[1
AND JUSTICE A SERIES FROMTHE DUTCH RESEARCH AND DOCUMENTATION CENTRE
DELINOUENT BEHAVIOR AMONG YOUNG PEOPLE IN THE WESTERN WORLD
FIRST RESULTS OF THE
INTERNATIONAL SELF-REPORT DELINOUENCY STUDY
ft%9L.i /0/
RDC - MINISTRY OF JUSTICE
KUGLER PUBUCATIONS - AMSTERDAM 1 NEW YORK EDITED BY JOSINE JUNGER-TAS,
GERT-JAN TERLOUW AND MALCOLM W. KLEIN
Delinquent
Delinquent behavior among young people in the western world : first resuits of the international self-report delinquency sutudy / ed. by Josine Junger-Tas, Gert-Jan Terlouw and Malcolm W. Klein. —Amsterdam [etc.] : Kugler Publications.—Iii. —(Studies on crimeandjustice)
Met lit. opg.
ISBN 90-6299-108-4 Trefw.: jeugdcriminaliteit.
Distributors:
For the U.S.A. and Canada:
Kugler Publications P.O. Box 1498
New York, NY 10009-9998 Telefax (+212) 477-0181 For all other countries:
Kugler Publications P.O. Box 11188
1001 GD Amsterdam, The Netherlands Telefax (+31.20) 638-0524
©Copyright 1994 Kugler Publications
All rights reserved. Nopartof this book may be translated or reproduced in any form by print, photoprint, microfilm, or any other means without the prior written permission of the publisher.
Preface vii The international self-report delinquency study: some methodologi
cal and theoretical issues
Josine Junger-Tas 1
Self-reported delinquency in Helsinki, Finland, 1992
Kauko Aromaa 14
Self-reported offending among young people in England and Wales
Benjamin Bowling, John Graham and Alec Ross 42 The self-reported delinquency study in Belfast, Northern Ireland
Joan McQuoid 65
Self-reported delinquency in The Netherlands
Gert-Jan Tertouw and Gerben J.N. Bruinsma 102
Self-reported delinquency in Liège, Belgium
Michel Bom and Claire Gavray 131
Self-reported juvenile delinquency in Mannheim, Germany
Peter Sutterer and Thomas Karger 156
Self-reported juvenile delinquency in Switzerland
Martin Kittias, Patrice Vittettaz and Juan Rabasa 186 The self-report delinquency study in Portugal
Etiana Gersâo and Manuet Lisboa 212
Self-reported juvenile delinquency in Spain
Rosemary Barberet, Cristina Rechea-Atberota and Juan
Montailés-Rodriguez 23$
Self-reported delinquency in three Italian cities
Uberto Gatti, Giovanni Fossa, Eteonora Lusetti, Maria Ida
Marugo, GaetanaRusso and Giovanni Battista Traverso 267 Key-findings of a preliminary self-report delinquency study in
Athens, Greece
Cattiope D. Spinettis, Eteni Apospori, Maria Kranidioti,
Yiota Symiyianni and Nina Angelopoutou 288
Self-reported delinquency in a midwestern American city
Ineke Haen Marskatt and Vincent J. Webb 319
Self-reported delinquency in Germany after the reunification
Klaus Boers, Michael Ctass and Peter Kurz 343
Self-reported delinquency at age 18: New Zealand’s Dunedin multi disciplinary health and development study
Te mmie E. Moffitt, Phil A. Sitva, Donald 1?. Lynam and Bill Henry 354 Delinquency in thirteen western countries: some preliminary
conciusions
Josine Junger-Tas 370
Epilogue
Malcolm W. Klein 381
Questionnaire for the ISRD 387
In 1988 some 30 experts in self-report methodology from 15 countries met in Noordwijkerhout, The Netherlands, for a NA Advanced Research Workshop. The workshop was organized by Professor Malcolm W Klein from the University of Southern California and by the Dutch Research and Documentation Center, with the objectives of discussing problems in meas uring self-reported crime and delinquency and examining the potential for conducting comparative research. The resuits of the workshop were laid down in a publication on self-report methodology, ‘Cross-National Research in Self-Reported Crime and Delinquency’ (Klein, 1989).
During the workshop some of the experts present felt that the moment had come to undertake a comparative self-report study which would in clude countries that had achieved a certain level of methodological knowi edge and research experience. Others, however, considered that there were too many methodological problems and pitfalis to start such an undertaking: one should concentrate first on solving these problems be fore attempting to realize a complex international comparative study.
But prudence and caution were overruled by those who wanted to go forward. They argued that it would be easier to solve difficult problems by trying them out in international practice. In application, the methodology of cross-national research would be improved, and scientific knowledge would be advanced.
The latter group, having won the argument, the International Self-Re port Delinquency Study was started in 1990 with the construction of a common-core measuring instrument. The Research and Documentation Center of the Dutch Ministry of Justice took the initiative for the study, coordinated the information and organized the work to be done.
Constructing a common questionnaire that would satisfy the partici pating countries was no simple matter. Nor was it easy to agree on the ba sic methodological requirements necessary to achieve comparability of re search outcomes. These issues called for a lot of consultation, including two meetings in The Hague. Moreover, financial constraints seriously re stricted the nature of the samples and sample size.
This volume presents the first resuits of the self-report study in 13 in dividual countries. A second volume will follow in 1995, which will present the results of in-depth comparative analysis.
Specialists in the field will no doubt discover many limitations and shortcomings in the research material.
However, la perfeetion n’est pas de ee monde. We learned a lot by conducting this study and have already found a number of elements that may be improved in a next survey.
More than sixteen authors collaborated in order to realize the ISRD study. Without their enthousiasm and perseverance, their willingness to accept a common format and the editors’ comments, it would not have been possible to achieve this undertaking. We are especially thankful to Malcolm Klein for his continued association with our efforts, his very
helpful comments and the hard work he put into correcting our (Amen can) English.
We express the hope that the reader will be as fascinated by the self-re port delinquency outcomes in the different countnies as we, the research ers, have been.
Josine Junger-Tas
DELINQUENCY STUDY: SOME METHODOLOGICAL AND
THEORETICAL ISSUES
Josine Junger-Tas
1. Introduction
There has always been some interest in comparative and cross-national research, but it has clearly increased in recent years. There are several reasons for this. One is that international contacts between policy makers as well as between scientists have become increasingly frequent. Another is that the need for collaboration in both these fields is ever more urgent.
It is very important that policy makers gather information on the ex- tent and nature of crime in comparable western countries and investigate how crime and delinquency relate to specific socioeconomic and cultural contexts. Delinquent behavior of young people is a problem in the entire western world: teenagers commit most of the crimes such as all kinds of theft, burgiary, vandalism and violence. New insights and knowledge may eventually have an impact on the development of criminal policies in van ous countries.
This need for policy makers is reflected in the United Nations Crime Prevention and Criminal Justice Program and in the Council of Europe’s efforts in bringing together scientists and policy makers of some 27 Euro pean countries with the objective of providing information on and improv ing and evaluating criminal policy.
However, a persistent and nagging problem in cross-national research is the lack of adequate instruments to enable reliable and valid com parisons. Major efforts have been made by the United Nations to achieve comparisons between nations on the basis of police and cniminal justice statistics (Vetere and Newman, 1977; Pease and Hukkila, 1990). Inter preting the results presents major problems, however, because nations differ widely in the way they organize their police and court systems, the way they define legal categories and the way they collect and present their statistics. In fact, the lack of uniform definitions of criminal acts, of common measuring instruments and of a common methodology makes comparisons among countries extremely hazardous. It is in this area that the needs of policy makers and the interests of criminologists coincide.
Address for correspondence: Miaistry of Justice, Research and Documentation Cen ter; EO. Box 20301, 2500 EH The Hague, The Netherlands
Criminologists have been interested for a long time in differences in crime between countries (Clinard, 1978; Friday, 1973; LePoole, 1977;
Levinson, 1977; Vetere and Newman, 1977). At the same time, criminolo gists have been trying, not without success, to develop measures of crime to complement the existing official statistics which reflect police activities rather than the ‘true’ level of crime. Victim surveys are widely used as a corrective and sensitive measure of crime trends, and two large-scale international comparative victimization studies have yielded extremely interesting results (Van Dijk, et al., 1990; Van Dijk and Mayhew, 1992).
In addition to victim surveys, another way to measure crime is through the use of self-reports. Both methods have their strengths and their limi tations. Self-reports, however, have certain dear advantages over victim suneys. They make it possible to measure non-personal and victimless crimes as well as characteristics of offenders. Moreover, they can estab lish the prevalence and incidence of criminal acts committed by offenders and the relationship between types of crimes and types of offenders.
The ISRD study has three main objectives:
— To achieve comparability of prevalence and frequency of different types of delinquent behavior in the participating countries.
— To contribute to the explanation of (differences in) delinquent behavior, that is to try to advance theory building by examining whether gener ally accepted theoretical notions can be applied to all participating countries, and, if not, what differential factors might account for it.
— To contribute to the solution of a multitude of methodological problems related to cross-national research as well as to self-report methodology.
Two meetings were held at the Dutch Research and Documentation Center to coordinate the study. The first one, in the spring of 1990, was spent in finalizing a common core measurement instrument. The second one, in the winter of 1990, discussed the resuits of pilot studies and made a number of changes in the survey instrument.
In this chapter T will review some of the methodological and theoretical problems that the participants were confronted with and the ways in which they tried to deal with them.
Stating this, it should be made dear that we are modest in our ambi tions. For many of the problems and dilemmas there were no perfect solu tions. The gap between the ideal design and reality forced some research ers to accept less than ideal solutions. Despite these caveats, developing this line of research remains a real challenge to all of us.
2. Some methodological problems: how are they managed?
2.1 Sampling method
One important question concerns the scope of the sample: national sam ples or city samples. Obviously if the objective is to compare national de linquency rates, nationally representative samples are needed. However, if we are interested in trying to explain some of the crime differences be
tween countries, then we could do with smaller local samples. Of course, we would have liked to do both, but, although much attention was given to this matter, the harsh reality of funding limitations dictated the options avallable: only four participating study groups—in Switzerland, Portugal, the United Kingdom and The Netherlands —had sufficient funds to allow them to draw a national random sample; the others opted for city samples
— Liège, Belfast, Athens, Dunedin, Helsinki, Omaha, and three cities in Italy (Genoa, Messina and Siena), while Spain used a stratified national sample of cities. This means that comparisons of national delinquency rates had to be limited to those countries that drew national random sam ples, while the study of correlates of crime could be extended to all sam ples. However, in view of the many socioeconomic and cultural differences between countries that we cannot possibly control, we had to avoid rigid classifications. Wisdom suggests that the best we could do was compare the relative rank ordering of the prevalence of specific types of crimes in different countries and cities of comparable urbanization level.
This is not so much an immediate problem since this book presents a first overview of the findings, without comparative analysis. The next vol ume will be devoted to cross-national comparisons. At that point we must consider whether we can control for individual study differences or whether we should limit comparisons to those participants who used simi lar sampling methods.
Apart from sample scope, different sampling methods were used, each with its own problems and pitfalls. One of the most frequently used sam pling methods was to select from some school population. Access was rela tively easy and questionnaires were administered to groups of students.
The shortcomings of this method are well known. Marginal students such as truants and drop-outs who are likely to be different from the average student on a number of variables including delinquency are excluded from the sample. Moreover, in some countries, the age limit for compulsory education is 16 years, which considerably restricts the age range of the sample. To this should be added that, as a consequence of extensive pri vacy regulations in a number of countries, access to schools is becoming more and more difficult and in some countries even impossible.
Several participants — Italy (three cities), Belgium (Liège), Nebraska (Omaha) and Finland (Helsinki) — had recourse to school samples, al though their procedures differed. The Belgian researcher made a random selection of schools, where he approached students of 14 to 18 years of age (the limit of compulsory education is 18 years). The sample of those aged 18 to 21 was based on a random selection of population files. In the three Italian cities the schools were randomly selected according to type (pre paratory schools, technical schools, etc.) and, within the schools, the sub jects were also selected at random. The age range was 14 to 19 years.
The Finnish school sample was restricted to all 9th graders (15-16 years) and llth graders (17-18 years). However, since compulsory educa tion ends at age 16, those who did not continue their education were ex cluded.
The Omaha school sample inciuded a random selection of public school students in 12 public high schools, aged 14 to 18.
Despite the efforts of these participants to achieve random selection of schools and subjects within the schools, there is no exact way of knowing to which extent truancy and/or drop-out may have introduced a bias in the findings of these studies, leading to a possible underestimation of delin quency in the target populations.
An obvious answer to this problem would be random sampling. How ever, random sampling does not ensure that one will achieve a rep resentative sample. Some marginalized groups of juveniles, such as school drop-outs, unemployed youths, ethnic minority juveniles and institutiona lized youths, are usually underrepresented. These youths are generally hard to reach, and even when one succeeds in getting in touch with them they often do not want to participate in such a study. Since one may an ticipate that delinquency levels among these groups are higher than among the average respondents, this sampling method may also lead to an underestimation of the extent of delinquency in the youth population.
One solution in such a case is to combine a random sample with a stratified sample. This procedure was used by Hindelang and his col leagues in the Seattie study (1981) in order to estimate delinquency in a general school population as well as to maximize variance on delinquency.
Another possibility is to draw a random sample and add specific groups of young people, such as students of vocational training schools, regular visi tors of special youth clubs (which was done in the Omaha pilot of the ISRD project), or ethnic minority children living in specific neighborhoods (which was done in the Dutch and English studies). Still another way to include ‘high risk’ juveniles, known to have higher delinquency rates than the average youths, is to include institutionalized juveniles. This was done in the Los Angeles pilot, where 155 detainees in Central Juvenile Hall were interviewed (Junger-Tas et al., 1992).
The following solutions were tried by some participants: Spain used a stratified sample, based on city size and randomly selected census secti ons. The sample is not representative for the country as a whole because rural areas were excluded. Moreover, as the samples were not propor tional, weighting was needed. But sample size was large (2,100) and, in order to add a number of high-risk juveniles, an additional 250 young peo ple in marginalized neighborhoods were interviewed. A comparable proce dure was followed in The Netherlands, where a stratified nation-wide ran dom sample of 17 cities was drawn. The cities were selected on the basis of urbanization degree and geographical distribution. The sample size of 914 includes an oversampling of 68 Turkish and Moroccan youths in the two largest Dutch cities. Switzerland, Portugal, the United Kingdom and Northern Ireland also drew random samples. Moreover, in the English study, two samples were added: a random sample in high crime areas and a booster sample of ethnic minorities. The only exception is New Zealand where the research is part of a longitudinal study. The sample is a birth
cohort and inciudes all young people aged 18 in one particular city (Dunedin).
2.2 Response
A number of factors may have an impact on the willingness of young peo ple to participate in a self-report delinquency study. One such factor is prior contacts with the juvenile justice system. Research among juveniles (Junger-Tas et al., 1983) has shown that such contacts have a serious im pact on response rates: the highest response was among those who Te ported no prior police contacts; the lowest response was among those who had police contacts as well as prosecutor contacts. Similar resuits have been found for adult males, where response was also related to being known by the police (Veendrick, 1976). Unfortunately, there is not much that can be done to remedy this situation if one draws a random sample.
The best option in this case is to oversample certain high-risk categories, which has indeed been done by a number of participants in the ISRD study.
Another problem may arise from differences in response rates between participating countries. The difficulty lies in explaining these differences.
Of course, they may be related to shortcomings in the studies themselves.
But they may also be related to unknown cultural differences among coun tries. Some countries may be ‘oversollicited’ by social or marketing re search, while others are perhaps not yet used to this type of research. In both cases the situation can lead to low response rates. It is important, therefore, to collect information about non-response so that we can get a better idea about its characteristics. Five questions were added in the ISRD instrument which tap the state of the home, the maintenance of the building and of the other buildings in the street, the general appearance of the neighborhood and the presence of signs of vandalism. These ques tions, which are completed by the interviewers, were added in order to have some additional, and somewhat standardized, socioeconomic infor mation. The information is collected for all potential respondents. As a re sult, non-response can be analyzed not only with respect to variables such as age and sex but also with respect to socio-economic status (SES). How ever, response also depends on the sampling method. In school samples, the response will generally be high, although Finland reports some cases of low response where only a few youngsters showed up in the special classes organized for the survey, due to the unusually fine weather for the season. This resulted in a total response rate of 75%, whereas this was 98% in Belgium. In the case of random samples non-response is generally higher. As far as the Dutch study is concerned, taking into account non existent addresses and persons who had moved, the overall response rate was 61%. Response was clearly higher in the middie-sized (57%) and small cities (66%) than in the four large cities — inciuding Amsterdam — (52%). The response rate among ethnic minorities was fairly similar to that of the Dutch sample, namely 63%. The Dutch study checked the re presentativeness of the sample on a number of variables, comparing it to
the same-age youth population. With respect to sex, age distribution, school involvement, education and employrnent status, there were only slight differences.
An unexpected factor was the exceptionally good summer, which start ed even before the month of June. Many young people did not keep their— even repeated — appointments and could not 5e traced. In Spain the re sponse rate was 64% and among the high-risk youth even higher, at 75%.
But in countries such as Portugal (95%) and Northern Ireland (92%), re sponse was much higher. Where data were reported on non-response, we noted contradictary findings with respect to SES of respondents and non respondents. In The Netherlands non-respondents lived in relatively more deteriorated living environments than respondents (in terms of upkeep of the home and street vandalism). In Spain, however, it was the other way around: there the respondents were living in a Iess attractive environ ment. As the other participants did not report on this matter, we cannot draw any definite conciusions at this stage of the study.
2.3 Data cottection
There was consensus about the preferred mode of data collection. Consider ing the kind of data to 5e collected, the best method would 5e face-to-face interviews. However, this was not one of the options in all studies. Where it was impossible, as in the case of school samples, according to the Finnish researcher, some of the students had difficulties with the rather complex nature and the length of the questionnaire. In Italy all data were collected by face-to-face interviews during school hours and in rooms that were made available for this purpose.
An important point with respect to both reliability and validity was the preparation of the interviewers. In the Dutch main study, interviewers were given a one-day training session organized by one of the researchers and the research agency, and, in addition, received elaborate written in structions. In the Dutch city sample research, undertaken by the Univer sity of Twente, a separate test-retest reliability study was conducted on a sub-sample of 65 youths.
The Dutch pilot study indicated that the presence of a third party dur ing the interview — frequently the mother — introduced some bias in the answers given: comparison of those who were alone with the interviewer and those accompanied by another person indicated a lower admission rate of delinquent acts in the second group. On the basis of these results particular attention was paid to this point during the interviewer train ing. As a consequence, in the main study the proportion of third parties present during the interview was reduced by almost one-tenth, the pro portion of those intervening was reduced by half, and the difference in ad mitted delinquency between the two groups was considerably less pro nounced in the main study than in the pilot (Terlouw and Junger-Tas, 1992). Of course in this respect a situation such as in the Italian and the Omaha study, where the interviewer and the respondent are alone in a room, is preferable by far.
Except in the case of school samples where the questionnaire was self administered, data collection was generally based on face-to-face inter views, sometimes combined with a self-administered section of the ques tionnaire. Most of the interviews took place in the home of the inter viewees. Unfortunately, in a number of cases the respondent was not alone in the room with the interviewer. The Netherlands and Northern Ireland report that in 10% to 8% of the cases a third person influenced the interviewees’ answers. This was one of the reasons that delinquency ques tions in the present study have been self-administered. For example, in the English study the sociodemographic questions were asked face-to-face, whule the questions on drugs and on the other offences were asked via self-completion forms in two separate booklets. In Germany a similar method was followed.
It is very fortunate that the response rates are generally satisfactory and not too disparate. Mthough we are not sure as to what extent the re suits might be biased by non-response, the fact that in some studies addi tional samples of high-risk juveniles have been drawn while in others the sample is comparable with population data, would suggest that we can have some confidence in the generalizability of the findings.
2.4 The survey instrument
The instrument that is now used by participants is the result of a long proc ess of negotiation among representatives of the different countries. It was no easy task to achieve consensus on the delinquency items that were to be inciuded in the core list, that is those items on which we wanted to have comparative data. Some compromises had to be made, essentially between U.S. participants and the European ones, because of the fact that a number of acts which are offenses in the U.S are not so in Europe (alcohol use and purchase) or are not prosecuted (cannabis use).
Globally the questions covered five groups of variables: prevalence and frequency of delinquent behavior; circumstances of the act; social reac tions to delinquency; social background variables; and some theoretical variables.
Prevatence and frequency are measured in different ways. All delin quency questions are introduced by ‘did you ever....’ (lifetime prevalence), followed by ‘did you do it tast year?’ (current prevalence). The next ques tion is then ‘how many times did you do it’ (frequency)*.
Circumstances of the act are measured by questions such as: did respondent act alone or with others, who was the victim, where did it hap- pen.
Sociat reactions implied discovery of the act and reactions to it by par ents, others or the police.
Sociat and demographic uariabtes include sex, age, family composition,
*See for list of offenses, Questionnaire,p.389
ethnicity, SES, education, school attendance or employment, leisure acti vities.
A limited number of theoreticat variables, related to social control theory, include relations with parents, parental supervision, school in volvement and performance, attitudes towards school and/or work and leisure activities.
Of course we are talking here of the common core instrument. Partici pants could, and did, add variables that were important in their own local situation. In this book, however, only the findings that relate to the core instrument will 5e reported.
There was much discussion about the level of detail that would stil! al low sufficient and reliable information. Several participants performed Te liability tests, repeating earlier questions at the end of the interview (Junger-Tas et al., 1992). In this pilot study, internal consistency of the answers to six questions ranged from about 90% for five of them to 75%
for shoplifting, a high frequency offense. The Spanish study similarly re peated a number of questions at the end of the questionnaire and com pared the answers with the earlier answers given. They found high levels of reliability.
A more serieus problem was the instrument’s validity. For comparative purposes we agreed on a common core operationalization of delinquency, including acts that are considered offenses if committed by an adult, in all participating countries. Status offenses are defined as ‘problem’ or deviant behavior but are not part of the delinquency definition as used in interna tional comparisons. Moreover, we used a set of ifitering questions, such as
‘have you ever...’ before going into details about the exact meaning and cir cumstances of the act. In order to optimize response validity, a number of studies used a mixed approach. The first and second part of the question naire, including all sociodemographic questions, was administered in face- to-face interviews. The third part of the questionnaire, that is, the selec tion of follow-up questions that refer to the admitted acts, was subse quently given to the respondent by the interviewer to complete.
Great care was also taken to have high item-specificity: for example, what exactly is meant by ‘threatening other people’? Did the boy steal the car er did he merely use it for joy-riding? Further, the reference period for the detailed questions was one year only, since longer periods increase the risk of memory bias.
2.5 Concurrent vatidity
Another nagging methodological problem is how to achieve some measure of concurrent validity. What other sources on delinquent behavior are avail able to control our data? In many studies police reports are used to control whether the offenses figuring in the police reports were mentioned by the respondents. Such studies have been done both for aduits (Veendrick, 1976) and for juveniles (Hindelang et al., 1981; Junger-Tas et al., 1983; Junger, 1990). The self-report method did not seem adequate in the case of adults:
there was considerable underreporting of burgiary, assaults and hit-and-
run offenses. As for juveniles the resuits suggest moderate to high validity.
However, there are indications that juveniles are far more inclined to talk with relative openness about their delinquent behavior than they are about their contacts with the juvenile justice system or the police. Unfortunately most participants in the ISRD study do not have the names and addresses needed for this type of control, while others —on the basis of legal regula tions —are not allowed to check validity this way.
One way to approach this problem is to check the data with data from other similar, or at least comparable, studies. This was tried by the Bel gian and Spanish researchers, but they were not able to achieve the corn parison. The questionnaires used were too dissimilar in several respects:
the reference period for committing delinquent acts, the euphemistic wording in the definition of the acts (pinching or borrowing instead of stealing) which almost certainly undermined validity, and differences in the delinquency items. Moreover, the geographical areas covered and the sample ages were different.
Partial validity controls were introduced in the Dutch study by compar ing delinquency data with resuits from two comparable studies: first, the bi-annual self-report study of a national random youth sample of 12- to 18-year-old minors, and second, a longitudinal self-report study of a 12- to 25-year-old sample. In the national youth study many questionnaire items are comparable to those of the ISRD study, while in the second study the ISRD questionnaire was used. Both data sets are also used to test exter nal validity. The comparison showed considerable similarity in the re suits. A sirnilar approach was used in the Omaha study, where the data were compared with two large self-report studies in the U.S — the Seattle study (Hindelang et al., 1981) and the National Youth Survey (Elliott et al., 1983). Considering the fact that the Ornaha sample is a school-based one, there was a surprising similarity in the relative rank-ordering of the offenses.
The Finnish study used different techniques to check validity, such as a comparison with a qualitative study on self-reported criminal behavior among a similar school population and a parallel study in another Finnish city. An interesting partial validation resulted from the consultation of the so-called ‘penalty books’ of the sampled schools, wherein disturbances and (school) problem behaviors are recorded. Although no individual rates can be compared, this would allow for comparison of troublesome behavior with delinquent behavior at the classroom level.
There is one validity problem left that is hard to solve: the validity of the self-report method administered to ethnic minorities. One earlier Dutch study found large variations in validity arnong three different eth nic groups (Junger, 1990). One way to check validity is to compare the self-report data with police data. This was done in Junger’s study.
Three ISRD participants (Switzerland, England and Wales, and The Netherlands) found that ethnic minorities reported fewer delinquent acts than their national counterparts. Unfortunately, there were no additional (police) data to operate a validity check so that the discrepancy between self-report data and official data on ethnic minorities remains essentially unexplained. 1f we also take into account the differential response validity
between blacks and whites in American self-report studies (Hindelang et al., 1981; Elliott and Ageton, 1980), we might venture the following tenta tive statements about this problem. The bulk of the research in this field tends to suggest that where specific ethnic groups are well-integrated in mainstream society — speaking the language and participating in the country’s economic and social life — response validity of these groups will not differ to any great extent from that of the national youth population.
Where new immigrant groups participate in self-report studies, validity may be low, if only for cultural reasons. In the case of long-standing, very deprived minority groups, validity will at least be questionable. In prac tice this would mean that any research on ethnic minorities using self-re port would require careful study of the social, economic and cultural posi tion of the groups in question before conciusions can be drawn on response validity. Of course it would be better stili to check response validity with another source of delinquent or problem behavior.
3. Some further methodological issues
Once the data collection of the individual studies was completed, there ap peared to be several requirements for a useful analysis, recalling that the focus of the analysis is the country/city rather than the individual respon dent. A first requirement was to reach agreement on at least the following dependent variables:
— the prevalence and frequency of delinquency either at the national level or at the city level (depending on the sample available);
— the nature of delinquency, for example the distribution of specific offenses or the proportion of property-related and violent offenders;
— the seriousness of delinquency measured, for example, by frequency andlor by the nature of the delinquent acts;
— the prevalence and frequency of ‘problem’ behavior (status offenses);
— the nature of problem behavior, such as differences in the amount of truancy and alcohol and drug use.
Second, in order to allow for simple comparisons in this first volume, and with an eye to later, more complex, comparative analyses in the next one, certain decisions had to 5e taken with respect to the independent variables, some of which had to be constructed on the basis of the existing data set.
The main independent variables were the following:
socioeconomic variables:
(un)employment father (un)employment mother education level
source of income (part-time) employment unemployment respondent school drop-out
degree of urbanization
demographic variables—composition of the population:
ethnicity/race age
sex
single-parent family presence of stepparent
cutturat variables with respect to adotescents and young adutts:
alcohol use drug use
peer group involvement
other variabtes retated to criminologicat theory:
school performance school commitment work commitment bond with parents supervision by parents bond with friends
organized leisure/sports participation
Some variables, such as education status, SES and ethnicity, present seri ous problems because of different defmitions among countries.
With respect to educationat status, long discussions and several trials to combine classifications based on the many different school systems re sulted in the following definition: education status is defined as the level of education achieved, measured in completed school years. For those still in school this is the current level; for those who have left school this is the highest level achieved, whether completed with a diploma/certificate or not. Three categories were distinguished:
— high: university or comparable higher education; gymnasium!lycée or preparatory school leading to university;
— medium: professional, vocational or training schools leading to me dium-skill jobs;
— Ïow: vocational training, apprenticeships, no training after leaving el ementary school, leading to low-skill jobs;
The assumption is that this classification is somewhat independent of age, since a youngster in a medium-level school will probably get a me dium-level job. Thus, although he ages, his educational status will remain the same.
SES presents similar definitional problems. The final proposal was to combine father’s professional status and the interviewer’s estimate of the status of the home*. Father’s professional status is classified as:
— lower class (unskilled job; unemployed)
— lower middle class (skilled job)
— middie class (employees; employers of middie-sized businesses)
*T}s would be impossihie in the case of school samples.
— higher middie class (professions; jobs requiring higher education)
— upper class (employers in big enterprises)
Home status is evaluated by the interviewer at the end of the interview on the basis of home appearance, street appearance, evidence of vandalism andlor whether neighborhood is run down.
Finally, ethnicity is essentially based on the place of birth of the father andlor on race.
All this is fairly straightforward, and it helped to standardize to a cer tam extent the analysis of the individual studies. The resuits give us information on either entire countries or on cities. This allowed us to place one country/city next to another and to examine the question of whether relationships that are established in one of them are also valid in others. In order to make these comparisons worthwhile, we recommended that researchers conduct the analyses of the individual studies along simi lar lines, offering a common format for presentation of the results. This was indeed achieved and makes reading the different chapters much easier and more interesting.
In conclusion, however, a word of caution is required. In designing the ISRD study we tried to keep sources of variation between countries within certain ilmits. For example, we decided to restrict the study to countries in the western world so that there would not be too many differences in the economic, social and cultural situation of the participating countries and to ensure a more or less similar understanding and interpretation of most of the delinquency items. Also, all participants in the ISRD study used the common measurement instrument. (The Dunedin and EIW-Ger man studies being partial exceptions.)
However, as will already 5e evident, despite these similarities, there are considerable differences between studies in sampling method, sample scope and data collection. To just exactly what extent and in what ways these differences will iniluence the studies’ resuits, we do not know. But of course they will have an impact on the closer comparisons to be con ducted later on.
These future much more complex comparative analyses will be devoted to trying to explain variations in delinquency prevalence and frequency as well as variations in the nature of delinquency across participating coun tries. In addition we will examine the differential distribution of indepen dent variables over the various individual country/city samples. To the ex- tent that these variables are related to delinquency, their differential dis tribution over the various samples might offer some explanation for varia tions in delinquency.
It is dear, however, that for the purpose of these comparisons, deci sions must be made on questions such as which countries or cities may 5e compared among themselves, what elements we might control for and what differences we would be willing to pass over for the sake of compari son. It will be of paramount importance to report carefully the steps that will 5e taken in this exercise, so that the scientific community can judge what has been done so far and support attempts to improve the methodol ogy of international comparative research.
References
Clinard, M.B. (1978), Cities with Littte Crime: The Case of Suitzertand. Boston, Mass., Cambridge University Press.
Elliot, D.S. and D. Huizinga (1983), Social class and delinquent behavior in a na tional youth panel: 1976—1980. Crintinotogy, 21: 2.
Elliot, DS. and S.S. Ageton (1980), Reconciling race and class differences in self-re ported and official estimates of delinquency. American Sociologicat Review 45: 1.
Friday, P.C. (1973), Problems in comparative criminology: comments on the feasi bility and implications of research. International Journat of Criminotogy and Pe notogy, 1:2.
Hindelang, M.J. (1981), Variations in sex-race-age specific incidence rates of of fending. American Sociotogicat Review, 46.
Hindelang, M.J., T. Hirschi and J.G. Weis (1981), Measuring Detinquency. Beverly Huis, Calif., Sage Publication.
Junger, M. (1990), Detinquency and Ethnicity: An Inuestigation on Sociat Factors relating to Detinquency among Moroccan, Turkish, Surinamese and Dutch Boys.
Boston, Mass. and Deventer, The Netherlands, Kluwer Law and Taxation Publi cations.
Junger-Tas, J. et at.(1983), Jeugddetinquentie: Achtergronden en Justitiële Reactie.
‘s Gravenhage, Staatsuitgevenj.
Junger-Tas, J. and G.J. Terlouw (1991), Het Nederlandse Publiek en het Criminali teitsprobleem (I)(II). Detikt en Detinkuent, March and April Issues.
Junger-Tas, J., M.W. Klein and X. Zhang (1992), Problems and dilemmas in corn parative self-report delinquency research. British Journal of Criminotogy, forth coming.
Klein, M.W., ed. (1989), Cross-national Research in Self-reported Crime and Detin quency. Deventer, The Netherlands, IGuwer Academic Publications.
Levinson, D. (1977), What have we learned from cross-cultural surveys? American Behaviorat Scientist, 20:5.
LePoole, F. (1977), Law and practice concerning the counterparts of ‘persons in need of supervision’ in some European countries with a particular empliasis on the Netherlands. In: Lee E. Teitelbaum, ed., Beyond Control, Status Offenders in the Juuenite Court. Cambridge, Mass., Ballinger Publication Company.
Pease, K. and K. Hukkila (1990), Criminat Justice Systems in Europe and North America. Helsinki, Finland, Helsinki Institute for Crime Prevention and Control.
(affihiated with the United Nations.)
Terlouw, G.J. and J. Junger-Tas (1992), The International Juvenile Self-report De linquency Study: Design and first resuits of the Dutch study. Paper presented at the 44thAnnual meeting of the American Society of Criminology. New Orleans, U.S.A.
Van Dijk, J.J.M., P. Mayhew and M. Killias (1990), Experiences of Crime Across the World. Boston, Mass., and Deventer, The Netherlands, ICuwer Law and Taxa tion Publications.
Van Dijk, J.J.M. and P. Mayhew (1992), Criminat Victimization in the Industriat ized World: Key Findings of the 1989 and 1992 International Crime Surueys. The Hague,The Netherlands, Directorate for Crime Prevention, Ministry of Justice.
Veendrick, L. (1976), Verborgen en Geregistreerde Criminaliteit in Groningen.
Groningen, The Netherlands, Rijksuniversiteit Groningen, Crirninologisch Insti tuut.
Vetere, E. and Gr. Newman (1977), International crime statistics: Au overview from a comparative perspective. Abstracts on Criminology and Penology, 17:3.
HELSINKI, FINLAND, 1992
Kauko Aromaa
1. Introduction
The 1992 ISRD was carried out in two cities in Finland: Turku and Hel sinki. Preliminary resuits for Turku have been reported separately (Aromaa and Laitinen, 1993) and are not inciuded in the present report. In this report, the description of the research site, therefore, refers to the whole country and to the city of Helsinki as the Finnish state capital. Fin land is an independent republic in Scandinavia, with 5 million inhabitants.
The country is sparsely populated, with 16.4 (1990) inhabitants per square kilometer. Independent since 1917, it has remained rather isolated from both Eastern and Western Europe; thus, the proportion of foreign citizens in the total population is about one percent even today. The recently acceler ated integration development in Europe, in particular the changes in the former Sovjet Union, are presently rather strongly influencing the amount and structure of foreigners—permanent inhabitants as well as visitors—in the country. The proportion of foreigners is increasing, most of the present growth being caused by immigration from areas of the former Soviet Union (Estonia and Russia, primarily). The Southern parts of the country inciude the largest cities, the largest being the state capital Helsinki with 500,000 inhabitants (greater Helsinki — the capital and its environs — has over 800,000 inhabitants).
2. Socioeconomic and sociodemographic information 2.1 Socioeconomic characteristics
Urbanization Finland is still not very urbanized as compared with the in dustrialized countries of Central Europe. About 60% of the population live in cities or towns; however, the size of the cities is rather small. Only the capital has as many as 500,000 inhabitants. The next cities in size—five in all—have between 100,000 and 170,000 inhabitants. About 40% (40.7% in 1990) of the population live in cities with at least 100,000 inhabitants.
Distribution of poputation over industry About half of the population is counted as belonging to the employed labor force. Of these, nearly 60% are employed in services, 30% in industry and construction and less than 10%
in agriculture.
Address for correspondence: National Research Institute of Legal Policy, P.O. Box 35, 00931 Helsinki, Finland.
In Helsinki, 55% of the population belong to the employed labor force. Of these, 77% are employed in services, 19% in industry and construction.
The employment situation has rapidly deteriorated since 1991, the Un employment rate being presently above 20% for the whole country (at the time of the fleidwork of this survey, it was about 17%).
Level of prosperity The consumption of energy was (1989) at the level of 5,847 Mtoe* per capita, 1.8 times the OECD European average (partly, how ever, depending on the cold climate due to the northern location of the coun try). The telephone density (1990) was 535 per 1000 inhabitants, or about the Italian, British, German, French, Dutch and Norwegian level. The gross national product (1990) was about 500 billion marks, or 26,000 USD per capita. The social welfare expenses were (1989) about 24% of the gross national product, about the Norwegian level (26), below the Danish level (30) and clearly below the Swedish one (35). The income distribution in Fin land is rather skewed, the househoids in the lowest decile earning less than one-third of the lower income limit of the highest decile (1989).
Education level Of the population, 50.4% of those at least 15 years old (1990) have completed their education. This inciudes 40.7% of the popula tion having upper secondary education, and 9.7% with higher education. In the age bracket 25-29, 83.4% have completed their education, inciuding 12.4% with higher education. In the age bracket 30-34, 77.5% have corn pleted their education, and this inciudes 16.7% with higher education. The figures are generally slightly higher for women than for men.
2.2 Demographic information
Ethnic composition Ethnically, Finland is very homogeneous. Only about 1% of the population are foreign citizens, the largest group of these being citizens of the (former) Sovjet Union. Racially, the Fjnns are mixed. The main stock apparently derjve from the East Baltic and Nordic races. The Finnish language belongs to the Finno-Ugrian group. Finland has a Swed ish-speaking minority (6% in 1990). In 1990, 296,738 persons spoke Swed ish, 2,427 spoke German, 3,569 spoke English, 3,884 spoke Russian and 1,734 persons Lappish as thejr mother tongue. The Swedish-, Russian- and Lappish-speaking minorities have a long history, the nomad Lapps being the prehistoric inhabitants who were edged out by the immigrating Finns around 1000 B.C. Between the njnth and thirteenth centurjes, Swedes en tered the country and consolidated Christianity. Until 1809, Finland was a province of Sweden. In 1809, Finland became an autonomous Grand Duchy of Russja and, in 1917, an independent republic (Puikkinen, 1983; Statisti cal Yearbook of Finland, 1991). For Helsinki (1993), the proportion of the Swedish-speaking population was 7.4% and the proportion of foreigners
*Mtoe is a standard measure for energy consumption, and stands for million tonnes of oil equivalent; toe=equivalent of 1 tonne of heavy fuel oil (1 tonne= 11,28 MWh).
2.5%. Half of the population with foreign nationalities live in the province of Uusimaa, i.e., in the greater Helsinki region.
Age distribution Like many other western countries, Finland has been ag ing rapidly in recent years:
Age bracket 1970 1981 1991
% % %
-24 43.1 35.0 32.0
25-34 13.4 18.8 15.0
35-44 12.4 14.2 16.7
45-54 11.3 11.2 12.6
55-64 10.5 9.4 10.1
65- 9.3 11.4 13.6
Total % 100.0 100.0 100.0
For example, while the size of the total population has grown only slightly (8% between 1970 and 1990), the number of men in the age bracket 15-24 years has fallen from 444,000 (1970) to 330,000 (1990), os by 25%. The popu lation of Helsinki is older than the average population. The proportion of 0 to 14-year-olds is (1991) below the national average, 15.0 versus 19.3%.
Also, the proportion of those at least 65 years old was higher, 14.4 versus 13.3%.
Divorce rates Divorce rates nowadays are not quite as indicative of ‘bro ken homes’ as before the late 1960s when cohabitation began to normalize (cf. Aromaa et al., 1980). The annual number of divorces has, in any case, stabilized: about 10,000 divorces since the late-1970s, rising to the level of 12,000 in the late-1980s and early 1990s (with an all-time high of 14,000 in 1989, due to a change in the divorce legislation). The level of 12,000 corre sponds to about 2.4 divorces per 1,000 inhabitants. The number of mar riages, around 30,000 annually in the late 1970s, has fallen to 25,000 in 1991. Calculated per 1,000 marriages, the 1991 rate of divorces was 0.52.
This rate has grown rapidly, in the first half of the 1980s, it was still about 0.32-0.35. From 1986 to 1991, the rate has been 0.38, 0.39, 0.47, 0.58, 0.54 and 0.52, respectively.
For Helsinki, divorce figures level with the national average. Their an nual average number for the period 1986-1990 was 1,605 (3.3 per thousand of the average population), against 3,436 marriages (7.0 per thousand). Cal culated per 1,000 marriages, the rate of divorces follows closely the national level, the rates for 1990-1992 being 0.50, 0.52 and 0.54. The comparability of the Helsinki rates with the national rates suffers a little, however, from the fact that cohabitation is most common in the region of the capital.
Distribution of different househotd types The number of households (1989) was 2,149,200. Their average size was 2.3 persons. The proportion of one person households was 34.9%, of two-person househoids 29.6%. At least
four persons lived in 20.4% of all househoids. Househoids with children totaled 30.4% (654,400). Of the househoids with children, 7.3% were one adult-one-child-househoids, and 3.7% had one adult and two chuldren.
The size of the Helsinki househoids (1990) was below the national aver age: 2.1 persons. Of the Helsinki households 40% had children; 40% of the families with children had only one parent.
2.3 Cutturat factors
Alcohol legistation poticy Like its Nordic neighbors Sweden and Norway, Finland has a state alcohol monopoly, allowing sale of alcoholic beverages— inciuding medium-strength beer—to those at least 18 years old, except for strong alcoholic beverages where the age limit is 21. Beer may be sold in all kinds of shops and served in any service place, but other alcoholic beverages are sold only in outlets of the state alcohol monopoly. Wine and hard liquor may be served in places licensed and supervised by the state monopoly. The price level of alcoholic beverages is regulated by an alcohol tax, resulting in a relatively high price level as compared with Western European averages.
According to consumption surveys, alcohol use is, among the young gen eration, increasing. The overall consumption level is about 7 liters of abso lute alcohol per capita, and is decreasing presently as ajoint consequence of the economic recession and the aging of the population.
Drug tegislation poticy All narcotics are prohibited. The list of narcotics inciudes a number of medically used legal drugs. Also the use of narcotics is criminalized.
The law distinguishes between petty, ordinary and aggravated narcotics offenses and two degrees of smuggling of narcotics. Standard penalties for ordinary narcotics offenses are fines. The maximum penalty for aggravated drug offenses is 10 years imprisonment. The proportion of prisoners witli a drug offense as the ‘main offense’ is about 3%. The number of persons appre hended for narcotics offenses (in 1991) was about 1900. In 1991, 931 persons were sentenced to punishments with drug offenses as their main offense. Of these, 708 were sentenced to fines, 91 to conditional imprisonment and 132 to unconditional imprisonment.
Studies on drug use indicate that it has remained at a rather constant level during the 1980s, with, for example, 8.5% of conscripts (1987) admit ting to experimentation with narcotics and about 1% admitting to having used narcotjcs at least ten times. However, recent data for 1992 indicate that experimentation has again become more popular, approaching the 15%
level of the early 1970s. For the 15-year-olds, a recent survey for 1992 esti mates that about 10% have experimented with drugs, solvent sniffing or classic narcotics (Kontula and Koskela, 1992; Rikollisuustilanne, 1992).
Proportion of mothers in paid emptoyment For decades, Finland has had one of the highest percentages of mothers in paid employment. In 1990, 64%
of all adult (15-74) women belonged to the labor force (as compared with 73% of the men).
The participation of women in the labor force as well as their employ ment rate is likely to decline slowly now as, due to the economic recession, a serious problem of a permanently high unemployment rate is forcing the mothers home.
Presence of strong youth cutture Youth researchers claim that, in recent years, youth culture has tended to vanish and disaggregate into numerous individualistic and variable youth career lifestyles (Lahteenmaa and Siurala, 1991). Gang formation is weak, partly because of the economic stagnation and, later on, the recession, and the presently very slowly pro ceeding urbanization of the country, together with the phenomenon of per manently small birth cohorts ever since the early 1970s.
Participation of youth in organized teisure There are no standard statis tics on how widely the young cohorts are involved in organized leisure.
However, sports and other organizations claim rather large membership figures. According to survey data for the 1980s, the total number of youth organizations has declined as a consequence of the decline of the support of political youth organizations. Also the proportion of youths involved in such organizations has decreased during the 1980s. In 1988, 35% out of all youths between ages 10 and 24 years were members of sports organiza tions, 19% belonged to hobby organizations, 13% to labor unions. Because of overlapping memberships, however, it is likely that the majority are not a member of any kind of youth organization. Membership does not, however, mean active participation. In a national survey made in 1986, 79% of the respondents in the age bracket 15-19 years did not participate in the activi ties of sports organizations, and only 4% did so at least once a week (Hir vonen, 1989; EL086).
For Helsinki, similar information is available for the end of the 1980s.
According to this information, Helsinki juveniles participated in the activi ties of youth organizations slightly less than the national average.
Organization membership or participation in the activities of youth or ganizations does not fully correspond to the idea of participation in organ ized activities. Thus, for a large proportion of youths, leisure time is ifiled by other ‘organized’ activities, such as music classes and lessons, horseback riding and a multitude of other organized activities that are not covered by
‘youth organizations’. Therefore, the picture given above is not quite correct if the idea is to find a rough measure of the amount of unorganized leisure time among juveniles.
3. Study design 3.1 Sampting method
A non-random sample of 9th (15-16-year-olds) and llth (17-18-year-olds) classes was drawn from the register of schools in Helsinki. The aim was to sample secondary schools representing variable areas, in particular on two dimensions: social status and the characteristic of special versus general
school. The schools involved are, for the ninth-graders, formally standard schools, and, with few exceptions, the whole birth cohort participates in the training which is practically compulsory up to the age of 16. For the llth graders, they are either gymnasiums or vocational schools. Before lOth grade, the Finnish school system does not differentiate by future status;
only then is there a division, and lOth-graders and beyond are either gymnasists or vocational school pupils (or, of course, working or in on-site vocational training or unemployed and could not be reached by the class room method). In the region of the capital, about 90% of the age bracket of the llth-graders are estimated to be in one or the other type of school.
By this method, the 9th-grade class sample may be considered repre sentative of all 9th grades, whereas the llth-grade class sample is biased at least in excluding those not participating in further formal education.
The different school types are selective to a certain extent: boys more often than girls choose to continue their education in vocational schools.
Therefore, vocational school samples tend to be boy-dominated (in 1991, the proportion of women in vocational schools was 30.8%). Also, girls tend to concentrate in certain vocational training directions just as do boys. Our sample of vocational school classes is very biased in this respect: we did not gain access to a school with giri-dominated directions, and therefore 82% of our vocational school respondents are boys.
A considerably larger proportion of the girls continue, after the 9th grade, in gymnasiums. Gymnasium samples therefore have an overrepre sentation of girls (in 1990, 60% of gymnasium pupils were girls).
Generally, school is entered in the year in which the child reaches the age of seven. Therefore, the 9th-grade pupils are almost always 1546 years old, and the 2nd-year gymnasists are 17-18 years old, as are the 2nd-year voca tional school trainees. In this last group, there is, however, a proportion of students who are continuing their training after a break, and these may be in their early, or even late, twenties.
The total raw Helsinki sample, as calculated from the list of pupils be longing to the 80 classes from 17 different schools, was 1,672. These com prised 41 9th-grade classes from nine schools (with 826 pupils), 15 gymna sium classes (from 6 schools, with 473 pupils) and 24 vocational school classes (from 2 schools, with 387 pupils).
3.2 Response rate and characteristics of non-response
The overall response rate was 75%, varying between 38 and 92%. As a Te sult, the net sample consisted of 1,254 respondents. The low (38%) response rate was found in one of the vocational school subsamples that was made in a deviant manner; all second-year pupils of one of the two vocational schools participated in the study. They were gathered together in the central hall of the school on a beautiful day late in the spring season, and, consequently, only one-third of the pupils arrived. The second-lowest response rate of 57%
was found in a gymnasium where all second classes were invited to gather in one hall for the survey and were thus almost openly given the chance not to participate.
In one school, the second-grade pupils happened to have a party on the previous night. Three classes from this school participated in the study;
their average response rate was about 75, but lowest in the morning and highest in the afternoon. This affected the boys, in particular.
Normally, the questionnaire was completed during a regular school les son, where average participation is formally expected behavior although, as a rule, 10-20% of the pupils seem to be absent on any school day and any time during the school day in the 9th grades, and 15-30% in the second grades of the gymnasiums.
The response rate was lower for boys (71%) than for girls (80%).
The manual school registers made it very difficult to try to assess the non-response more accurately; however, in a classroom sample, it is obvious that those not present are likely to comprise some of the most crime- and problem-intensive pupils. Many of the missing pupils are, of course, absent for random or problem-irrelevant reasons, but, for some, their absence is an indication of truancy and other problem behavior. The teachers of each class were asked to give some general observations on this, and they usually agreed that this was the case. However, they usually did not think that the absent pupils deviated very much—only a littie bit—from those present.
3.3 Data-cottection method
The data were collected by a self-administered, anonymous questionnaire that each respondent completed individually in class, during a regular les son, and under teacher andior researcher supervision; usually, the teacher was not present when the questionnaires were completed. Despite at tempts to prevent extensive group work, some pupil interaction could not be excluded by this technique, reminding one of the warning by Sipil (1982) who says that group questionnaires provide a particular problem.
His conclusion was that the variable of school-class influenced the self-re port results more than other relevant factors, meaning that the group situ ation apparently inftuenced the answers to a considerable extent.
The questionnaire used in the Helsinki study was not fully identical with the one apparently used in other countries. The differences are minor, how ever. First, we added three of our own offense questions: one on drunk driv ing, one on riding a bicycle while intoxicated and one on glue/solvent sniff ing. Second, the items concerning alcohol use in our version of the question naire use the wording referring to ‘illegal’ use of alcohol, a formulation that seems to have caused some problems. Third, in the drug battery, we did not have the questions on selling drugs, hard or soft: our ‘final’ version did not have these items as it was based on the English-language version that was available to us at that time. 1 do not consider this a great problem since Finland continues to be the Western European country with the smallest drug market (also relatively speaking). The drug use questions managed to locate only a very small number of respondents with experience of narcotic substances, if non-medical use of drugs and solvent sniffing are excluded.
3.4 Retiability
The considerable length of the questionnaire means that there is an obvious risk of low-quality responses, particularly for those not well accustomed to completing lengthy questionnaires. We were able to detect some such in stances, but there is no accurate estimate of the degree of this problem.
Not all respondents take this kind of study seriously. Kontula and Merilainen’s (1988) remark on a typical problem of the technique applied in this case may also be repeated: we may always expect some ‘smart alecks’
in this type of design. Not surprisingly, we could identify a small number of obviously non-serious answers. Other means of controlling this problem were not available.
Another approach to assessing the reliability of the present data is to consider the parallel study made in Turku (Aromaa and Laitinen, 1993).
The Turku study was made in a largely identical way, and thus its results should be rather similar to the Helsinki results minus the likely difference caused by the city size. Helsinki being considerably larger than Turku (500,000 versus 170,000 population), we should also expect higher delin quency and problem behavior rates. A first observation in support of this expectation is that the 9th-grade pupils of Turku had an almost non-exist ent non-response: only a handful of the pupils in the sampled classes were absent. Also, the Turku respondents generally displayed a rate of delin quent and problem behavior — both prevalence and frequency — that was very close to the Helsinki results, with hardly any significant differences.
This indirectly supports the idea that the Turku respondents, on the whole, ought to display slightly less delinquency and problem behavior than their Helsinki peers.
3.5 Vatidity
So far, we have two separate validity checks: the studies of Honkatukia (1992, 1993) and Kivivuori (1992, 1993). These studies are qualitative analyses of non-standardized interviews about the self-admitted criminal behavior of youths who belonged to similar classes as those participating in the quantitative study. Some of the youths in Kivivuori’s and Honkatukia’s youth club samples also belonged to the questionnaire sample. Their re sults showed that children contacted outside of the school setting—in youth clubs that mostly attract youths who tend to skip classes or play truant to a massive degree —admit to many more delinquent acts, more serious delin quent acts and more systematic contacts with the underworld of adult criminals than those contacted in school surroundings. Qualitative data derived through essays (with the theme ‘as 1 committed a crime’) authored by all pupils in a number of school classes would seem to repeat the general patterns of delinquent behavior emerging from the questionnaires.
Other direct means of validation have not been attempted—for instance, the possibility of comparing the individual responses with police registers was not possible because of the guaranteed anonymity of the study. How ever, there is a partial possibility by comparing the police data with the
survey data as a bulk: compare the number of suspects in the age bracket 15-17 in the police records with the number of respondents (as a percentage of the corresponding age cohort, assuming that the respondents are a repre sentative sample of the cohort) reporting that the police were informed of one or several of their crimes last year. The calculations reveal that the size order of the delinquent acts covered by the questionnaire is typically higher than the police-recorded figures. The number of offenses detected by the police was, however, clearly lower than the number of police-recorded sus pects aged 15-17. This difference is in the correct direction as the survey only covered the most recent offense, where each respondent only can ap pear once, and the police records give the number of all incidents where there is an identified suspect. The police figures should be higher than the survey rates of offenses known to the police derived in this way also because some of the youths belonging to the relevant age bracket did not participate in the survey.
4. Delinquency and problem behavior
Tables 1 and lijn the Appendix report the main data for this chapter. At present, the focus is mainly on various types of problem behavior and of fending, less on the individual offenders.
4.1 Prevatence (Table 1, Appendix)
The prevalence rates may be seen as being more robust indicators of delin quent behavior than the frequency figures. Simultaneously they are, how ever, Iess sensitive to individual variations in the intensity of a given behavior type. Prevalence rates serve to differentiate between those who have littie or no experience in offending and those who do, but they are not very good at differentiating between high- and low-intensity offenders. As has been stated many times over (e.g, Kyvsgaard, 1991, 1992; Sarnecki, 1986, 1990) the most intensively delinquent youths are the core problem groups, rather than those with some experience of offending in general, in dicated by the prevalence results. However, it is also obvious that a high prevalence of offending is at least to some extent related to a high intensity (frequency) of delinquent behaviors. In Table 1, the ‘ever’ prevalence sum scales indicate that almost everybody (i.e., over 90%) has at least some ex perience with some problem÷alcohol behavior (mainly alcohol and truancy) as well as of ‘other youth-related offenses’ (mainly fare dodging). Property crimes are reported by three-quarters, shoplifting and stealing at school being its most common forms. Violence, then, is reported by less than two thirds, and even then it is mainly vandalism. Other violence is limited to participation in ‘riots’, and concrete personal violence is rather uncommon:
only a few percent admit to such acts. The drug use scale indicates that one respondent in five has had some experience with drugs or narcotics (almost exclusively soft narcotics). In our questionnaire, drugs also inciude glue and solvents that are usually sniffed and the use of which, as far as known,
