« Previous | Contents | Next »
Listen
THE 2000 SCOTTISH CRIME SURVEY: OVERVIEW REPORT
Chapter 1 Introduction
Background
The Scottish Crime Survey (SCS) measures crime against people living in private households throughout Scotland. The 2000 SCS was the fifth such survey to be carried out in Scotland since the early 1980s. In 1982 and 1988 the Crime Survey in Scotland formed part of the British Crime Survey (BCS). However, in 1993 the first independent SCS was run in Scotland and was repeated in 1996 and 2000. The (BCS) has been carried out in England and Wales in 1982, 1984, 1988, and every two years from 1992 to 2000.
The main aims of the SCS are to:
- provide an alternative measure of crime to the police recorded crime statistics
- examine trends in the level and nature of crime over time
- assess varying crime risks
- collect information on a range of other crime-related issues (e.g. concern about crime; attitudes to the police; drug misuse; domestic violence)
Purpose and Limitations of Crime Surveys
The principal value of surveys such as the SCS is that they provide an alternative index of crime to set alongside the police recorded crime statistics. For a variety of reasons, not all incidents of victimisation are reported to the police and, of those that are, not all are subsequently recorded by them. Crime surveys can provide a more complete picture of victimisation levels by including those incidents that the police do not record. Moreover, by repeating national surveys at regular intervals, they can provide valuable trend information. Crime surveys also fill in some details of our picture of crime by collecting information about the nature and impact of victimisation that is not available from other sources.
Surveys such as the SCS do, however, have a number of important limitations, and these should be borne in mind when interpreting the data. The SCS does not provide a complete picture of crime in Scotland. It is a survey of adults in private households; as such it does not collect information about crimes committed against corporate bodies, individuals not resident in households and those aged under16.
Also, it should be understood that the survey cannot provide a perfectly reliable measure of crime. One limitation is that the SCS is dependent on the willingness of the public to take part in the survey. Not all are willing, and this may lead to some response bias. Also, it cannot be assumed that the information provided by individuals is entirely accurate: the requirement to remember incidents and report them accurately is quite demanding. There is also a possibility that public perceptions of crime and victimisation may change between surveys and that this may affect recall and the nature of how matters are reported.
The SCS results, like those from other sample surveys, are subject to sampling error and the design effect of clustering the sample. The findings should therefore not be seen as exact measures but as indicators set within margins of error.
The SCS does not, therefore, claim to measure the 'true' level of crime in Scotland. For many kinds of crime, however, including the most prevalent, it provides a better indicator of levels and trends in victimisation than police recorded crime statistics.
The Design of the Survey
The results reported here are based on responses from a representative sample of 5,059 adults (aged 16 or over), interviewed in their homes between January and May 2000 by representatives of MVA. The interviews were obtained from an issued sample of 7,6517,651 addresses across 334 sample points on the Scottish mainland and islands.
Despite the survey design and methodology being identical to the 1993 and 1996 sweeps, and many of the same interviewers being used, the 2000 survey was much more difficult to complete than previously. It is not easy to give definitive reasons for this, but it does reflect difficulties widely experienced in the market research industry in recent years (see Appendix B for further details). As a result of these difficulties the survey took longer to complete than before, and the response rate fell from 77% in 1996 to 71% in 2000.
In addition, 403 young people between the ages of 12 and 15, resident in households selected for the Survey, completed a short questionnaire which covered their experience of crime, policing and other issues. Additional 'boost' interviews were also conducted with 380 adults and 44 young people belonging to ethnic minorities.
The main difference between the independent SCS run since 1993 and earlier surveys lies in their geographical coverage. In 1982 and 1988, the sample for the BCSs was drawn solely from central and southern Scotland. In 1993, 1996 and 2000, geographical coverage was extended to include the whole of mainland Scotland and the larger islands, thus giving a more comprehensive picture of crime in Scotland.
The structure of the interview is detailed in Appendix B. Apart from some adjustments to the questionnaire content the 2000 SCS was virtually identical in design and survey methodology to the 1996 and 1993 surveys. In all key respects other than geographic cover, the structure of the 1993 SCS was based on that of the 1988 BCS (Scotland). However, there were a few methodological and design changes between the 1996 and 2000 SCS. These are discussed in Appendix B. In summary, these were:
- the inclusion of an ethnic minority booster sample
- the addition of a 'screener question' on domestic violence
- the omission of questions on self-reported petty offending in the self-completion questionnaire
- change in the format of the young persons' questionnaire
- removal of self-reported drug taking questions in the young persons' questionnaire
- changes to the 'public perception of the police' section of the survey
- some new questions in the victim form
Aims and Content of this Report
For the first time in 2000, the SCS findings are reported in a modular format rather than a single document. This report provides an overview of the key findings, concentrating mainly on the extent and distribution of crime. Separate reports providing further, more detailed information on individual subjects such as vehicle crime, housebreaking and young people will be published in due course (see back page).
The contents of this report are introduced below.
Chapter 2 provides estimates of the level of crime in Scotland in 1999.
Chapter 3 examines trends in different types of crime, comparing SCS 2000 estimates of crime levels with those of previous crime surveys.
Chapter 4 compares SCS findings with police recorded crime statistics
Chapter 5 compares the SCS findings for Scotland with those from the British Crime Survey (BCS) for England and Wales
Chapter 6 looks at variations in the risk of victimisation between types of people and households.
Chapter 7 discusses fear of crime and public reactions to crime.
It should be noted that the SCS provides a very extensive range of information. It cannot all be presented in a single report: even the range of modules proposed will not by any means exhaust its potential for analyses and reporting. The data will provide a valuable source for secondary analysis to support other, related, research projects.
In addition to this report, a technical report is available which gives a detailed account of the survey design methodology and conduct, the data weighting, and the calculation of the principal indicators with their standard errors and design effects.
Methodological Notes To allow meaningful comparisons with earlier survey results, comparisons with surveys prior to the 1993 SCS are based on the subset of interviews carried out in the same areas of the country covered previously (this was 4261 interviews in 1993, 4292 in 1996 and 4271 in 2000). This is necessary because generally lower rates of victimisation in Grampian and the Highlands and Islands would otherwise serve to lower estimates of crime for 1993 onwards relative to earlier years. Unless otherwise stated, all data included in this report on proportions of households, individuals, incidents of crime etc., are weighted. Where the raw number of responses is given (signified by 'n=…'), this is unweighted. Details of the weighting processes and factors used are given in Appendix B. One of the key aims of the 2000 SCS was to measure the level and type of personal and household victimisation experienced by members of the Scottish public. From the data gathered in the SCS, it is possible to calculate three main types of indicator of the levels of crime: - estimates of total crime in Scotland - these are incidence rates for each type of crime which are grossed up to reflect estimates for the total population using projected population estimates from the 1991 Census. These are used within this report to show the extent of crime in Scotland, crime trends, and as the basis for comparisons with statistics of crimes recorded by the police.
- incidence rates (or 'victimisation rates') - these are derived from the number of incidents reported in the sample, expressed as a rate per 10,000 units. Units could be adults aged 16 or over, households or vehicle owning households. These are used to provide a more precise indication of changes in crime rates between surveys, as, unlike estimates, they are not affected by changing population or household numbers.
- prevalence rates - these show the percentage of respondents/households who reported at least one SCS crime. Prevalence rates relate to people/households, not to incidents, and therefore must be the same as, or lower than, incidence rates. This is because some people and households were victims of crime on more than one occasion. They are used within this report to give an indication of crime risks.
The sample size of the survey is the minimum deemed to be necessary to estimate the main indicators. It follows that these indicators must normally be presented for the whole of Scotland. Estimates derived from sub-sets of the data are subject to rapidly widening confidence intervals. Where findings are said to be 'significant', this is true at the 95% confidence limit (p£ 0.05). For example, if victimisation rates for a certain type of crime were said to have changed significantly between 1995 and 1999, we are at least 95% certain that the difference is not due to some chance effect of the survey sample. Statistical tests, such as two tailed T-tests and X 2 tests, were carried out where appropriate. In the presentation of tables, an asterisk (*) is used when a percentage is less than 1% but greater than zero. In tables where figures have been rounded, a discrepancy may be apparent between the sum of the constituent items and the totals as shown. |
« Previous | Contents | Next »