ICVS

Methodology

Main characteristic of the ICVS is standardization. Data can only be compared if all aspects of the study are standardized. Two types of methodologies have been developed over the years, Cati methodology for the countries with high telephone penetration, and Face to face methodology for the countries with low telephone penetration. In most cases the latter are restricted to the capital city.

Furthermore, the handling of the data and the analysis of the data are standardized as well.

CATI surveys

The technical management of most of the surveys in the industrialized countries has been carried out by Interview, a Dutch surveying company. Interview subcontracted fieldwork to survey companies in the participating countries, while maintaining responsibility for the questionnaire, sample selection and inteview procedures. The survey on Malta was done according to the Face to Face method, supervised by UNICRI.

Sampling: a sample of between 1000 and 2000 households was drawn by random dialing of telephone numbers. Non relavant contacts (like companies) were ignored. Within a household, there was a random selection of a household member aged over 16. In case of a refusal, this household member was not replaced. The process continues until the agreed amount of completed interviews were reached. An exeption to this procedure is Finland, a random selection of individual were drawn from the population register. Also an exeption was Northern Ireland and some rural parts of Spain (1989), since telephone penatration was low the interviews were taken face to face, but also computer assisted. The survey in Japan was also Face to face.

response rates: in the eleven industrialized countries in the 1996 sweep taken as a whole, 67% of the respondents selected for interview agreed to take part. this was an improvement on the overall response rate of 60% for the twelve countries of the 1992 sweep and on the 43% response rate in 1989. In 1996, response varied from 40% in the USA to 80% or more in Austria, Finland and Northern Ireland. For the seven countries which took part both in 1992 and 1996, the response rate was about the same or better in five, but fell slightly in two (the Netherlands and USA). For the three countries which had surveys in 1996 and 1989, responses were lower in Switzerland but higher in the other two.

CATI: the interviews were done by telephone. The interviewer reads the questions (and instructions) from a computer screen. The answers are directly entered into the computer system and used to select the next question. (For instance, the items on car crimes were skipped if the household has no cars.)

Face to face surveys

UNICRI was responsible for the face to face questionnaire and monitoring of the ICVS in the developing countries and countries in transition. The face to Face methodology was adopted in all developing countries and the countries in central and East Europe, with the only exception of Solvenia were CATI was used instead. Fieldwork included the undertaking of feasibility/training missions and the carrying out of pilot studies in the countries which were participating for the first time, and of course the fully fledged surveys. The whole procedure was based on the ‘manual for the Face to Face surveys of the International Crime victim Surveys’ (Alvazzi del Frate, 1995). Deviations of the described procedures were only made in close deliberation with UNICRI.

Fieldwork in nine capital cities in Eastern Central Europe was done by Gallup, an international surveying company. Some of the surveys were done Face to Face, others by telephone. Samples sizes were 1500 cases.

One of the objectives of the missions was to get acquainted with the country’s criminological situation, law enforcement and criminal justice needs in the area of crime prevention. this leads to specific needs as regard to, for example, additional questions to be included in the questionnaire and the development of sampling design. However, due to the comparative nature of the project, the proposed changes were kept within the main structure and content of the standard questionnaire and sampling.

translation of the questionnaire to the local language(s) was the responsibility of the national co-ordinator. Translation was done by people who have expert knowledge of criminology and survey methodology and have expert knowledge in their own languages and in English, French or Spanish -of which solid versions exist since 1989-. This were team efforts and translations back and forth ensure reliable translations. In some of the countries however the interviewers, having to work in several local dialects, were provided with the translation in the language of the majority linguistic group while translations into dialects were provided on the spot, that is to say, during the interviewing process by the interviewers themselves.

fieldwork In most countries the survey was carried out by and ad hoc team of interviewers, the national co-ordinators often relied on senior students. In one country, they were recruited among members of the local community upon recommendation of the community leaders. Sometimes it was important to match interviewers and respondents in terms of ethnicity, religious or tribal affiliation or other such characteristics. some other traits need to be taken into account. For example, in some cultures male respondents were not willing to talk with female interviewers and vice versa, customs preclude the use of male interviewers to interview female respondents particularly for such sensitive issues as sexual victimisation. In some countries, data collection was subcontracted to survey companies.

Samples of 1.000 respondents were generally drawn from the population of the largest city, although in a few countries the surveys covered either several cities. In some of the surveys there was and additional sample of 200 cases form a rural area, or a national sample. Sampling was generally hierarchical, it started with identifying administrative area’s within the city, followed by a step-by-step procedure aiming at identifying areas, streets, blocks, households and finally the household member aged over 16 whose birthday is next.

A systematic collection of response rates and refusals was only initiated with the 1997 version of the face to face questionnaire. As regards previous surveys, this information was provided by the national co-ordinators in their final reports. In 1996, the response rates in developing countries was on average 95%, ranging from a minimum of 86% in Botswana to a maximum of 99% in South Africa, The Philippines and Bolivia. Rate of refusal was on average 3.4%. For the countries in Central and East Europe, the average response rate was 81.3 %, lowest were Lithuania (54%) and Slovak Republic(56%), all others were above 73%. It was observed that in some countries the refusal rate was higher due to the vicinity of recent war conflicts which may have increased the general level of suspicion. In these countries the fear of strangers was so widespread that the national co-ordinator suggested including a series of questions dealing with attitudes towards opening the door to strangers and the use of entryphones.

The interviews were done face to face, the questions were read from paper and the answers marked on the questionnaires. Clear routings were indicated to avoid asking questions that were not relevant.

Weighting and data management

weighting

[update to 2000] A number weighting variables are computed to compensate for over- and undersampling, deliberate or otherwise, of particular groups within the population. In most of the surveys we had a sample of households, this means that from the individual perspective, members from large households are undersampled and therefore individuals living in small households are oversampled. Both household and individual weights are computed. The individual weights are basically proportional to the household size. In case of a random sample of individuals, the houshold weight is reversed proportional to the household size. Furthermore the weights compensate for age and gender distribution and, for nationwide surveys, regions within the country.

data management

[update to 2000] The surveys in the industrialized countries were organized by Interview B.V. After the surveys were completed they merged the seperate datafiles into one well documented datafile. Specific data entry software was developed for the Face to Face surveys. In case countryspecific items were added to the questionnaire, the software was adjusted. These datafiles were merged into one larger file. In turn, the five main files (1989, 1992, 1996 CATI and the 1992, 1996 Face to Face databases) were matched and merged into one master file. Some countries used their own software for entering the data, these files were matched and added seperately. At this moment the master file consists of 135.465 cases from 92 differend surveys in 56 countries.

Data analysis

The project runs since 1989. A standard set of data analysis tools have been developed over the years to ensure that choice in data analysis that have been made in the past are applied over time and also between countries. Also when different teams of data analysts and authors work on the data. This guarantees that the statistics that are made public the by the International Working Group are consistent and standardized. As always, the authors carry responsibility for the accuracy of whatever is published.