though for implementation it is necessary to customize many of its elements. In sum, the principle of this objective has been achieved: we have successfully designed a methodology with which to process security data from collection to analysis. However, the translation to implementation of our methodology in the field turned out to be more complex than we had anticipated.
3. Data warehouse.
We have successfully reached this objective. The data warehouse, designed to store and query for analysis information from the perceived security survey, the actual security databases and media analyses, is operational.
4. Validation study.
This objective has successfully been achieved. We have:
1) designed and implemented the data collection methods as planned;
2) collected all data as originally planned;
3) successfully loaded all the data into the data warehouse in preparation for analysis; and
4) conducted statistical analyses on the data from the validation study to assess the conceptual model and provide proof of the CPSI principle.
Project results:
WP 2 - Conceptual model
a. Objectives and summary of progress
The objective of WP2 was to investigate the conceptual boundaries and structure of the problem space concerned with actual security and perceived security. This included:
- the ethical and legal aspects of the problem field (WP 2.1.);
- the structure of the total problem space (including the interrelationship between factors) (WP 2.2);
- the investigation of public opinion as related to actual and perceived security (WP 2.3.); and
- a background study and a literature review (WP 2.4).
All of the sub-WPs were carried out successfully, giving an overview of the problem space to be studied, and bounding this space in a practical manner which has allowed for the formulation of a 'realistic' survey questionnaire which could be compared to the actual security data available through census and crime statistics, and other data. WP2 also resulted in the development of a prototype 'Comparative intervention efficacy' model and demonstrated that it is feasible to develop an analytic hierarchy of comparative goals and goal criteria for different types of interventions.
Ethical and legal issues (TNO)
a. Objectives and summary of progress
The objectives of WP 2.1 were to address relevant ethical and legal issues as they became apparent during the project (the legal issues were addressed in a written brief). This was done in close cooperation with the researchers, including participation in project workshops. The overall objective was to develop a common understanding among the researchers about the ethical issues and possible ways to deal with them. This ensured a more all encompassing view on the research project.
A number of ethical issues were identified:
1) In the first morphological analysis, the available group of people was a bit biased (mainly Caucasian, highly educated men). In the following workshops more attention was given to the diversity of the group of experts and their fields.
2) The ethical dimension of different definitions of actual and perceived security, and the problems with gathering data on crime and on personal opinions / dispositions were identified.
3) Based on the report on the Dutch legislation on data protection, it was decided that the survey data, as analyzed by the researchers, were not personal data. It was, however, necessary to recode should too few respondents be in a category. For future use (local) governments are warned that making the geographical analysis units smaller might lead to problems regarding privacy and personal data because indirect identification of respondents is possible.
4) It was decided that involvement of the security industry in the project could lead to perverse results as the industry needs people to feel insecure so that they purchase security technologies.
5) Users of the CPSI system are warned not to upload data from very different sources such as social security, car insurance etc. Combining data from very different sources risks violating contextual privacy. Advanced data mining techniques cause problems with regard to the informed consent of respondents.
6) With regard to the legal issue the most pressing issue is that different EU countries have different national implementations of the data protection directive. This means that some survey questions especially those concerning sensitive issues cannot be asked in every EU country.
b. Significant results
- A report on the Dutch data protection legislation with regard to the CPSI survey was completed.
- A report on the national interpretation of the Eu Data Protection Directive is completed.
- The final report on Ethical and Legal issues is delivered.
- An article was written that is already available online: Van Gorp, A. C. and Van der Molen S. (2009), Parallel, embedded or just part of the team; ethicists cooperating in a European security research project, Science and Engineering Ethics, published online at DOI 10.1007/s11948-009-9187-5.
- A manuscript was written for a special issue in Security Journal: Van Gorp A. C. Is an informed consent form sufficient: Ethical and legal issues in perceived security research.
- Interviews have been held with about two thirds of the project participants.
c. Deviations from project plan
None
d. Statement of the use of resources
The total budget for WP 2.1 was increased during the first reporting period. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
Morphological analysis (FOI)
a. Objectives and summary of progress
The objective was to structure and investigate rigorously the dimensions of the total problem space addressed in the project. Each of the WP's required specific expertise in various disciplines. Given these various backgrounds and the need for a common conceptual framework, a series of problem structuring workshops were carried out in order to create among the project members - and outside expertise where needed - a common problem structure and modelling framework, as well as to specify the conditions for the future work packages.
The workshops utilised computer-aided morphological analysis, a non-quantified modelling method developed at FOI. The workshops resulted in a number of morphological inference models defining the total (dimensional) problem space that the project addressed as well as the interrelationships between the problem variables.
b. Significant results
The initial morphological fields for actual security and perceived security produced multi-dimensional typologies which were wider and more varied than were found to exist in the research literature, especially as concerns the relationship between 'ethic' (objective, statistical) variables and 'emic' (subjective, motivational) variables.
The dimensions of perceived security could be reduced to three key variables:
(a) types of threat;
b) locations; and
c) a set of demographic parameters without significant loss of information.
c. Deviations from project plan
None
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings. Note that the budget for WP2.2 was decreased due to the fact that less budget was needed than had originally been anticipated. The budget that resulted from this decrease was transferred to various other beneficiaries for work in other WPs. This decrease in no way affected the execution of the work as it is described in Annex 1.
Public opinion (University of Kent)
a. Objectives and summary of progress
This WP endeavoured to develop an understanding of how public opinion relates to and influences security. In particular, we were interested in developing an overview of how attitudes towards security influence people's behaviour and indirectly bear upon public policy. This sociological analysis should contribute to the development of a conceptual model of perceived security.
The study has resulted in a 'Review of existing surveys on public opinion of security' which includes material on:
1) what surveys tell us about the EU public's perception of insecurity;
2) how opinion surveys relate to security; and
3) opinions on specific threats such as immigration, climate change and economic crisis.
b. Significant results
- Review of existing surveys on public opinion of security
- Inventory of European crime surveys and crime survey summary
- Recommendations for future research
c. Deviations from project plan
WP 2.3 started late due to issues with the moment at which University of Kent was able to begin work. However, this has had no negative impact on other tasks, resources or planning.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
Background data (SCP)
a. Objectives and summary of progress
The objective was to examine the state of the art, i.e. to generate insights on the basis of literature about the determinants of perceived security, the relation between actual and perceived security. These insights have contributed to the development of a conceptual model of perceived security.
Besides being a broad literature study, WP2.4 addresses the important links between individual and contextual factors and the fear of crime, along with coping strategies. These individual and contextual determinants are gathered in a conceptual model explaining fear of crime. The process of data collection is described and the conceptual model is then translated into indicators on which data can be gathered either from official registrations or from a local survey newly designed for the present purpose.
b. Significant results
- A fear of crime literature review
- List of factors for designing the methodology identified in the literature review
- Comparison with the factors developed in the morphological analysis
c. Deviations from project plan
None
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
WP 3 - Data warehouse
Overview
The development of a data warehouse is necessary to create a central environment for all the data needed to make analysis of the security work field happen. To conduct the analysis work, we need to integrate all the 'security related' data from the location of the validation study and distribute the information that is derived from the 'security related data' to the authorised users.
Information analysis report (Sogeti)
a. Objectives and summary of progress
The objective of WP 3.1 is to describe all the requirements for the data as needed in order to make the analysis work. It contains a summary of all definitions of the data captured in the system.
The following topics were addressed:
- Identification of the data dictionary
- Design of the data models for data storage and reporting
- Design of the loading process including the data delivery interfaces
Technical design documents were delivered as input for the construction of the database environment itself. This WP has been completed.
b. Significant results
All subjects in the CPSI model have been addressed and are integrated into one information model for the data warehouse.
c. Deviations from project plan
Originally, the media analysis was designed to be delivered by TEMIS. The TEMIS data were aggregated according to the nature of the desired analyses. However, for the pilot we opted for the inclusion of the more 'raw' RSS feeds generated by JRC.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded, note though that the total budget for WP 3.1 was increased during the first reporting period; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
Database environment (VLC)
a. Objectives and summary of progress
The objective of WP 3.2 is to construct the database environment together with the procedures to load data and the tooling. The activities, which were executed in the first phase, mainly concerned the technical preparation of this work package. For the second phase the technical design documents from WP 3.1 are used as input. This work package has been completed.
b. Significant results
- The required hardware and software were selected, bought, installed and practiced with.
- The data warehouse was designed and constructed.
- Data was loaded into the data warehouse.
- Reports were built.
- Data extractions were generated for analytical purposes.
c. Deviations from project plan
The construction of the pre-processing step was skipped, for the following reason. Although, a standard interface was designed, in practice it is hard to conform to that interface. It requires a lot of manual preparation for the delivering party. Therefore, it was decided not to automate that step. This has no further impact on other work in either this or other WPs.
d. Statement of the use of resources
The total budget for WP 3.2 was increased during the first and second reporting periods. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Another part was spent for purchasing software necessary for the data warehouse. Minor costs were incurred for workshops and project meetings.
WP 4 - Data collection / validation
WP 4.1 and 4.2 (SCP)
a. Objectives and summary of progress
WP 4.1 contains a list of the databases acquired for inclusion in the data warehouse. In order to select the relevant variables within these databases, we used the outcome of the morphological analyses and the literature review.
WP 4.2 contains of two elements: selection of a city for the pilot study and designing a questionnaire. After ample consultation within the consortium the decision was made to conduct a survey on perceived security in only one city: Amsterdam. This city has its own department for research and statistics (O+S Amsterdam) and has a permanent survey among the Amsterdam population on safety and security. O+S were prepared to cooperate with us and permitted us to link our survey to their own.
In consultation with O+S Amsterdam we designed a questionnaire. The research period took three months and the netto-response was 49 % (N = 1222). Some 5 % of the response seemed of no use because of incomplete filling-in of the questionnaire.
b. Significant results
To validate the conceptual model, data was needed on the social context of the respondents participating in the local study (WP4.2). We needed information on the level of actual crime, demographical and social composition and on the physical structure of the neighbourhoods people live in. Since our local study was carried out in Amsterdam, we contacted O+S (the Department for Research and Statistics of the Municipality). O+S has an independent position within the municipality of Amsterdam that can be compared to that of Statistics Netherlands (CBS) within the national government. The variables listed are complementary to the survey data collected in Amsterdam. A list of variables available from O+S to validate the conceptual model covers police statistics, demographical and social composition and physical structure.
The survey in Amsterdam (4.2) went well; the cooperation with the research organization O+S was excellent.
c. Deviations from project plan
All went as planned; no deviations.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings. Note that the budget for WP's 4.1 and 4.2 was decreased due to the fact that less budget was needed than had originally been anticipated. The budget that resulted from this decrease was transferred to various other beneficiaries for work in other WPs. This decrease in no way affected the execution of the work as it is described in Annex 1.
WP 4.3 (Temis, JRC)
a. Objectives and summary progress
The main goal of the processing of the textual (media) data in the CPSI project is to extract key content from various sources in English and Dutch and to export it for populating the CPSI data warehouse with security concepts alone or in relation to geographic locations concerned. The added value of using text mining technologies resides in the possibility of extracting, from unstructured text, key content from textual media content.
The extracted information provides various analysis dimensions as factors for media salience of security related issues. For example:
- timely analysis of security related topics in news articles;
- analysis of geographic distribution of security related topics in news articles.
b. Significant results
A pilot application was set up that annotated documents for a period of seven months starting from November 2008 until the end of May 2009. During this period, more than 23 000 documents in English and Dutch were annotated. The export files of the structured information for import into the data warehouse contain circa 73 200 lines in the defined format: document identification; document date; source name; crawling category; document position; security concept; extraction type; location.
Achievements for WP 4.3 are the following:
- A specific knowledge component for extracting security concepts in English and Dutch was built. The structure of this component uses the conceptual model coming from the morphological analysis WP. The following categories of security issues have been retained for the pilot application: discriminations, drugs, economic downturn, property crime, terrorism and violence and vandalism.
- The media texts crawled by EMM of JRC are annotated with the software of TEMIS for information extraction and text analytics called Luxid®.
- A specific export procedure that exports the structured media content for the import into the CPSI data warehouse was developed. For example, a relation between the security issue 'murdered by Islamic militants' and the location 'south-central Somalia' extracted from the text 'murdered by Islamic militants in south-central Somalia' is exported with the reference to the media text from which it comes.
- An access to the user interface of Luxid® and set up analysis scenarios for exploiting the media content was provided. For example, negative sentiment words (like accuse, expensive, greatest difficulties, worst, conflict, fear, unable, error, etc.) or opinion verbs (like criticise) can be put in correlation with security issues by using the analysis tools of the platform. This permits identification of key content in the media texts regarding the perception and positions toward security, for example: 'Advertising firms censor signs for fear of vandalism', 'Report blames racism for rise in youth violence', 'US to blame for much of Mexican drug violence'.
Depending on the scope of investigations, the pilot system can be extended to other countries and languages, other information sources and other extraction needs. For example, in order to analyse the opinion of countries, organisations, official institutions, influential persons, governments etc. or individual persons towards security issues, one of many future possibilities is to expand the knowledge components to allow for the extraction of opinions in relationship to the opinion holders. The extraction of opinions expressed in the media may constitute in the future a component of an information intelligence platform supporting predictive modelling tasks.
The following tasks may be supported:
- Provide an overall landscape of security perception by different media sources: Identify the tonality of a media text and the communication strategies of a journalist or globally of a media source that are set up when talking about security topics.
- Identify groups of influence: Identify social networks according to the point of view on security issues shared by social movements (organisations, official institutions, influent persons, governments etc.).
c. Deviation from project plan
None.
d. Statement of the use of resources
The total budget for WP 4.3 was increased during the second reporting period. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
WP 4.4 (SFU - CESS)
a. Objectives and summary progress
Work in WP 4.4 and its deliverable took place in a long-term frame and on different levels, including literature study, desk research, mixed working groups with end user involvement (including cross-fertilization with ESRIF WG10 and WG11 according to the CPSI dissemination plan). Report on cultural issues on two levels (society and political selection of risks according to the Douglas / Wildavsky model) were concluded, as well as a report on the country case studies conducted (AT, BG, FR, GE, IT, NL, SE, UK - reflecting the composition of the end user advisory group), summarising and illustrating results; paper on culture-sensitive communication guidelines to prevent or reduce gaps between citizens' actual and perceived security.
b. Significant results
High social fear of crime countries (crime perceived as a problem 'out there', such as prominent in Austria, the Netherlands and Sweden) have under-average personal fear of crime but an over-average number of offences reported to police, indicating low actual security. Personal fear of crime in this type of country is also lower than the victimisation level would suggest.
High personal fear of crime countries (crime perceived as an individual(ised) issue, such as prominent in Bulgaria and Italy) can still have average victimisation levels. They have however an under-average number of offences reported to police (high actual security) and average or under-average social fear of crime. Personal fear of crime is thus more detached from actual security than social fear of crime. More social fear of crime is associated with less personal fear of crime. This leads to the hypothesis that social fear of crime reduces personal fear of crime.
The EU itself does not generally seem to be a source of citizens' fear of crime in their own country; France, Germany and the UK appear to be exceptions. At the same time, citizens perceive national interventions to be most suitable to enhance their security against crime; the exception is Italy, where citizens appear to prefer EU solutions also to national security problems. For the UK, statistical analysis, while supporting no strong interpretation, points to a possible interpretation in a similar direction.
Despite public support for EU decision-making and action against crime, the EU itself is generally not perceived as a locus of successful interventions to enhance citizen security against crime; the exception is Italy. For the UK again, statistical analysis, while supporting no strong interpretation, points to a possible interpretation in a similar direction.
CPSI country analysis so far has shown no consistent association between acceptance of technological solutions for security problems (such as CCTV) and relationship between level of societal and of personal fear of crime, level of victimization, relationship between felt and actual personal / social security and victimization.
CPSI country analysis however has shown consistent association between acceptance of CCTV and cultural attitudes towards technology: countries in which technology is interpreted as part of the security problem (e.g. critical infrastructure protection, information technology as object of offence and source of insecurity), public CCTV acceptance is lower than in countries where technology is interpreted as part of the solution (e.g. information technology as a foundation for coordinated, efficient prevention and response).
On the level of cultural selection of risk, analyzed by Member State security research policies / programmes, in cultural factors were found to generally have the strongest positive impact on (developing) a comprehensive approach to citizen security at the national level; they were found to generally have the strongest evidence of negative impact on splits in thematic thrust (such as society vs. technology-centred security interventions).
Citizens' knowledge is the key factor for their perception of security and interventions. Knowledge also mediates between felt and actual security. This study thus highlights the need for continued interpretation of security as an information and cognition issue. While cultural factors have a positive impact on solutions to security problems when these are perceived by the public as national problems, the EU must minimise its negative impact on solutions to security problems when these are perceived by the public as 'European problems'. At the same time, security was found to continue to be mainly a national cultural value rather than representing a value common to European citizens.
Moreover, knowledge turned out to be the first-rank effectiveness criteria for security-enhancing interventions. Strong knowledge and interpretative contexts present on the national level are a cultural factor that decreases citizens' individual perception of insecurity. That means social fear of crime reduces personal fear of crime. Actual insecurity particularly increases social fear of crime (perception of crime as a problem 'out there') but decreases personal fear of crime (perception of crime as an individual concern).
Interventions should in the first place focus on a comprehensive definition or articulation of risk based on exchange of knowledge. Policy interventions based, in contrast, on perceived security have been found to increase citizens' fear of crime.
While cultural factors have a positive impact on solutions to security problems when these are perceived by the public as national problems, they have a negative impact on solutions to security problems when these are perceived by the public as 'European problems'.
Effective communication strategies to reduce gaps between actual and perceived (in)security need to determine - even before a choice of concrete tools or measures - the cultural context in which the communication-based intervention is to be implemented and organise planning according to whether personal fear of crime or social fear of crime is the predominant societal context, for instance:
- In countries with a social fear of crime culture, communication-based interventions typically lower citizens' perception of (in)security, and this type of gap between actual and perceived security is mainly due to styles of social meaning making as rooted in the social context.
- In countries with a personal fear of crime culture, communication-based interventions typically increase citizens' perception of (in)security, and this type of gap between actual and perceived security is mainly attributable to disproportionate media reporting.
c. Deviation from project plan
In order to avoid duplication, the risk of which only became visible during the actual work, project review meetings concluding WP4.4 activity concentrate on the national aggregation level, covering two perspectives: Political culture (society) and political selection of risk (politics of citizen security). The selection of these perspectives reflects the two-fold research interest of CPSI: citizens' perceptions of security on the one hand and interventions on the other. Further, to avoid duplication with WPs dedicated to survey and quantitative analysis, WP4.4 concentrated on the national level of aggregation and include eight country reports (Austria, Bulgaria, France, Germany, Italy, the Netherlands, Sweden, United Kingdom - reflecting the composition of the end-user advisory group).
Work on the 'definition of culture-sensitive media strategies for threat / security situation communication and public awareness-raising in order to close gaps between actual and perceived security' was postponed because project reviews showed that links with WP4.3 (University of Kent) needed to be strongly explored. The development of communication guidelines was consigned to WP4.4 because the results of the culture analysis yielded several points to directly connect to, which had not been anticipated in the DoW.
Work on some parts of WP4.4 showed that cultural issues as analysed in the WP4.4 framework (method of security culture) do not provide enough information to produce envisaged typologies such as 'Typology of security perception differences between different typified locations (cities, towns, regions, country)'. During project review, it was therefore decided to cover these issues in the final deliverable on the basis of availability of cross-WP data and research results.
Introduction of WP4.4 results into the conceptual model though the data warehouse could not be implemented as planned because several attempts to construct a translation mechanism between the cultural data available and the quantitative information able to be grasped by the data warehouse failed.
d. Statement of the use of resources
The total budget for WP 4.4 was increased during the first reporting period. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
e. Actions taken
Work relating to culture-sensitive media strategies were conducted together with University of Kent, following WP4.3 course of action and responsibility. Integration of WP4.4 cultural data and research results into to conceptual model was accomplished though morphological snalysis.
WP 5 - Analyses
a. Objectives and summary of progress per task
The objective was the development of a series of inference models describing the empirically based relationships between the parameters originally mapped out (theoretically) in WP 2. These models will map out solution spaces to the total problem space developed in WP 2, and should allow for multiple drivers both in order to make comparisons between different countries (where applicable), cultural groups and social strata, and to allow any parameter in the problem space to be treated as an independent variable.
This objective has been achieved. We have contributed to the work packages preceding WP5 in such a manner that we are assured that the necessary theoretical models (formulated in conceptual structural equation models (SEM)) are in place and that the collected data is of the nature and detail that testing of the CPSI methodology was possible.
The WP breaks down into three separate tasks: quantitative analysis, qualitative analysis and synthesis. The main task of this work package is to analyse the data resulting from earlier WPs and synthesise the findings of WPs 5.1 and 5.2. in WP 5.3.
WP 5.1 Quantitative analysis
a. Objectives and summary of progress
The aim of this WP 5.1 'Quantitative analysis' is to demonstrate how quantitative analyses can be performed with the data collected through the CPSI method. In a pilot study in Amsterdam (the Netherlands), the data collection and analyses were demonstrated and a general model was tested. Based on the results, a standardised EU monitor is presented that can be used to concisely and fully measure perceived security. In order to be able to do that the following procedure was used.
An overall CPSI model was developed, which hypothesises how actual security and perceived security are causally related and how this relationship is influenced by background factors of citizens and media salience of security topics at the neighbourhood level. By using data from different sources, aspects of this model were tested at the neighbourhood level.
Before testing the overall model, variables to be used in this analysis need to be identified and selected. Four sources of data are used:
1) perceived security data collected by means of a survey of individual citizens (WP 4.2 Perceived security: questionnaire);
2) actual data procured from police statistics at a neighbourhood level (WP 4.1 Actual security: databases);
3) background data on neighbourhoods at a neighbourhood level procured from the bureau of statistics (WP3.1 Design and construction data warehouse, WP3.2 Integrate databases); and
4) media salience of security topics procured through the media analysis procedure (WP4.3 Media content analysis). Each data set was scrutinized separately for data integrity and fitness for purpose.
b. Significant results
The brunt of the work dealt with reducing the number of variables (questionnaire items) of the survey. The original questionnaire consisted of more than 500 items, which cannot all be used in the testing of the overall model at the neighbourhood level. A selection procedure was used to reduce the number of variables to 59 and thus reduce the questionnaire items to 'a standardised EU monitor'. This selection procedure took a four-step approach. First of all, items that were not relevant for CPSI were omitted; these were items that were used by the Amsterdam Department of Research and Statistics in a standard security monitor (VMR) that was part of the questionnaire but were of no relevance to the CPSI pilot. Secondly, a data reduction procedure was used to narrow down the number of items, based on the theoretical model of perceived insecurity (WP 4.2 Perceived security: questionnaire). Thirdly, an empirical procedure was used to further reduce the number of items. By factor analysing the remaining items and identifying the highest loading variables on the extracted factors, an empirical base was achieved for the hypothesised theoretical model of perceived insecurity. Finally, these items were constructed into scales at the aggregate neighbourhood level and ready for use to test the overall CPSI model. We suggest for the future implementation of the CPSI method using the 'standardised EU monitor'.
For testing the overall CPSI model of perceived insecurity, the survey data at the aggregate neighbourhood level was merged with the actual security data, data on the salience of media topics, and background data at the neighbourhood level. The resulting file was used for the testing of the overall CPSI model. In this analysis, hypothesised relationships were tested and explored and gave rise to relationships of interest and practical use for future CPSI implementation. For example, perceived security was only partially explained by actual police statistics on crime and demographics of the neighbourhood. However, neighbourhood quality ratings and social cohesion appeared to be more powerful predictors of perceived security: neighbourhoods that are rated as secure and cohesive report lower levels of fear. Media salience is also related to perceived security in several ways.
In general, the pilot demonstrates that the CPSI methodology can be used as intended. Implementation of the methodology at local or higher levels of public administration is feasible and allows for a unique integration of data from different sources. As such, it can supplement policy assessment by testing of interventions and enhance transparency of citizen security.
c. Deviations from project plan
There were no deviations from the project plan.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for RTD-related workshops and project meetings.
WP 5.2 Qualitative analysis
a. Objectives and summary of progress
The qualitative analysis of the CPSI Security Survey concerning perceived security, victimisation and other project data was meant to give a broad overview of some of the core results in an easy-to-use and easy-to-understand form. It is not a statistical study in the sense of identifying 'statistically significant' relationships or correlations. This was left to WP 5.1: Quantitative analysis.
The form of the qualitative analysis is a number of computer-supported morphological inference models, by which the user can examine and compare relative demographic differences in threat perception and feelings of security, reported victimisation and the use of media sources. This study, together with the WP 5.1 (Quantitative analysis) provides a basis for drawing conclusions concerning relationships between actual and perceived security (WP 5.3 - Synthesis), which is one of the prerequisites for identifying appropriate security interventions.
The qualitative analysis was also concerned with:
1) identifying principle factors involved in media-based opinion building as well as examining the feasibility of mapping out the complex structure of interrelations between these factors (from WP 2.3); and
2) a comparison of national (public) security cultures based on a number of national security indicators compiled by WP 4.4.
b. Significant results
Three of the most interesting indicators for studying the survey material on perceived threat or insecurity in a population were gender, age and immigrant status. Some general trends were uncovered by the morphological models developed around these indicators. Among other things, the 'threat' that the survey population was clearly most worried about is 'discrimination'. This was the case across gender, age groups and immigrant status. The only slight deviation is that age group '15-25 yrs' seems to be the most concerned about discrimination of all the age groups. (However, this age group was the smallest sample). The general rank order of the most worried about types of threats is:
1) discrimination;
2) crime on the streets;
3) economic crisis and unemployment; and
4) lack of integration by immigrants.
Also notable is that as concerns general feelings of insecurity, women tend to feel more insecure than men (if only marginally), feelings of general insecurity tend to decrease marginally with age, and there is no stated difference in general feelings of insecurity between Dutch born and Immigrant responders.
As concerns media-based opinion building, it is fully feasible to utilise the morphological modelling process to identify and hypothesise the relationships between demographic variables and variables concerning media use and influence, by which these hypotheses may be tested. However, as the work revealed, every relationship is virtually a study in itself. Thus the main utility of the prototype model was its structuring of the possible types of relationships involved.
c. Deviations from project plan
There were no deviations of the project plan.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for RTD-related workshops and project meetings.
WP 5.3 Synthesis
a. Objectives and summary of progress
The objective of WP 5.3 Synthesis is to synthesise the results of WPs 5.1 Quantitative analysis and 5.2 Qualitative analysis. The qualitative and quantitative approaches are assumed to be complimentary. This WP shows how the two approaches can mutually address the systematic analysis of issues concerning citizen security.
The quantitative analysis focused on the available data obtained in WP 4. The previously identified causal relationships between the relevant security factors were tested for the strength of the relationships. Various statistical techniques have been used among others, correlation, factor, reliability analysis and structural equation modelling.
The analyses gave insight into how actual security relates to perceived security and how this relationship is mediated or moderated by media (public opinion) indices and cultural or background factors. Furthermore, the data was analysed for psychometric purposes at the individual level, for the analysis of relationships between actual security and perceived security the data were analysed at the neighbourhood level. No longitudinal data were obtained within the time frame of the project, thus the effect of interventions could not be tested.
The second approach is qualitative in nature; this analysis described the results of the computer-aided inference models describing solution spaces, and their consequences. This activity focused on the development of a series of inference models describing the empirically based relationships between the parameters originally mapped out (theoretically) in WP2. These models map out solution spaces to the total problem space developed in WP2, and allow for multiple drivers both in order to make comparisons between different countries (where applicable), cultural groups and social strata, and allow any parameter in the problem space to be treated as an independent variable.
A number of modelling workshops were carried out with the project beneficiaries, University of Kent and the Centre for European Security Studies, and including specialised expertise where needed. The workshops utilised such methods as computer-aided morphological analysis (MA), multi-criteria decision analysis (e.g. AHP), quality function deployment (QFD). Trained facilitators knowledgeable in the methods employed facilitated the modelling groups.
In order to achieve a synthesis of both types of analyses (qualitative and quantitative) we made use of the overall model used in the quantitative analyses in WP 5.1. This basic model postulates that perceived insecurity is dependent on actual security and that this relationship can be mediated by either media or background factors.
First, we described the communalities and discrepancies between the overall concepts. We then described how the two approaches converge or diverge with respect to hypothesised relationships between the concepts in the overall CPSI model. Finally, we discussed how the overall model can facilitate understanding and how the CPSI method can facilitate evidence generation for all sorts of queries about citizen security.
b. Significant results
Our synthesis of analyses demonstrated that qualitative and quantitative approaches are complementary in analysing complex and comprehensive issues such as citizen security. As such, a broad spectrum of questions can be asked and tested: concepts are defined, relationships are identified, context is given, and technology is made available to collect, store, and analyse data. However, intensive interaction with stakeholders is crucial to select the specific questions to be addressed. Only stakeholders, such as a security officer or a policy maker, can truly recognize the urgency of which questions to ask in the domain of citizen security and prioritise research efforts.
Based on the four factors of the CPSI model, hypotheses can be tested at a high level of abstraction. At the same time, stakeholders might inspire testing of more specific hypotheses. In general, all relationships between factors of the model can be studied using the CPSI method. For example, the end user of CPSI can use the method to examine the following relationships:
- measure the level of perceived security at any given time;
- measure the level of perceived security in any neighbourhood;
- measure the relationships between perceived security and other factors (such as actual security, media, or background) at any given time;
- measure the change in perceived security over time in any neighbourhood;
- measure the change in relationships between perceived security and other factors over time;
- measure differences between (types of) neighbourhoods in terms of perceived security and actual security;
- examine the effect of changing salience of media reports on perceived security over time.
The CPSI methodology allows for testing the effects of implemented interventions on actual and perceived security at the neighbourhood level. This implies longitudinal data collection at neighbourhood level, i.e. measuring before and after implementation of the intervention.
c. Deviations from project plan
There were no deviations of the project plan.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for RTD-related workshops and project meetings.
WP 6 - Dissemination
a. Objectives and summary of progress
A CPSI web portal was established (see http://www.cpsi-fp7.eu online), featuring a video with basic information about the project, event announcements, WP abstracts, PowerPoint presentations, papers, etc. It was updated regularly. The website contains information on project-related workshops and symposia (or parts of symposia) held or taken part in by CPSI project members, the project and its methodology, discussion of intermediate results, reference to (experts from) other FP7-SEC projects of relevant parts of scope (e.g. CAST, FORESEC). CPSI dissemination actions (such as appearance at symposia) were coordinated on the basis of a regularly updated dissemination plan and a conference calendar, identifying relevant opportunities for dissemination.
An international final symposium was held, with 68 participants from 15 countries and media reporting. E-lectures were recorded, edited and made available on the CPSI project website.
b. Significant results
CPSI was presented on the Third European Security Conference Initiative (ESCI) in 2008, organised by the WWEDU/Centre for European Security Studies and attended by some 100 participants from academia, industry and policy. CPSI partners have made relevant contributions to several ESRIF working groups, including WG10 and WG11, where CPSI research results were also exploited for the purpose of producing themes for further security research. In addition, research for CPSI WP4 deliverables (especially report on cultural issues) was disseminated into ESRIF WG10 contribution to part 2 of the ESRIF final report.
CPSI results were communicated to media representatives during the CPSI final symposium and a DVD was produced and delivered, containing all deliverables and other documents.
c. Deviations of project plan
The end-user advisory group made itself available for review of CPSI deliverables from an end user point of view, which was originally not foreseen and is a positive deviation. Depending on the time frame for and kind of this review feedback, it now and again impacted the schedule for submission of deliverables and required some additional text work and person months resources.
WP6 objectives were amended by ESRIF-related dissemination objective as ESRIF only started after CPSI DoW was finalised and turned out to entail working groups relevant for and open to input from CPSI. This was a positive deviation and did not considerably impact resources. Rather, it opened up space for discussing CPSI intermediate results and still better designing them to end-user and European needs.
The CPSI summer school could not be held although it was fully planned and stuffed and advertised throughout Europe. There was not sufficient turnout on the students' side. Elements from the summer school curriculum were resumed at the CPSI final symposium, and e-learning content for the CPSI website that had been planned to be gained from the summer school was generated from appropriate presentations at the CPSI final symposium.
d. Statement of the use of resources
The total budget for WP 6 was increased during the first and second reporting periods. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Major costs were incurred for organisation of the final event.
The CEUSS budget increase addressed a recalculation related to the transfer from WWEDU to SFU, resulting in more personnel costs for WP6. Additional budget was transferred to CEUSS in order to cover dissemination / reporting activities in relation with the CPSI final symposium that had not been foreseen (such as reports in stakeholder magazines and a full-scale conference report). Additional budget was allocated to Sogeti, to cover costs for the designing and developing of the final CD-ROM.
Minor costs were incurred for workshops and project meetings.
Potential impact:
The dissemination plan is described in detail later in this report and contains an overview of all dissemination activities carried out in the second reporting period and shortly after the termination of the project.
The exploitable foreground in CPSI consists mainly of two aspects. First and foremost is the implementation of the CPSI system in end users' organisations. This is indeed the primary goal of the CPSI project. The purpose of implementing this system would be to help end-users:
1) increase their insight into perceived and actual security in their area, the determinants of this security and mediating factors; and
2) to increase the effectiveness of the allocation of resources (financial, personnel, material).
Indeed, a better understanding of what affects security in the desired direction helps those responsible for strategic and operational planning decide what interventions to develop and implement for which locations. The second exploitation possibility we anticipate is in regards to providing support and advice to relevant organisations based on the knowledge developed in CPSI. This could be, for example, consulting in the area of statistical analyses suited to complex data sets, similar to that used in CPSI, or in the area of cultural considerations as they relate to security. In this way, it is possible for us to assist parties who have their own extensive registration systems (and hence to not need or want the CPSI system as a whole), but who cannot maximise the information possibilities of their systems. To illustrate, TNO recently visited the city of Rotterdam to discuss their registration system. Though their security information registration system is very good and extensive, they have two areas in which we may be able to advise them based on knowledge developed in CPSI:
1) how to statistically combine and interpret their data on actual and perceived security and interventions; and
2) possibilities for advanced analysis using structural equation modelling.
Another concrete possibility is to participate in a pilot study, currently being planned by the Dutch Ministry of Internal Affairs and Kingdom Relations, regarding the toolkit of security perception and the 'Table of 12,' which is an overview of 12 types of security interventions that can be implemented in the field.
In the case of both exploitation possibilities, we anticipate the 'target audience' is primarily local, regional and national governments. Secondary target audiences include organisations responsible for crisis management such as Sweden's MSB (Swedish civil contingencies agency) or the Red Cross. CPSI has had excellent relations with the Austrian Red Cross throughout the project.
For both exploitation possibilities, we expect the first exploitation within the next year.
In principle no additional development of the CPSI system is needed before it can be implemented in the end-user's organisation. However, as the CPSI system is not an off-the-shelf product that can be implemented as a sort of plug-and-play application, implementing CPSI in an end user's organisation will comprise some custom fitting of the system to the needs and constraints of that user. Examples of custom characteristics are:
- cultural considerations regarding what questions can be asked and what cannot (e.g. political preference, income, ethnicity);
- specific considerations relating to the security culture of the particular country (e.g. centralised security policy versus de-centralised);
- coding of criminal offenses;
- incorporation of information specifically requested by the end-user;
- integration with existing registration and software systems;
- translation of keyword list for media analyses and choice of media sources;
- translation of standardised EU security monitor;
- development of software routines to automatically perform statistical analyses according to the needs and desires of the end-user.
There are different aspects of the impact of the CPSI system and its implementation. First, in the short term, it will provide a better snapshot of security related concepts than is currently available in most areas. In the short term, questions can be answered such as: What is the relation between media use and perceived security by neighbourhood? Or Which types of crimes are most related to a low perception of security? Second, in the long term, it will be possible to monitor trends in security. Then, questions can be answered such as: What is the effect of 'yuppification' of a neighbourhood on crime? On perceived security? Or When and where does camera surveillance work best? Third, the overall impact of a widespread use of the CPSI system is a more unified, systematic and integrated registration and analysis of security data. This supports a better overview of best practices, which benefits not only the primary user of the system, but, through a sharing of experiences and information, also the common good. This impact cannot be qualified as such and may take a number of years to truly 'mature.' However, the sooner we start implementing the system, the sooner we can reap the benefits.
In addition to the positive effects the output of a project may have on public security, there may also be a negative societal impact. This addresses things such as the effect of the product on privacy, civil liberties and human rights in terms of the violation of ethical, moral or legal rights and principles. As regards the CPSI project, during the execution of the project we made every effort to ensure the privacy of the participants through the use of suitable anonymity procedures and the aggregation of the data to the group level before conducting analyses, in order to insure that participants' identity could be neither directly nor indirectly ascertained. Furthermore, in the dissemination of the empirical study, we were careful to not make known the individual neighbourhoods in which data were collected. In this way, we hoped to provide additional 'protection' for the participants.
Aside from how we dealt with concerns about the societal impact of the execution of the project, we also considered the broader scope of what the actual implementation of the CPSI system can mean for concerns about privacy and other essential values. CPSI has designed a system to monitor security-related information for the betterment of society as a whole and the citizen as an individual. In order to assist the end-user in applying the system in a responsible way, we have drawn up an overview of recommendations and caveats, which we recommend the end-user follow. This overview is included in the final project CD-ROM. These include:
1. 'Privacy and informed consent issues'
Informed consent must be acquired when collecting data from respondents, such as is the case with the standardised security survey. It is essential that respondents understand what is being asked of them and what will be done with the information they provide. The nature of the informed consent form in your country may be legally regulated, so check with local authorities for more information. We recommend care when using any data at the level of the individual citizen or small group; protecting individuals' data and their right to privacy is paramount.
2. 'Use of surveys: Statistical significance of results and reporting'
For those individuals with a layman's knowledge of statistics, it is often difficult to correctly interpret results and put them in perspective. For this reason, we advise caution when reporting results, particularly when these results are used to set the political agenda.
3. 'Secure protocols are needed to secure personal data'
The end-user should consider procedures to securely upload the data. At this moment there are no secured lines or protocols inherent in the CPSI system but especially if the data warehouse is used for monitoring and data are uploaded regularly then there need to be secure lines or protocols.
4. 'Socially desirable answers'
People tend to provide answers that they think the researcher wants to hear. In this way, there are indications that there is a difference between 'reported public opinion' and true public opinion. There are ways to construct surveys to minimize this effect. Within the CPSI team there is a lot of expertise available on this issue. For end-users, it means that expert advice may be called upon if needed.
5. 'The gathering of personal data is governed by national interpretation of the EU directive 95/46/EC'
The gathering and handling of personal data is strictly regulated by the EC and the individual member states, who each maintain their own guidelines. When using personal data, anonymity of the respondent must be guaranteed and your use of this data may need to be formally registered. Please check with your local authority for the guidelines in your country.
6. 'Some countries recognise more restrictions with regard to the processing and obtaining of special personal data than others'
So-called intimate or special data for example about one's sexual preferences or innermost thoughts is legally not allowed to be collected in some countries. Countries such as Spain, Greece, Portugal, Germany, Sweden and Denmark strictly regulate the obtaining of special personal data. Certain questions can therefore not be asked in some EU countries. The national implementation of EU Directive 95/46/EC needs to be consulted before a survey is conducted.
Project website:
The address for the project website is http://www.cpsi-fp7.eu The website has been developed and maintained by CESS. CESS will continue to maintain the website after CPSI's termination.
It currently contains general information on the project, such as a project summary, a project outline including the project videos, and descriptions of the partners. The website also contains links to papers and presentations related to CPSI, an overview of scientific journals and other websites relevant for this field, conference calendars, e-lectures and abstracts of CPSI deliverables.
The project's four main objectives were to:
- develop a conceptual model of actual and perceived security and their determinants;
- design a methodology to register and process security-related data;
- develop a data warehouse to store amassed data; and
- carry out an empirical proof-of-principle study to test the model, methodology and data warehouse.
In CPSI we focused on security related to 'everyday' crime, such as theft, assault and vandalism. The CPSI methodology, however, can be applied to other areas of security as well, such as terrorism or financial security. During the life of the project, we carried out many activities towards realising these objectives.
Project context and objectives:
The project has the following objectives:
1. develop a conceptual model to describe the relationship between actual and perceived security and the determinants that influence this relationship;
2. develop a methodology to collect, quantify, organise, analyse and interpret data on the factors described in the conceptual model;
3. develop a data warehouse to aggregate and store data from different sources;
4. conduct a validation study to test the model, methodology and data warehouse in the field.
The degree to which we have realised these objects will now be addressed in more detail:
1. Conceptual model
The development of the conceptual model has been completed with the completion of work package (WP) 5. The first steps in this process were undertaken in WPs 2.2 (Morphological analysis) and 2.4 (Background data). Together, these work packages yielded a preliminary conceptual model which formed the basis for:
1) the perceived security survey;
2) the information extracted from existing databases on actual security;
3) the media analyses and ultimately;
4) the structure of the data warehouse. In sum, this objective has been reached as planned.
2. Methodology
The CPSI methodology is an integration of techniques for data collection, data storage, and data analysis and interpretation. It is specifically focused on improving understanding of the relationship between perceived security and:
1) actual security;
2) security reporting in the media; and
3) interventions intended to improve security.
Originally, we had anticipated being able to develop a method that could be used off-the-shelf by end users to monitor and better understand perceived security in their area. However, during the project, we discovered that much of the work needed to implement the CPSI methodology cannot be structured in a one-size-fits-all manner. As a result, we have developed a standardized methodological concept, though for implementation it is necessary to customize many of its elements. In sum, the principle of this objective has been achieved: we have successfully designed a methodology with which to process security data from collection to analysis. However, the translation to implementation of our methodology in the field turned out to be more complex than we had anticipated.
3. Data warehouse.
We have successfully reached this objective. The data warehouse, designed to store and query for analysis information from the perceived security survey, the actual security databases and media analyses, is operational.
4. Validation study.
This objective has successfully been achieved. We have:
1) designed and implemented the data collection methods as planned;
2) collected all data as originally planned;
3) successfully loaded all the data into the data warehouse in preparation for analysis; and
4) conducted statistical analyses on the data from the validation study to assess the conceptual model and provide proof of the CPSI principle.
Project results:
WP 2 - Conceptual model
a. Objectives and summary of progress
The objective of WP2 was to investigate the conceptual boundaries and structure of the problem space concerned with actual security and perceived security. This included:
- the ethical and legal aspects of the problem field (WP 2.1.);
- the structure of the total problem space (including the interrelationship between factors) (WP 2.2);
- the investigation of public opinion as related to actual and perceived security (WP 2.3.); and
- a background study and a literature review (WP 2.4).
All of the sub-WPs were carried out successfully, giving an overview of the problem space to be studied, and bounding this space in a practical manner which has allowed for the formulation of a 'realistic' survey questionnaire which could be compared to the actual security data available through census and crime statistics, and other data. WP2 also resulted in the development of a prototype 'Comparative intervention efficacy' model and demonstrated that it is feasible to develop an analytic hierarchy of comparative goals and goal criteria for different types of interventions.
Ethical and legal issues (TNO)
a. Objectives and summary of progress
The objectives of WP 2.1 were to address relevant ethical and legal issues as they became apparent during the project (the legal issues were addressed in a written brief). This was done in close cooperation with the researchers, including participation in project workshops. The overall objective was to develop a common understanding among the researchers about the ethical issues and possible ways to deal with them. This ensured a more all encompassing view on the research project.
A number of ethical issues were identified:
1) In the first morphological analysis, the available group of people was a bit biased (mainly Caucasian, highly educated men). In the following workshops more attention was given to the diversity of the group of experts and their fields.
2) The ethical dimension of different definitions of actual and perceived security, and the problems with gathering data on crime and on personal opinions / dispositions were identified.
3) Based on the report on the Dutch legislation on data protection, it was decided that the survey data, as analyzed by the researchers, were not personal data. It was, however, necessary to recode should too few respondents be in a category. For future use (local) governments are warned that making the geographical analysis units smaller might lead to problems regarding privacy and personal data because indirect identification of respondents is possible.
4) It was decided that involvement of the security industry in the project could lead to perverse results as the industry needs people to feel insecure so that they purchase security technologies.
5) Users of the CPSI system are warned not to upload data from very different sources such as social security, car insurance etc. Combining data from very different sources risks violating contextual privacy. Advanced data mining techniques cause problems with regard to the informed consent of respondents.
6) With regard to the legal issue the most pressing issue is that different EU countries have different national implementations of the data protection directive. This means that some survey questions especially those concerning sensitive issues cannot be asked in every EU country.
b. Significant results
- A report on the Dutch data protection legislation with regard to the CPSI survey was completed.
- A report on the national interpretation of the Eu Data Protection Directive is completed.
- The final report on Ethical and Legal issues is delivered.
- An article was written that is already available online: Van Gorp, A. C. and Van der Molen S. (2009), Parallel, embedded or just part of the team; ethicists cooperating in a European security research project, Science and Engineering Ethics, published online at DOI 10.1007/s11948-009-9187-5.
- A manuscript was written for a special issue in Security Journal: Van Gorp A. C. Is an informed consent form sufficient: Ethical and legal issues in perceived security research.
- Interviews have been held with about two thirds of the project participants.
c. Deviations from project plan
None
d. Statement of the use of resources
The total budget for WP 2.1 was increased during the first reporting period. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
Morphological analysis (FOI)
a. Objectives and summary of progress
The objective was to structure and investigate rigorously the dimensions of the total problem space addressed in the project. Each of the WP's required specific expertise in various disciplines. Given these various backgrounds and the need for a common conceptual framework, a series of problem structuring workshops were carried out in order to create among the project members - and outside expertise where needed - a common problem structure and modelling framework, as well as to specify the conditions for the future work packages.
The workshops utilised computer-aided morphological analysis, a non-quantified modelling method developed at FOI. The workshops resulted in a number of morphological inference models defining the total (dimensional) problem space that the project addressed as well as the interrelationships between the problem variables.
b. Significant results
The initial morphological fields for actual security and perceived security produced multi-dimensional typologies which were wider and more varied than were found to exist in the research literature, especially as concerns the relationship between 'ethic' (objective, statistical) variables and 'emic' (subjective, motivational) variables.
The dimensions of perceived security could be reduced to three key variables:
(a) types of threat;
b) locations; and
c) a set of demographic parameters without significant loss of information.
c. Deviations from project plan
None
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings. Note that the budget for WP2.2 was decreased due to the fact that less budget was needed than had originally been anticipated. The budget that resulted from this decrease was transferred to various other beneficiaries for work in other WPs. This decrease in no way affected the execution of the work as it is described in Annex 1.
Public opinion (University of Kent)
a. Objectives and summary of progress
This WP endeavoured to develop an understanding of how public opinion relates to and influences security. In particular, we were interested in developing an overview of how attitudes towards security influence people's behaviour and indirectly bear upon public policy. This sociological analysis should contribute to the development of a conceptual model of perceived security.
The study has resulted in a 'Review of existing surveys on public opinion of security' which includes material on:
1) what surveys tell us about the EU public's perception of insecurity;
2) how opinion surveys relate to security; and
3) opinions on specific threats such as immigration, climate change and economic crisis.
b. Significant results
- Review of existing surveys on public opinion of security
- Inventory of European crime surveys and crime survey summary
- Recommendations for future research
c. Deviations from project plan
WP 2.3 started late due to issues with the moment at which University of Kent was able to begin work. However, this has had no negative impact on other tasks, resources or planning.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
Background data (SCP)
a. Objectives and summary of progress
The objective was to examine the state of the art, i.e. to generate insights on the basis of literature about the determinants of perceived security, the relation between actual and perceived security. These insights have contributed to the development of a conceptual model of perceived security.
Besides being a broad literature study, WP2.4 addresses the important links between individual and contextual factors and the fear of crime, along with coping strategies. These individual and contextual determinants are gathered in a conceptual model explaining fear of crime. The process of data collection is described and the conceptual model is then translated into indicators on which data can be gathered either from official registrations or from a local survey newly designed for the present purpose.
b. Significant results
- A fear of crime literature review
- List of factors for designing the methodology identified in the literature review
- Comparison with the factors developed in the morphological analysis
c. Deviations from project plan
None
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
WP 3 - Data warehouse
Overview
The development of a data warehouse is necessary to create a central environment for all the data needed to make analysis of the security work field happen. To conduct the analysis work, we need to integrate all the 'security related' data from the location of the validation study and distribute the information that is derived from the 'security related data' to the authorised users.
Information analysis report (Sogeti)
a. Objectives and summary of progress
The objective of WP 3.1 is to describe all the requirements for the data as needed in order to make the analysis work. It contains a summary of all definitions of the data captured in the system.
The following topics were addressed:
- Identification of the data dictionary
- Design of the data models for data storage and reporting
- Design of the loading process including the data delivery interfaces
Technical design documents were delivered as input for the construction of the database environment itself. This WP has been completed.
b. Significant results
All subjects in the CPSI model have been addressed and are integrated into one information model for the data warehouse.
c. Deviations from project plan
Originally, the media analysis was designed to be delivered by TEMIS. The TEMIS data were aggregated according to the nature of the desired analyses. However, for the pilot we opted for the inclusion of the more 'raw' RSS feeds generated by JRC.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded, note though that the total budget for WP 3.1 was increased during the first reporting period; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
Database environment (VLC)
a. Objectives and summary of progress
The objective of WP 3.2 is to construct the database environment together with the procedures to load data and the tooling. The activities, which were executed in the first phase, mainly concerned the technical preparation of this work package. For the second phase the technical design documents from WP 3.1 are used as input. This work package has been completed.
b. Significant results
- The required hardware and software were selected, bought, installed and practiced with.
- The data warehouse was designed and constructed.
- Data was loaded into the data warehouse.
- Reports were built.
- Data extractions were generated for analytical purposes.
c. Deviations from project plan
The construction of the pre-processing step was skipped, for the following reason. Although, a standard interface was designed, in practice it is hard to conform to that interface. It requires a lot of manual preparation for the delivering party. Therefore, it was decided not to automate that step. This has no further impact on other work in either this or other WPs.
d. Statement of the use of resources
The total budget for WP 3.2 was increased during the first and second reporting periods. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Another part was spent for purchasing software necessary for the data warehouse. Minor costs were incurred for workshops and project meetings.
WP 4 - Data collection / validation
WP 4.1 and 4.2 (SCP)
a. Objectives and summary of progress
WP 4.1 contains a list of the databases acquired for inclusion in the data warehouse. In order to select the relevant variables within these databases, we used the outcome of the morphological analyses and the literature review.
WP 4.2 contains of two elements: selection of a city for the pilot study and designing a questionnaire. After ample consultation within the consortium the decision was made to conduct a survey on perceived security in only one city: Amsterdam. This city has its own department for research and statistics (O+S Amsterdam) and has a permanent survey among the Amsterdam population on safety and security. O+S were prepared to cooperate with us and permitted us to link our survey to their own.
In consultation with O+S Amsterdam we designed a questionnaire. The research period took three months and the netto-response was 49 % (N = 1222). Some 5 % of the response seemed of no use because of incomplete filling-in of the questionnaire.
b. Significant results
To validate the conceptual model, data was needed on the social context of the respondents participating in the local study (WP4.2). We needed information on the level of actual crime, demographical and social composition and on the physical structure of the neighbourhoods people live in. Since our local study was carried out in Amsterdam, we contacted O+S (the Department for Research and Statistics of the Municipality). O+S has an independent position within the municipality of Amsterdam that can be compared to that of Statistics Netherlands (CBS) within the national government. The variables listed are complementary to the survey data collected in Amsterdam. A list of variables available from O+S to validate the conceptual model covers police statistics, demographical and social composition and physical structure.
The survey in Amsterdam (4.2) went well; the cooperation with the research organization O+S was excellent.
c. Deviations from project plan
All went as planned; no deviations.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings. Note that the budget for WP's 4.1 and 4.2 was decreased due to the fact that less budget was needed than had originally been anticipated. The budget that resulted from this decrease was transferred to various other beneficiaries for work in other WPs. This decrease in no way affected the execution of the work as it is described in Annex 1.
WP 4.3 (Temis, JRC)
a. Objectives and summary progress
The main goal of the processing of the textual (media) data in the CPSI project is to extract key content from various sources in English and Dutch and to export it for populating the CPSI data warehouse with security concepts alone or in relation to geographic locations concerned. The added value of using text mining technologies resides in the possibility of extracting, from unstructured text, key content from textual media content.
The extracted information provides various analysis dimensions as factors for media salience of security related issues. For example:
- timely analysis of security related topics in news articles;
- analysis of geographic distribution of security related topics in news articles.
b. Significant results
A pilot application was set up that annotated documents for a period of seven months starting from November 2008 until the end of May 2009. During this period, more than 23 000 documents in English and Dutch were annotated. The export files of the structured information for import into the data warehouse contain circa 73 200 lines in the defined format: document identification; document date; source name; crawling category; document position; security concept; extraction type; location.
Achievements for WP 4.3 are the following:
- A specific knowledge component for extracting security concepts in English and Dutch was built. The structure of this component uses the conceptual model coming from the morphological analysis WP. The following categories of security issues have been retained for the pilot application: discriminations, drugs, economic downturn, property crime, terrorism and violence and vandalism.
- The media texts crawled by EMM of JRC are annotated with the software of TEMIS for information extraction and text analytics called Luxid®.
- A specific export procedure that exports the structured media content for the import into the CPSI data warehouse was developed. For example, a relation between the security issue 'murdered by Islamic militants' and the location 'south-central Somalia' extracted from the text 'murdered by Islamic militants in south-central Somalia' is exported with the reference to the media text from which it comes.
- An access to the user interface of Luxid® and set up analysis scenarios for exploiting the media content was provided. For example, negative sentiment words (like accuse, expensive, greatest difficulties, worst, conflict, fear, unable, error, etc.) or opinion verbs (like criticise) can be put in correlation with security issues by using the analysis tools of the platform. This permits identification of key content in the media texts regarding the perception and positions toward security, for example: 'Advertising firms censor signs for fear of vandalism', 'Report blames racism for rise in youth violence', 'US to blame for much of Mexican drug violence'.
Depending on the scope of investigations, the pilot system can be extended to other countries and languages, other information sources and other extraction needs. For example, in order to analyse the opinion of countries, organisations, official institutions, influential persons, governments etc. or individual persons towards security issues, one of many future possibilities is to expand the knowledge components to allow for the extraction of opinions in relationship to the opinion holders. The extraction of opinions expressed in the media may constitute in the future a component of an information intelligence platform supporting predictive modelling tasks.
The following tasks may be supported:
- Provide an overall landscape of security perception by different media sources: Identify the tonality of a media text and the communication strategies of a journalist or globally of a media source that are set up when talking about security topics.
- Identify groups of influence: Identify social networks according to the point of view on security issues shared by social movements (organisations, official institutions, influent persons, governments etc.).
c. Deviation from project plan
None.
d. Statement of the use of resources
The total budget for WP 4.3 was increased during the second reporting period. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
WP 4.4 (SFU - CESS)
a. Objectives and summary progress
Work in WP 4.4 and its deliverable took place in a long-term frame and on different levels, including literature study, desk research, mixed working groups with end user involvement (including cross-fertilization with ESRIF WG10 and WG11 according to the CPSI dissemination plan). Report on cultural issues on two levels (society and political selection of risks according to the Douglas / Wildavsky model) were concluded, as well as a report on the country case studies conducted (AT, BG, FR, GE, IT, NL, SE, UK - reflecting the composition of the end user advisory group), summarising and illustrating results; paper on culture-sensitive communication guidelines to prevent or reduce gaps between citizens' actual and perceived security.
b. Significant results
High social fear of crime countries (crime perceived as a problem 'out there', such as prominent in Austria, the Netherlands and Sweden) have under-average personal fear of crime but an over-average number of offences reported to police, indicating low actual security. Personal fear of crime in this type of country is also lower than the victimisation level would suggest.
High personal fear of crime countries (crime perceived as an individual(ised) issue, such as prominent in Bulgaria and Italy) can still have average victimisation levels. They have however an under-average number of offences reported to police (high actual security) and average or under-average social fear of crime. Personal fear of crime is thus more detached from actual security than social fear of crime. More social fear of crime is associated with less personal fear of crime. This leads to the hypothesis that social fear of crime reduces personal fear of crime.
The EU itself does not generally seem to be a source of citizens' fear of crime in their own country; France, Germany and the UK appear to be exceptions. At the same time, citizens perceive national interventions to be most suitable to enhance their security against crime; the exception is Italy, where citizens appear to prefer EU solutions also to national security problems. For the UK, statistical analysis, while supporting no strong interpretation, points to a possible interpretation in a similar direction.
Despite public support for EU decision-making and action against crime, the EU itself is generally not perceived as a locus of successful interventions to enhance citizen security against crime; the exception is Italy. For the UK again, statistical analysis, while supporting no strong interpretation, points to a possible interpretation in a similar direction.
CPSI country analysis so far has shown no consistent association between acceptance of technological solutions for security problems (such as CCTV) and relationship between level of societal and of personal fear of crime, level of victimization, relationship between felt and actual personal / social security and victimization.
CPSI country analysis however has shown consistent association between acceptance of CCTV and cultural attitudes towards technology: countries in which technology is interpreted as part of the security problem (e.g. critical infrastructure protection, information technology as object of offence and source of insecurity), public CCTV acceptance is lower than in countries where technology is interpreted as part of the solution (e.g. information technology as a foundation for coordinated, efficient prevention and response).
On the level of cultural selection of risk, analyzed by Member State security research policies / programmes, in cultural factors were found to generally have the strongest positive impact on (developing) a comprehensive approach to citizen security at the national level; they were found to generally have the strongest evidence of negative impact on splits in thematic thrust (such as society vs. technology-centred security interventions).
Citizens' knowledge is the key factor for their perception of security and interventions. Knowledge also mediates between felt and actual security. This study thus highlights the need for continued interpretation of security as an information and cognition issue. While cultural factors have a positive impact on solutions to security problems when these are perceived by the public as national problems, the EU must minimise its negative impact on solutions to security problems when these are perceived by the public as 'European problems'. At the same time, security was found to continue to be mainly a national cultural value rather than representing a value common to European citizens.
Moreover, knowledge turned out to be the first-rank effectiveness criteria for security-enhancing interventions. Strong knowledge and interpretative contexts present on the national level are a cultural factor that decreases citizens' individual perception of insecurity. That means social fear of crime reduces personal fear of crime. Actual insecurity particularly increases social fear of crime (perception of crime as a problem 'out there') but decreases personal fear of crime (perception of crime as an individual concern).
Interventions should in the first place focus on a comprehensive definition or articulation of risk based on exchange of knowledge. Policy interventions based, in contrast, on perceived security have been found to increase citizens' fear of crime.
While cultural factors have a positive impact on solutions to security problems when these are perceived by the public as national problems, they have a negative impact on solutions to security problems when these are perceived by the public as 'European problems'.
Effective communication strategies to reduce gaps between actual and perceived (in)security need to determine - even before a choice of concrete tools or measures - the cultural context in which the communication-based intervention is to be implemented and organise planning according to whether personal fear of crime or social fear of crime is the predominant societal context, for instance:
- In countries with a social fear of crime culture, communication-based interventions typically lower citizens' perception of (in)security, and this type of gap between actual and perceived security is mainly due to styles of social meaning making as rooted in the social context.
- In countries with a personal fear of crime culture, communication-based interventions typically increase citizens' perception of (in)security, and this type of gap between actual and perceived security is mainly attributable to disproportionate media reporting.
c. Deviation from project plan
In order to avoid duplication, the risk of which only became visible during the actual work, project review meetings concluding WP4.4 activity concentrate on the national aggregation level, covering two perspectives: Political culture (society) and political selection of risk (politics of citizen security). The selection of these perspectives reflects the two-fold research interest of CPSI: citizens' perceptions of security on the one hand and interventions on the other. Further, to avoid duplication with WPs dedicated to survey and quantitative analysis, WP4.4 concentrated on the national level of aggregation and include eight country reports (Austria, Bulgaria, France, Germany, Italy, the Netherlands, Sweden, United Kingdom - reflecting the composition of the end-user advisory group).
Work on the 'definition of culture-sensitive media strategies for threat / security situation communication and public awareness-raising in order to close gaps between actual and perceived security' was postponed because project reviews showed that links with WP4.3 (University of Kent) needed to be strongly explored. The development of communication guidelines was consigned to WP4.4 because the results of the culture analysis yielded several points to directly connect to, which had not been anticipated in the DoW.
Work on some parts of WP4.4 showed that cultural issues as analysed in the WP4.4 framework (method of security culture) do not provide enough information to produce envisaged typologies such as 'Typology of security perception differences between different typified locations (cities, towns, regions, country)'. During project review, it was therefore decided to cover these issues in the final deliverable on the basis of availability of cross-WP data and research results.
Introduction of WP4.4 results into the conceptual model though the data warehouse could not be implemented as planned because several attempts to construct a translation mechanism between the cultural data available and the quantitative information able to be grasped by the data warehouse failed.
d. Statement of the use of resources
The total budget for WP 4.4 was increased during the first reporting period. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Minor costs were incurred for workshops and project meetings.
e. Actions taken
Work relating to culture-sensitive media strategies were conducted together with University of Kent, following WP4.3 course of action and responsibility. Integration of WP4.4 cultural data and research results into to conceptual model was accomplished though morphological snalysis.
WP 5 - Analyses
a. Objectives and summary of progress per task
The objective was the development of a series of inference models describing the empirically based relationships between the parameters originally mapped out (theoretically) in WP 2. These models will map out solution spaces to the total problem space developed in WP 2, and should allow for multiple drivers both in order to make comparisons between different countries (where applicable), cultural groups and social strata, and to allow any parameter in the problem space to be treated as an independent variable.
This objective has been achieved. We have contributed to the work packages preceding WP5 in such a manner that we are assured that the necessary theoretical models (formulated in conceptual structural equation models (SEM)) are in place and that the collected data is of the nature and detail that testing of the CPSI methodology was possible.
The WP breaks down into three separate tasks: quantitative analysis, qualitative analysis and synthesis. The main task of this work package is to analyse the data resulting from earlier WPs and synthesise the findings of WPs 5.1 and 5.2. in WP 5.3.
WP 5.1 Quantitative analysis
a. Objectives and summary of progress
The aim of this WP 5.1 'Quantitative analysis' is to demonstrate how quantitative analyses can be performed with the data collected through the CPSI method. In a pilot study in Amsterdam (the Netherlands), the data collection and analyses were demonstrated and a general model was tested. Based on the results, a standardised EU monitor is presented that can be used to concisely and fully measure perceived security. In order to be able to do that the following procedure was used.
An overall CPSI model was developed, which hypothesises how actual security and perceived security are causally related and how this relationship is influenced by background factors of citizens and media salience of security topics at the neighbourhood level. By using data from different sources, aspects of this model were tested at the neighbourhood level.
Before testing the overall model, variables to be used in this analysis need to be identified and selected. Four sources of data are used:
1) perceived security data collected by means of a survey of individual citizens (WP 4.2 Perceived security: questionnaire);
2) actual data procured from police statistics at a neighbourhood level (WP 4.1 Actual security: databases);
3) background data on neighbourhoods at a neighbourhood level procured from the bureau of statistics (WP3.1 Design and construction data warehouse, WP3.2 Integrate databases); and
4) media salience of security topics procured through the media analysis procedure (WP4.3 Media content analysis). Each data set was scrutinized separately for data integrity and fitness for purpose.
b. Significant results
The brunt of the work dealt with reducing the number of variables (questionnaire items) of the survey. The original questionnaire consisted of more than 500 items, which cannot all be used in the testing of the overall model at the neighbourhood level. A selection procedure was used to reduce the number of variables to 59 and thus reduce the questionnaire items to 'a standardised EU monitor'. This selection procedure took a four-step approach. First of all, items that were not relevant for CPSI were omitted; these were items that were used by the Amsterdam Department of Research and Statistics in a standard security monitor (VMR) that was part of the questionnaire but were of no relevance to the CPSI pilot. Secondly, a data reduction procedure was used to narrow down the number of items, based on the theoretical model of perceived insecurity (WP 4.2 Perceived security: questionnaire). Thirdly, an empirical procedure was used to further reduce the number of items. By factor analysing the remaining items and identifying the highest loading variables on the extracted factors, an empirical base was achieved for the hypothesised theoretical model of perceived insecurity. Finally, these items were constructed into scales at the aggregate neighbourhood level and ready for use to test the overall CPSI model. We suggest for the future implementation of the CPSI method using the 'standardised EU monitor'.
For testing the overall CPSI model of perceived insecurity, the survey data at the aggregate neighbourhood level was merged with the actual security data, data on the salience of media topics, and background data at the neighbourhood level. The resulting file was used for the testing of the overall CPSI model. In this analysis, hypothesised relationships were tested and explored and gave rise to relationships of interest and practical use for future CPSI implementation. For example, perceived security was only partially explained by actual police statistics on crime and demographics of the neighbourhood. However, neighbourhood quality ratings and social cohesion appeared to be more powerful predictors of perceived security: neighbourhoods that are rated as secure and cohesive report lower levels of fear. Media salience is also related to perceived security in several ways.
In general, the pilot demonstrates that the CPSI methodology can be used as intended. Implementation of the methodology at local or higher levels of public administration is feasible and allows for a unique integration of data from different sources. As such, it can supplement policy assessment by testing of interventions and enhance transparency of citizen security.
c. Deviations from project plan
There were no deviations from the project plan.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for RTD-related workshops and project meetings.
WP 5.2 Qualitative analysis
a. Objectives and summary of progress
The qualitative analysis of the CPSI Security Survey concerning perceived security, victimisation and other project data was meant to give a broad overview of some of the core results in an easy-to-use and easy-to-understand form. It is not a statistical study in the sense of identifying 'statistically significant' relationships or correlations. This was left to WP 5.1: Quantitative analysis.
The form of the qualitative analysis is a number of computer-supported morphological inference models, by which the user can examine and compare relative demographic differences in threat perception and feelings of security, reported victimisation and the use of media sources. This study, together with the WP 5.1 (Quantitative analysis) provides a basis for drawing conclusions concerning relationships between actual and perceived security (WP 5.3 - Synthesis), which is one of the prerequisites for identifying appropriate security interventions.
The qualitative analysis was also concerned with:
1) identifying principle factors involved in media-based opinion building as well as examining the feasibility of mapping out the complex structure of interrelations between these factors (from WP 2.3); and
2) a comparison of national (public) security cultures based on a number of national security indicators compiled by WP 4.4.
b. Significant results
Three of the most interesting indicators for studying the survey material on perceived threat or insecurity in a population were gender, age and immigrant status. Some general trends were uncovered by the morphological models developed around these indicators. Among other things, the 'threat' that the survey population was clearly most worried about is 'discrimination'. This was the case across gender, age groups and immigrant status. The only slight deviation is that age group '15-25 yrs' seems to be the most concerned about discrimination of all the age groups. (However, this age group was the smallest sample). The general rank order of the most worried about types of threats is:
1) discrimination;
2) crime on the streets;
3) economic crisis and unemployment; and
4) lack of integration by immigrants.
Also notable is that as concerns general feelings of insecurity, women tend to feel more insecure than men (if only marginally), feelings of general insecurity tend to decrease marginally with age, and there is no stated difference in general feelings of insecurity between Dutch born and Immigrant responders.
As concerns media-based opinion building, it is fully feasible to utilise the morphological modelling process to identify and hypothesise the relationships between demographic variables and variables concerning media use and influence, by which these hypotheses may be tested. However, as the work revealed, every relationship is virtually a study in itself. Thus the main utility of the prototype model was its structuring of the possible types of relationships involved.
c. Deviations from project plan
There were no deviations of the project plan.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for RTD-related workshops and project meetings.
WP 5.3 Synthesis
a. Objectives and summary of progress
The objective of WP 5.3 Synthesis is to synthesise the results of WPs 5.1 Quantitative analysis and 5.2 Qualitative analysis. The qualitative and quantitative approaches are assumed to be complimentary. This WP shows how the two approaches can mutually address the systematic analysis of issues concerning citizen security.
The quantitative analysis focused on the available data obtained in WP 4. The previously identified causal relationships between the relevant security factors were tested for the strength of the relationships. Various statistical techniques have been used among others, correlation, factor, reliability analysis and structural equation modelling.
The analyses gave insight into how actual security relates to perceived security and how this relationship is mediated or moderated by media (public opinion) indices and cultural or background factors. Furthermore, the data was analysed for psychometric purposes at the individual level, for the analysis of relationships between actual security and perceived security the data were analysed at the neighbourhood level. No longitudinal data were obtained within the time frame of the project, thus the effect of interventions could not be tested.
The second approach is qualitative in nature; this analysis described the results of the computer-aided inference models describing solution spaces, and their consequences. This activity focused on the development of a series of inference models describing the empirically based relationships between the parameters originally mapped out (theoretically) in WP2. These models map out solution spaces to the total problem space developed in WP2, and allow for multiple drivers both in order to make comparisons between different countries (where applicable), cultural groups and social strata, and allow any parameter in the problem space to be treated as an independent variable.
A number of modelling workshops were carried out with the project beneficiaries, University of Kent and the Centre for European Security Studies, and including specialised expertise where needed. The workshops utilised such methods as computer-aided morphological analysis (MA), multi-criteria decision analysis (e.g. AHP), quality function deployment (QFD). Trained facilitators knowledgeable in the methods employed facilitated the modelling groups.
In order to achieve a synthesis of both types of analyses (qualitative and quantitative) we made use of the overall model used in the quantitative analyses in WP 5.1. This basic model postulates that perceived insecurity is dependent on actual security and that this relationship can be mediated by either media or background factors.
First, we described the communalities and discrepancies between the overall concepts. We then described how the two approaches converge or diverge with respect to hypothesised relationships between the concepts in the overall CPSI model. Finally, we discussed how the overall model can facilitate understanding and how the CPSI method can facilitate evidence generation for all sorts of queries about citizen security.
b. Significant results
Our synthesis of analyses demonstrated that qualitative and quantitative approaches are complementary in analysing complex and comprehensive issues such as citizen security. As such, a broad spectrum of questions can be asked and tested: concepts are defined, relationships are identified, context is given, and technology is made available to collect, store, and analyse data. However, intensive interaction with stakeholders is crucial to select the specific questions to be addressed. Only stakeholders, such as a security officer or a policy maker, can truly recognize the urgency of which questions to ask in the domain of citizen security and prioritise research efforts.
Based on the four factors of the CPSI model, hypotheses can be tested at a high level of abstraction. At the same time, stakeholders might inspire testing of more specific hypotheses. In general, all relationships between factors of the model can be studied using the CPSI method. For example, the end user of CPSI can use the method to examine the following relationships:
- measure the level of perceived security at any given time;
- measure the level of perceived security in any neighbourhood;
- measure the relationships between perceived security and other factors (such as actual security, media, or background) at any given time;
- measure the change in perceived security over time in any neighbourhood;
- measure the change in relationships between perceived security and other factors over time;
- measure differences between (types of) neighbourhoods in terms of perceived security and actual security;
- examine the effect of changing salience of media reports on perceived security over time.
The CPSI methodology allows for testing the effects of implemented interventions on actual and perceived security at the neighbourhood level. This implies longitudinal data collection at neighbourhood level, i.e. measuring before and after implementation of the intervention.
c. Deviations from project plan
There were no deviations of the project plan.
d. Statement of the use of resources
No person month limits according to Annex 1 were exceeded; the main costs incurred were personnel costs. Minor costs were incurred for RTD-related workshops and project meetings.
WP 6 - Dissemination
a. Objectives and summary of progress
A CPSI web portal was established (see http://www.cpsi-fp7.eu online), featuring a video with basic information about the project, event announcements, WP abstracts, PowerPoint presentations, papers, etc. It was updated regularly. The website contains information on project-related workshops and symposia (or parts of symposia) held or taken part in by CPSI project members, the project and its methodology, discussion of intermediate results, reference to (experts from) other FP7-SEC projects of relevant parts of scope (e.g. CAST, FORESEC). CPSI dissemination actions (such as appearance at symposia) were coordinated on the basis of a regularly updated dissemination plan and a conference calendar, identifying relevant opportunities for dissemination.
An international final symposium was held, with 68 participants from 15 countries and media reporting. E-lectures were recorded, edited and made available on the CPSI project website.
b. Significant results
CPSI was presented on the Third European Security Conference Initiative (ESCI) in 2008, organised by the WWEDU/Centre for European Security Studies and attended by some 100 participants from academia, industry and policy. CPSI partners have made relevant contributions to several ESRIF working groups, including WG10 and WG11, where CPSI research results were also exploited for the purpose of producing themes for further security research. In addition, research for CPSI WP4 deliverables (especially report on cultural issues) was disseminated into ESRIF WG10 contribution to part 2 of the ESRIF final report.
CPSI results were communicated to media representatives during the CPSI final symposium and a DVD was produced and delivered, containing all deliverables and other documents.
c. Deviations of project plan
The end-user advisory group made itself available for review of CPSI deliverables from an end user point of view, which was originally not foreseen and is a positive deviation. Depending on the time frame for and kind of this review feedback, it now and again impacted the schedule for submission of deliverables and required some additional text work and person months resources.
WP6 objectives were amended by ESRIF-related dissemination objective as ESRIF only started after CPSI DoW was finalised and turned out to entail working groups relevant for and open to input from CPSI. This was a positive deviation and did not considerably impact resources. Rather, it opened up space for discussing CPSI intermediate results and still better designing them to end-user and European needs.
The CPSI summer school could not be held although it was fully planned and stuffed and advertised throughout Europe. There was not sufficient turnout on the students' side. Elements from the summer school curriculum were resumed at the CPSI final symposium, and e-learning content for the CPSI website that had been planned to be gained from the summer school was generated from appropriate presentations at the CPSI final symposium.
d. Statement of the use of resources
The total budget for WP 6 was increased during the first and second reporting periods. As a result, the person months estimated in Annex 1 were exceeded. The main costs incurred were personnel costs. Major costs were incurred for organisation of the final event.
The CEUSS budget increase addressed a recalculation related to the transfer from WWEDU to SFU, resulting in more personnel costs for WP6. Additional budget was transferred to CEUSS in order to cover dissemination / reporting activities in relation with the CPSI final symposium that had not been foreseen (such as reports in stakeholder magazines and a full-scale conference report). Additional budget was allocated to Sogeti, to cover costs for the designing and developing of the final CD-ROM.
Minor costs were incurred for workshops and project meetings.
Potential impact:
The dissemination plan is described in detail later in this report and contains an overview of all dissemination activities carried out in the second reporting period and shortly after the termination of the project.
The exploitable foreground in CPSI consists mainly of two aspects. First and foremost is the implementation of the CPSI system in end users' organisations. This is indeed the primary goal of the CPSI project. The purpose of implementing this system would be to help end-users:
1) increase their insight into perceived and actual security in their area, the determinants of this security and mediating factors; and
2) to increase the effectiveness of the allocation of resources (financial, personnel, material).
Indeed, a better understanding of what affects security in the desired direction helps those responsible for strategic and operational planning decide what interventions to develop and implement for which locations. The second exploitation possibility we anticipate is in regards to providing support and advice to relevant organisations based on the knowledge developed in CPSI. This could be, for example, consulting in the area of statistical analyses suited to complex data sets, similar to that used in CPSI, or in the area of cultural considerations as they relate to security. In this way, it is possible for us to assist parties who have their own extensive registration systems (and hence to not need or want the CPSI system as a whole), but who cannot maximise the information possibilities of their systems. To illustrate, TNO recently visited the city of Rotterdam to discuss their registration system. Though their security information registration system is very good and extensive, they have two areas in which we may be able to advise them based on knowledge developed in CPSI:
1) how to statistically combine and interpret their data on actual and perceived security and interventions; and
2) possibilities for advanced analysis using structural equation modelling.
Another concrete possibility is to participate in a pilot study, currently being planned by the Dutch Ministry of Internal Affairs and Kingdom Relations, regarding the toolkit of security perception and the 'Table of 12,' which is an overview of 12 types of security interventions that can be implemented in the field.
In the case of both exploitation possibilities, we anticipate the 'target audience' is primarily local, regional and national governments. Secondary target audiences include organisations responsible for crisis management such as Sweden's MSB (Swedish civil contingencies agency) or the Red Cross. CPSI has had excellent relations with the Austrian Red Cross throughout the project.
For both exploitation possibilities, we expect the first exploitation within the next year.
In principle no additional development of the CPSI system is needed before it can be implemented in the end-user's organisation. However, as the CPSI system is not an off-the-shelf product that can be implemented as a sort of plug-and-play application, implementing CPSI in an end user's organisation will comprise some custom fitting of the system to the needs and constraints of that user. Examples of custom characteristics are:
- cultural considerations regarding what questions can be asked and what cannot (e.g. political preference, income, ethnicity);
- specific considerations relating to the security culture of the particular country (e.g. centralised security policy versus de-centralised);
- coding of criminal offenses;
- incorporation of information specifically requested by the end-user;
- integration with existing registration and software systems;
- translation of keyword list for media analyses and choice of media sources;
- translation of standardised EU security monitor;
- development of software routines to automatically perform statistical analyses according to the needs and desires of the end-user.
There are different aspects of the impact of the CPSI system and its implementation. First, in the short term, it will provide a better snapshot of security related concepts than is currently available in most areas. In the short term, questions can be answered such as: What is the relation between media use and perceived security by neighbourhood? Or Which types of crimes are most related to a low perception of security? Second, in the long term, it will be possible to monitor trends in security. Then, questions can be answered such as: What is the effect of 'yuppification' of a neighbourhood on crime? On perceived security? Or When and where does camera surveillance work best? Third, the overall impact of a widespread use of the CPSI system is a more unified, systematic and integrated registration and analysis of security data. This supports a better overview of best practices, which benefits not only the primary user of the system, but, through a sharing of experiences and information, also the common good. This impact cannot be qualified as such and may take a number of years to truly 'mature.' However, the sooner we start implementing the system, the sooner we can reap the benefits.
In addition to the positive effects the output of a project may have on public security, there may also be a negative societal impact. This addresses things such as the effect of the product on privacy, civil liberties and human rights in terms of the violation of ethical, moral or legal rights and principles. As regards the CPSI project, during the execution of the project we made every effort to ensure the privacy of the participants through the use of suitable anonymity procedures and the aggregation of the data to the group level before conducting analyses, in order to insure that participants' identity could be neither directly nor indirectly ascertained. Furthermore, in the dissemination of the empirical study, we were careful to not make known the individual neighbourhoods in which data were collected. In this way, we hoped to provide additional 'protection' for the participants.
Aside from how we dealt with concerns about the societal impact of the execution of the project, we also considered the broader scope of what the actual implementation of the CPSI system can mean for concerns about privacy and other essential values. CPSI has designed a system to monitor security-related information for the betterment of society as a whole and the citizen as an individual. In order to assist the end-user in applying the system in a responsible way, we have drawn up an overview of recommendations and caveats, which we recommend the end-user follow. This overview is included in the final project CD-ROM. These include:
1. 'Privacy and informed consent issues'
Informed consent must be acquired when collecting data from respondents, such as is the case with the standardised security survey. It is essential that respondents understand what is being asked of them and what will be done with the information they provide. The nature of the informed consent form in your country may be legally regulated, so check with local authorities for more information. We recommend care when using any data at the level of the individual citizen or small group; protecting individuals' data and their right to privacy is paramount.
2. 'Use of surveys: Statistical significance of results and reporting'
For those individuals with a layman's knowledge of statistics, it is often difficult to correctly interpret results and put them in perspective. For this reason, we advise caution when reporting results, particularly when these results are used to set the political agenda.
3. 'Secure protocols are needed to secure personal data'
The end-user should consider procedures to securely upload the data. At this moment there are no secured lines or protocols inherent in the CPSI system but especially if the data warehouse is used for monitoring and data are uploaded regularly then there need to be secure lines or protocols.
4. 'Socially desirable answers'
People tend to provide answers that they think the researcher wants to hear. In this way, there are indications that there is a difference between 'reported public opinion' and true public opinion. There are ways to construct surveys to minimize this effect. Within the CPSI team there is a lot of expertise available on this issue. For end-users, it means that expert advice may be called upon if needed.
5. 'The gathering of personal data is governed by national interpretation of the EU directive 95/46/EC'
The gathering and handling of personal data is strictly regulated by the EC and the individual member states, who each maintain their own guidelines. When using personal data, anonymity of the respondent must be guaranteed and your use of this data may need to be formally registered. Please check with your local authority for the guidelines in your country.
6. 'Some countries recognise more restrictions with regard to the processing and obtaining of special personal data than others'
So-called intimate or special data for example about one's sexual preferences or innermost thoughts is legally not allowed to be collected in some countries. Countries such as Spain, Greece, Portugal, Germany, Sweden and Denmark strictly regulate the obtaining of special personal data. Certain questions can therefore not be asked in some EU countries. The national implementation of EU Directive 95/46/EC needs to be consulted before a survey is conducted.
Project website:
The address for the project website is http://www.cpsi-fp7.eu The website has been developed and maintained by CESS. CESS will continue to maintain the website after CPSI's termination.
It currently contains general information on the project, such as a project summary, a project outline including the project videos, and descriptions of the partners. The website also contains links to papers and presentations related to CPSI, an overview of scientific journals and other websites relevant for this field, conference calendars, e-lectures and abstracts of CPSI deliverables.
