INTRODUCTION
Criminologists use a wide range of evidence types in their research. They do so in part because all types of evidence are fallible and it would be a mistake to depend on only a few kinds and in part because different research questions call for different types of evidence. The choice of evidence to draw on in practice is often also made on the basis of familiarity with data sources and with methods of their analysis.
In what follows we present first an account of the types of evidence that are currently used in British criminology before going on to suggest that there is scope for broadening the range of sources typically used, emphasising in particular the scope for making greater and better use of secondary ones. Finally we discuss in detail the scope for more use of victimisation surveys, focusing in particular on the Crime Survey for England and Wales (CSEW, previously known as the British Crime Survey). We also note ways in which this already high quality but underused resource could be further improved to allow for more informative and more useful criminological research.
EVIDENCE USED IN BRITISH CRIMINOLOGY
The evidence presented here for the types of evidence used in criminology is, in common with all evidence, fallible. The choice of evidence for types of evidence use was opportunistic and could certainly be improved. This was a quick and dirty look at the types of evidence that are constructed and drawn on in British criminology.
All journal articles that appeared in the British Journal of Criminology (BJC) from the first in 2013 to the penultimate in 2015 were scanned and coded for the original evidence that was assembled, analysed and reportedFootnote 1. There were 163 articles in all. Of course a proportion of articles did not report research findings: 28 per cent comprised commentaries and half a dozen focused on theory and method. Moreover the BJC is an international journal, where many papers from non-UK institutions are published, many focus on crime issues in non-UK jurisdictions, and many are written by non-UK citizens. There was no way realistically to subdivide the papers into British and non-British and no comparison is made with journals published from other countries. Moreover, what is published in the BJC may in part reflect the predilections of the editor, editorial board and reviewers as well as decisions by authors on where to submit their work. This is, thus, at best a rough and ready account of what occurs in British criminology. For the purpose of this paper it was the best that was possible and provides at least indicative evidence of the types of evidence that are used in the UK's leading criminology journal. Others may like to repeat the exercise to see whether there have been changes over time and whether different evidence constellations are found in papers in journals published elsewhere.
So what do we find? Figure 1 shows the nationality of first author's institutional affiliation. The UK comes top and furnishes just under 40 per cent, with the US and Australia coming second and third, which is not surprising given that most residents in both have English as their first language. Figure 1 suggests that ‘British’ criminology figures large albeit not exclusively in the journal and it may well be that those submitting from other places do so in the belief that their work accords with the interests and methods used in British criminology.
Country of affiliation of first author, papers appearing in the British Journal of Criminology, 2013–2015.
The data types used in the published papers reporting findings is shown in Figure 2. Three quarters (77%) used only one type of data, a sixth two (18%), and the remainder three or more. Despite much talk of multiple methods they continue to be the exception.
Types of data used in papers appearing in the British Journal of Criminology, 2013–2015.
It is clear from Figure 2, however, that a wide range of methods and data sources were used, but the interview was by far the most common means of collecting data. Of the 50 studies drawing on interviews, 40 per cent were relatively small scale with 30 or fewer participants where mainly qualitative methods of analysis would be possible (although ten had more than 100 and two more than 200). In all focus groups, observational studies, ethnography and the use of visual materials accounted for a quarter of methods used and would require mainly qualitative analysis (24% of all data types drawn on in studies reporting research findings).
There are plenty of examples of clever use of non-standard data sources, for example changing costs of cleaning buses as a measure of the effectiveness of an initiative aiming to reduce vandalism on buses (Poyner Reference Poyner and Clarke1992); health records for sexually transmitted disease amongst the very young as an indicator of illicit sexual activity; hospital records as an indicator of levels of violent behaviour (Shepherd et al. Reference Shepherd, Sutherland and Newcombe2006); police pre-existing interest in illegally parked cars in disabled driver bays as an indicator of generalized offending and antisocial behaviour (Chenery et al. Reference Chenery, Henshaw and Pease1999). The promise of unobtrusive measures of behaviour of interest has quite a long history in social science and has been advocated in the past for criminological research (Garwood et al. Reference Garwood, Rogerson, Pease, Jupp, Davies and Francis2000). Exhibit one is drawn from current research in Australia by one of us to examine patterns of crime and antisocial behaviour in open spaces. The refuse collection department of the local authority has been using this form to code all items of rubbish being picked up in parks and other open spaces in one area on each collection day. This is being used as an indicator of patterns of behaviour occurring there and of changes occurring in them.
Australian rubbish data collection form.
Equally there are plenty of examples of informative analyses of existing data sets. Secondary analysis of existing data sets, however, was used fairly infrequently in the BJC papers scanned over the past three years, occurring in less than a third of findings-reporting papers (30%). Administrative data were drawn on around a sixth of papers reporting findings (16%). Analysis of victimisation surveys appeared in only five papers, where we suggest there is scope for improvement both in use by criminologists and in the design of the CSEW in particular to create the potential for more informative analysis.
It is not the purpose of this paper to criticize the data sources created and used in papers published in the BJC. They have all clearly reached a standard where they have been accepted by a major journal. Rather, the suggestion being made here is that there may be much greater scope for conducting informative secondary analyses of data sets that have been created for administrative purposes, national scale survey research or which can be collected using administrative methods.
Let us turn now to police recorded crime and the CSEW as the most frequently used sources of data on crime patterns and changes in them.
IDENTIFYING AND UNDERSTANDING CRIME PATTERNS
The main sources for crime statistics in the UK are police recorded crime data and crime survey data. The former includes basic information about crimes that have come to the attention of the police, such as, if known, time and place of incident, target's or victim's demographic information and perhaps employment, Modus Operandi of the offender, and any other evidence that may contribute to offender detection and criminal case solving.
POLICE RECORDED CRIME DATA
A number of limitations render police recorded crime unreliable as a source for measuring crime across jurisdictions and/ or over time. The most common limitations of police recorded crime include non-reporting by the public; non-standardised recording practice across police forces and over time; as well as changes in offence classification and legal definitions over time (Hough and Mayhew Reference Hough and Mayhew1982; Van Dijk and Tseloni Reference Van Dijk, Tseloni, van Dijk, Tseloni and Farrell2012). Police crime statistics have therefore been removed from those designated as ‘National Statistics’, whose statistical rigour is recognised by the independent UK Statistics Authority. This removal occurred in 2013, only a year after initial designation. With the following health warning, however, police recorded crime is still included in the Office for National Statistics (ONS) crime reports and publications alongside crime survey – based estimates (ONS 2016).
“In accordance with the Statistics and Registration Service Act 2007, statistics based on police recorded crime data have been assessed against the Code of Practice for Official Statistics and found not to meet the required standard for designation as National Statistics. Data from the Crime Survey for England and Wales (CSEW) continue to be badged as National Statistics” (ONS 2016).
CRIME SURVEYS
Crime surveys were initially introduced to supplement police recorded crime statistics. In addition to victims' crime experiences, police reporting and satisfaction, crime surveys provide information about a wide array of related issues - the general population's perceptions and attitudes about crime, disorder and the Criminal Justice System (CJS), including the police, as well as detailed demographic, socio-economic and other factual information about the survey participants, their household and neighbourhood.
The role of crime surveys however has expanded beyond ‘counting crime’. The detailed knowledge on victim /target as well as non-victim characteristics conveyed by crime surveys has provided the foundations for theory development, especially on the topics of victimisation; fear of crime; and legitimacy of and trust in the CJS. For instance, the lifestyle, routine activities, social disorganisation and repeat victimisation strands of victimisation theory drew on (Farrell and Pease Reference Farrell and Pease1993; Hindelang et al. Reference Hindelang, Gottfredson and Garofalo1978; Reiss Reference Reiss, Fienberg and Reiss1980; Sampson and Groves Reference Sampson and Groves1989) innovative crime survey data which enabled:
(i) Statistically reliable and detailed comparisons between victims and non-victims; and
(ii) Analyses of crime focused on the crime target or victim rather than the crime incident (which is the base of police statistics).
Population-based empirically evidenced theories (utilising crime survey data) have informed crime prevention policy interventions, such as those directing resources and police protection to repeat victims (Pease Reference Pease1998; Pease and Tseloni Reference Pease and Tseloni2014). Police analysts now routinely try to identify repeat victims within police recorded crime statistics.
British Crime Survey
The first national crime survey in the UK was the 1982 British Crime Survey (BCS). The 1982 BCS which was at the time an ad hoc survey did not have the ‘1982’ attached to its name – it was simply the British Crime Survey. The BCS was initiated by the former Research and Planning Unit of the Home Office for England and Wales in mid-1981. The Scottish Home and Health Department followed suit and Scotland was added (Hough and Mayhew Reference Hough and Mayhew1982). The BCS which measured crime in 1981 had at the time an uncertain futureFootnote 2 (as indicated by the lack of year in the title), a limited budget and unsurprisingly a small sample of roughly 10,000 respondents. “The full value of the survey statistics will accrue with repeats, when information about trends will become available. Until such time, the BCS can yield information only about 1981.” (Hough and Mayhew Reference Hough and Mayhew1982: 7). This is worth remembering especially given the media ‘outrage’ on crime rates ‘hikes’ when cybercrime was first measured in England and Wales last year (Watson Reference Watson2015). The Home Office archived reports provide the justification for and detailed information on the first BCS findings. At the time of writing this the BCS has been in existence for over 30 years and evolved to the Crime Survey for England and Wales (CSEW). In what follows it will be referred to as the CSEW. A brief history and an overview of the topics currently covered by the CSEW is given in the next section.
Other UK crime surveys
Four crime surveys, in addition to the CSEW, currently exist in the UK. Scotland, which was included only in the first (1982) BCS, has had its own independent crime survey since 1993. That year the Central Research Unit of the Scottish Office conducted the first Scottish Crime Survey (SCS) independently of the BCS but with fairly similar sampling design to enable cross-national comparisons (Anderson and Leitch Reference Anderson and Leitch1996). The SCS has now evolved to the Scottish Crime and Justice Survey (SCJS) which has been running continuously since 2008/09 and prior to that year on an ad hoc basis (Grant Reference Grant2015). The Northern Ireland Crime Survey (NICS) which was introduced over 20 years ago has also run continuously since 2005 with just a few sweeps between 1994/95 and 2003/04 (Department of Justice 2015).
The Home Office has introduced two more crime surveys in addition to the CSEW: the Offending, Crime and Justice Survey and the Commercial Victimisation Survey. The Offending, Crime and Justice Survey (OCJS) is a longitudinal crime survey with rotated panel design with some new entries at each sweep examining both victimisation and self-reported offending of the general population. It was administered annually from 2003 to 2006. However high attrition rates have made the OCJS panel sample size inadequate for victimisation and offending studies.
The crime surveys discussed hitherto are based on population samples and cover individuals' and households' experience and perception of crime and related topics. The last survey mentioned here, the Commercial Victimisation Survey (CVS), examines businesses' crime experiences and prevention. The CVS which was initiated in 1993 and repeated in 2002 has been administered annually since 2012. The CVS has included the wholesale and retail sectors in all sweeps (although with changing definitions), but other sectors have been covered on an ad hoc basis. The full list of Home Office data sets is available at: http://www.esds.ac.uk/findingData/dephomeofficelist.asp?print=1 (accessed 3 February 2016).
Annual updates on the CSEW and the other UK crime surveys are provided at the Crime Surveys User Conference organised by the UK Data Service. Ad hoc presentations and updates on crime statistics, including crime survey data, are held at conferences and workshops organised by Learned Societies with such an interest. The most obvious interest groups include the Crime and Justice Statistics Network of the British Society of Criminology; the Social Statistics Section of the Royal Statistical Society and, last but not least, the Institute for Advanced Legal Studies which motivated this article - based on two presentations at the 2015 National Training Day: ‘Sources and Methods in Criminology and Criminal Justice’. Virtual opinion and information exchanges, including responding to relevant UK Statistics Authority, Home Office, and the Ministry of Justice data – related consultations, also occur within the online JISCMAIL ‘Crime-Justice-Stats’ forum and the StatsUserNet ‘Crime and Justice Statistics Network’, both open to anyone with such an interest.
CRIME SURVEY FOR ENGLAND AND WALES
Short history and data sources
The Crime Survey for England and Wales (CSEW, previously called the British Crime Survey) “is viewed as a gold-standard survey of its kind” (Flatley Reference Flatley, Bruinsma and Weisburd2014: 199). It is a nationwide crime survey for England and Wales. As a repeated cross-section survey, it had been conducted in eight sweeps (1982, 1984, 1988, 1992, 1994, 1996, 1998 and 2000) before becoming an annual survey in 2001 (Kershaw et al., Reference Kershaw, Chivite-Matthews, Thomas and Aust2001). The CSEW was first administered by the former Research and Planning Unit, Research, Development and Statistics Directorate of the Home Office from 1981 until 2012. Following an independent review of official crime statistics by the National Statistician which had been requested by the Home Secretary, responsibility for crime statistics, including the CSEW, transferred from the Home Office to the Office for National Statistics (ONS) on 1st April 2012 (UK Statistics Authority 2010; Government Statistical Service 2011). The CSEW has been conducted by different survey companies over the years. These have been subcontracted by the Home Office Research and Planning Unit and, since 2011, the ONS. Primary data analysis is conducted in house while the companies prepare the survey reports.
Since its appearance, the CSEW has undergone a large number of improvements, such as changes in sample design, procedures, questionnaire content and collected information, and / or administrative responsibility for the survey. However welcome such changes may be they make a discussion of the CSEW without a specific time reference quite laborious. With apologies to the readers who are not interested in survey methodology the following overview is indispensable in order to appreciate what the CSEW data can or cannot tell us about crime and the CJS experiences and perceptions. It is only an overview - details about the CSEW sampling design, procedures and questionnaire over the years can be found in the respective technical reports (for example, Grant et al. Reference Grant, Harvey, Bolling and Clemens2006; Hales Reference Hales1993; Hales and Stratford Reference Hales and Stratford1997; Reference Hales and Stratford1999; Hales et al. Reference Hales, Henderson, Becher and Collins2000; Kershaw et al. Reference Kershaw, Chivite-Matthews, Thomas and Aust2001; NOP Market Research Limited, 1985; 1989; Simmons and colleagues, Reference Simmons2002; White and Malbon, Reference White and Malbon1995; Wood, Reference Wood1984) and the ONS User Guides from 2011/12 onwards (available at: http://www.ons.gov.uk/ons/guide-method/method-quality/specific/crime-statistics-methodology/user-guides/index.html (accessed 4 February 2016). Flatley (Reference Flatley, Bruinsma and Weisburd2014) gives an excellent overview of the CSEW and the survey's impact on British criminological thinking and crime prevention policy.
Population coverage and sampling frame
The CSEW represents all England and Wales residents 16 or older and since 2009/10 also children aged 10–15 years old living in private accommodation. The CSEW sampling frame, a notional ‘catalogue’ of the population from which the sample is selected, has been the Postcode Address File since 1992 whilst prior to 1992 the Electoral Register was used. Therefore the survey's population coverage differs from the two nations' actual population. Students in university accommodation, pensioners in old people's care homes, disabled people in care homes, permanent residents in hotels or motels, members of staff and their families living in hospitals, prisons, schools etc., people living in trailers, homeless and children under the age of 10 are not covered by the survey.
Sampling design, sample size and representativeness
The CSEW uses a stratified multi-stage cross-section sample design, which (a) had over-representation of inner cityFootnote 3 constituencies until 1998 and has had overrepresented low density areas since 2001/02, as well as (b) included ethnic minority booster samples until 1996. The random sample selection stages involve the following geographical strata: selection of constituencies (prior to the 2000 CSEW after first having divided them into inner city and non-inner city ones); within each selected constituency selection of postcode sectors; within each sector selection of segments; within each selected segment selection of addresses; and, in case of multi-occupancy selected addresses, selection of household. At the last stage one adult member per household is randomly selected by the interviewers to undertake the survey (since 1996 following the same methodology as in the 1992 sweepFootnote 4 and small variations in the remainder four sweeps, 1982–1988 and 1994).
In the survey's initial cross-sectional design the entire sample was introduced during the first quarter of the fieldwork year until 2000. Since 2001 the CSEW has used a continuous sampling design with monthly allocated interviews (Kershaw et al., Reference Kershaw, Chivite-Matthews, Thomas and Aust2001; Simmons and colleagues, Reference Simmons2002). Furthermore “the sample design has been refined in recent years to move to a partially clustered sample design involving an unclustered sample of addresses being drawn in the most densely populated areas, with more clustered designs in the medium population density and low population density areas” (Flatley Reference Flatley, Bruinsma and Weisburd2014: 195).
The current CSEW sample size is roughly 35,000 households per year. However, as already mentioned it started with just under 10,000 interviews in 1982, roughly doubled in 2000, further increased to about 33,000 in 2001/02, then reached 45,000 interviews annually to ensure partial sample representativeness within Police Force Areas in 2004/05 and finally has dropped again to its current size since 2012/13.
The survey has consistent high response rates of 78% on average (Jansson Reference Jansson2007) albeit currently maintained at 70% (Williams and Holcekova Reference Williams and Holcekova2015). It represents the adult population in England and Wales with the caveats mentioned in the previous sub-section. To ensure this a number of weights (based on calibration of population estimates originally from the Labour Force Survey and currently the 2011 Census) are provided in the public domain CSEW data sets. Their purpose is to “adjust for the different selection probabilities involved in the respective [sample] designs” (Flatley Reference Flatley, Bruinsma and Weisburd2014: 196). Thus they redress any population group non-randomness and non-representativeness introduced at (i) the first (prior to 2000 CSEW inner city and current Police Force Area weight) and (ii) last two stages of the sample selection process, including differential non-response (household and individual weights) as well as (iii) from the Home Office/ ONS crime counting rules (incident or series weight) (TNS 2014). The CSEW crime counting rules are overviewed in a later sub-section.
Reference Period
Prior to 2001 the CSEW reference period varied from about 12 to 15 months. Respondents were required to report any incidents that happened since first January of the year prior to the interview. Since the fieldwork period begun in January and the vast majority of interviews were conducted by the end of March [92.7% (Hales, Reference Hales1993: 13)], the reference period for most respondents was between 12 and 15 months. Events reported in the interview year were excluded from crime estimates for the reference year, although they were retained in the data for other uses than annual crime estimates. The continuous sampling design introduced since 2001 uses a rolling 12 month reference period.
Events that occurred outside the reference period may however be reported because respondents in retrospective surveys, such as the CSEW, tend to ‘forward telescope’ events occurring before the reference period and report them as eligible events (Schneider Reference Schneider1981). A panel sample design can be used to ‘bound’ the reference period whilst respondents are asked to report events that happen ‘since the last interview’. The CSEW does not employ a bounding procedure. Respondents are asked to date the events that they mention, but they are not offered a cognitive benchmark similar to a prior interview. Since 2001 the respondents have been offered a ‘life event calendar’, such as Christmas, Bank holidays, birthdays and other memory pinpoints, for the 12 months prior to their interview to help them place events in a meaningful way (Kershaw et al. Reference Kershaw, Chivite-Matthews, Thomas and Aust2001). Interviewers however have no listings of previous victimisation events against which to check reports in the current interview. ‘Telescoping’ would result in an overestimate of more serious events rather than more minor ones (Aye Maung Reference Aye Maung and Walker1995). The use of ‘life event calendars’ in the CSEW since 2001 in principle reduces ‘forward telescoping’ but how much has yet to be estimated (Kershaw et al. Reference Kershaw, Chivite-Matthews, Thomas and Aust2001).
Mode of Interviews
The CSEW was conducted via face to face interviews up until the 1992 sweep. Since 1994 Computer Assisted Personal Interviewing (CAPI) and for sensitive questions, such as domestic violence, Computer Assisted Self-administered Interviewing (CASI) has been implemented (Hales Reference Hales1993; Hales and Stratford Reference Hales and Stratford1997; Hales et al. Reference Hales, Henderson, Becher and Collins2000; Kershaw et al. Reference Kershaw, Chivite-Matthews, Thomas and Aust2001; White and Malbon Reference White and Malbon1995). Information on the effects of computer assisted interviewing suggests that using computers in the interview process will lead to increases in the reporting of victimisation (Hubble and Wilder Reference Hubble and Wilder1995). “Evidence from the USA suggests that there may be a slight rise in the recording of sensitive data” (White and Malbon Reference White and Malbon1995: 2). Indeed when CASI was implemented for questions about domestic violence in the 1996 CSEW the recorded prevalence of such incidents significantly increased (Mirrlees-Black Reference Mirrlees-Black1999).
Questionnaire Design and Content
The CSEW questionnaire comprises the following parts: the Household Details (previously called Address File), the Main Questionnaire, the Demographics and Media Consumption Section, Special Modules (former Follow-up Questionnaires), the Victim Forms and ad hoc special topics modules.
The Demographic Section collects information about the characteristics of the respondent, his/ her household and the ‘household representative person’ (previously termed ‘head of household’). It also asks about respondent's media consumption and sources of information on crime and related issues.
The Main Questionnaire collects information on crime experiences and perceptions - fear of and worry about crime, perceived incivilities and crime trends in the respondent's neighbourhood. In recent sweeps new forms of crime, such as mobile phone theft, plastic card and mass marketing fraud and cybercrime, are also investigated in the Main Questionnaire rather than the Victim Forms. This has been done in order to preserve the CSEW – based crime estimates time series' continuity and validity. Respondents aged 16–59 are currently asked about drugs and alcohol abuse, offending behaviour, domestic violence, sexual victimisation and its nature and stalking via a set of self-completion modules. Rotating self-completion modules have been recently introduced to cover the high demand for knowledge of specialist criminological topics without increasing interview length and associated survey's costs (financial and respondent fatigue).
The screening questions of the Main Questionnaire focus on crime experiences and are organised so that there is a one to one correspondence between question and crime type. The screeners are designed to elicit mention of particular crime events, using the legalistic approach to screening (Cantor and Lynch Reference Cantor and Lynch2000). Each question is designed to prompt mention of a particular type of crime event, e.g., burglary. All of the particulars necessary to classify the event as a specific type of crime are included in the screening question. As a result, the cues are embedded in long and complex sentences. One unintended consequence of this is the research-evidenced positive relationship between victimisation and education because more educated respondents are able to recognise, recall and report victimisation events (Hough Reference Hough1987). The screeners also ask the respondent to report the number of occurrences of each crime type.
The selected household member for the CSEW interview reports about any type of crime against his/her household involving personal or household property, including motor-vehicles, as well as any violent crime, theft or intimidation he/she has personally experienced. Until 1996 only female respondents were interviewed on incidents of sexual assault, rape, and obscene phone calls whereas all respondents thereafter.
Respondents who report crime experiences in the screener questions of the Main Questionnaire are then given up to sixFootnote 5 further modules, the Victim Forms, to complete. Victim Forms are completed for crimes reported in reverse order of the screeners – that is the most serious crimes have priority over less serious ones. The descending order of crime seriousness is: rape and sexual assault, robbery, assault, theft from person, burglary, theft from dwelling, vehicle theft and vandalism (Hales Reference Hales1993). In addition the CSEW includes data on threats, fraud and cybercrime which however are not part of the time series yet. Out of the six Victim Forms per victim the first three are long Victim Forms. They collect detailed information about the crime incident, its circumstances, offenders' modus operandi, related loss and /or harm, reporting to the police and reasons for non-reporting, the criminal justice system response to the crime and victim's satisfaction. A further three (short) Victim Forms collect less detailed information but sufficient to classify the incident for counting purposes. Back office checks ensure that the reported incident is correctly classified into one of the crime categories which also resemble as much as possible the statutory definitions of offences (TNS 2014: 58). This information is used to classify the incident as the most serious crime type that occurred during the event.Footnote 6
This practice encourages full reporting (and counting) of the most serious and less common crime types without of course any guarantee in case of serious multiple victimisation. For this reason the screening questions of the Main Questionnaire have been used in the early CSEW - based academic work (Chenery et al. Reference Chenery, Ellingworth, Tseloni and Pease1996; Osborn and Tseloni Reference Osborn and Tseloni1998). Indeed crimes can be analysed without using the data from the Victim Forms. It is important to keep in mind however that when using the screeners less serious crimes, which happened in the course of more serious ones and the respondent has not subsumed them in the latter, may be over-counted and crime types may be incorrectly defined.
The Special Modules and former Follow-Up Questionnaires have been administered to different portions of the CSEW sample to maximise the number of themes covered. Prior to 2001/02 two partially or completely (depending on year) Follow-Up Questionnaires were administered to respective randomly selected sub-samples. These collected detailed information about attitudes and perceptions of respondents on concepts relevant to fear of crime and victimisation theories, such as lifestyle and routine activities, crime prevention measures, self-reported offending, attitudes towards and experience of the police, misfortunes, attitudes towards offenders' punishment etc. Apart from the Follow-Up questionnaires certain themes, such as self-reported drug use, sexual victimisation and young people crime experiences, were investigated in periodic questionnaire supplements. The increase of the sample size since 2001/02 has allowed the introduction of at least four questionnaire modules (administered to four randomly selected sub-samples) to explore topical issues, such as experiences of the police, attitudes to the Criminal Justice System, crime prevention and security, online security and anti-social behaviour (Downham Reference Downham2015).
The CSEW for 10–15 years old is set out in the same way as the main survey, including the victimisation module, but the questions refer to schooling experiences, perceptions of crime, crime prevention and security. There is also a self-completion module covering bullying, street gangs, truancy and alcohol or drug use.
Series
Some crimes, such as domestic violence, are recurrent events that cannot be clearly distinguished. To overcome this, the CSEW classifies as series any repeat incidents which were very similar in that “the same thing was done under the same circumstances and probably by the same people” (Hales Reference Hales1993: 12, Main Questionnaire, Appendix B). There is no threshold number of events required for invoking the series procedure but their number is truncated at five when used for calculating national crime rates. In addition only one Victim Form is completed for such recurrent victimisations, namely series, collecting information about the most recent event in the series.
THE WAY FORWARD FOR THE CSEW: THE USERS' PERSPECTIVE
Based on the authors' and their colleagues (notably Professor Ken Pease and Professor Graham Farrell) past research a number of limitations of the CSEW have been identified which have hindered further in depth analyses and as a consequence limited potential insights for crime prevention policy. Particular suggestions for further improving the use of this already gem-like criminological data source are outlined in this section.
Sample size
It would be very helpful for the CSEW sample size to be large enough to be representative at the Police Force Area level. This would (a) facilitate local level policing, crime and crime perceptions research, (b) offer an abundance of practical policy and crime prevention insights at Local Authority level and (c) help with the Crime and Safety Partnerships and their PCCs budgets as they will not need to commission local victim surveys (usually of questionable quality, statistical standards and therefore usefulness).
Survey methodology
To facilitate repeat victimisation and crime seriousness escalation research, especially looking at transition probabilities of repeat victimisation the CSEW could incorporate a panelFootnote 7 of respondents who are surveyed across different years. Keeping the same CSEW respondents across 2 or more years (waves) would offer ample insights on the above (and perhaps increase the rate of repeats).
Sampling points - common across a range of central national social surveys
Crime victimisation, its impact and seriousness, experience of ASB, other crime perceptions, attitudes towards the police and the criminal justice system and other CSEW examined topics are linked to individuals' and their areas' social capital (Sampson et al. Reference Sampson, Raundenbush and Earls1997). Other measurements, such as political persuasion, for instance, are linked to attitudes and trust to the criminal justice system and punitiveness. A single survey examining all these issues is impossible due to respondents' fatigue and associated financial and quality costs. Other national surveys, such as the British Household Panel Survey, Understanding Society, and the Citizenship Survey, investigate various aspects of life that are arguably related to crime and crime perceptions. At the time of writing linked surveys exist but the CSEW is not one of them. Linking responses across comparable ONS surveys offers immense possibilities for social research, theory development and policy. The next paragraph provides a bit more information on social capital measurements.
Questionnaire content
The CSEW includes few social capital measurementsFootnote 8 on an ad hoc basis notwithstanding their significance for informal ‘guardianship', level of affinity felt to the area and resilience to victimisation. Social capital indicators refer to formal and informal connections between neighbours, citizens and their perceived access to local and /or national government's decision mechanisms. The theoretical aspects of social capital refer to bonding, bridging, trust, sense of community, social cohesion, social support and solidarity. Social capital constructs include: number of close friends that the respondent can rely on or confide in; number of friends to borrow money from; mutual aid or perceived solidarity which refers to helping the community in case of an emergency, as well as collective efficacy and collective mobilisation or non-formal social participationFootnote 9 (Sampson Reference Sampson, Cullen, Wright and Blevins2006; Sampson et al. Reference Sampson, Raundenbush and Earls1997).
Prior victimisation by burglary, car crime, and assault in the period before the reference year is the most important predictor of recent victimisation risk (Hope et al. Reference Hope, Bryan, Trickett and Osborn2001) and frequency (Osborn and Tseloni Reference Osborn and Tseloni1998) against victims with otherwise identical socio-demographic, economic, lifestyle, and area of residence characteristics. Do prior crime experiences indicate long-term victimisation which may be conceptualised similarly to long-term unemployment or simply a constellation of bad spells? However these questions were dropped after the 1992 sweep. Prior victimisation screener questions are important but far from perfect as the reported crime experience is not time – bounded (open to telescoping problems), validated or cross-checked. Using a panel element for examining repeat and multiple victimisation in depth, as suggested earlier, can address these problems.
Examining an abundance of individual and area information in relation to crime experience is one of the greatest advantages of crime survey data compared to police recorded crime. Past research on area characteristics (usually concurrent and interacting with individual and household ones) affecting victimisation risk and frequency (Osborn et al. Reference DR, Trickett and Elder1992; Osborn and Tseloni Reference Osborn and Tseloni1998; Tseloni Reference Tseloni2006; Tseloni and Pease Reference Tseloni and Pease2015) as well as fear of crime (Brunton-Smith and Sturgis Reference Brunton-Smith and Sturgis2011) has relied appropriately on Census and police data.Footnote 10 Rich as it may be Census area profiling however lacks information on respondent's area of residence place or location use, such as residential or mixed use, and environmental clues. Information on distance from city centres and areas with concentrated leisure activities would also help explain victimisation. Interviewers' assessment and /or official data from ONS/ local authority on sample point ‘place use’ and distance from commercial and leisure activities centres would address the single limitation of CSEW compared to police recorded crime data for victimisation and situational crime prevention research.
A contested issue since the origins of the CSEW has been the measurement of fear of crime (Ditton and Farrall Reference Ditton, Farrall, Hough and Maxfield2007; Sutton and Tseloni 2011). It has been shown that more exact wording and asking fear of crime questions in different sections of the questionnaire elicits responses that offer a more accurate picture of this issue (Farrall and Ditton Reference Farrall and Ditton1999; Gray et al. Reference Gray, Jackson and Farrall2008; Tseloni and Zarafonitou Reference Tseloni and Zarafonitou2008).
For improving crime prevention CSEW- based insights the number of long Victim Forms could be increased from three so that detailed incident information may be gathered for and possible links made across more than three crimes per victim.
Finally, employing consistent questioning over time allows reliable time series not just for crime rates which is something the CSEW can be proud of, but also crime – related variables, such as routine activities and trust to the police and the criminal justice system. By contrast, absolute income levels are not comparable over time and in this case providing income data as quartiles would be more appropriate for time series. Further, new crimes are defined or emerge, such as the newly introduced law relating to domestic violence (Serious Crime Act Section 76, 2015), which includes coercive control and emotional abuse. To keep up with the signs of the time additional crime types within the CSEW could be introduced in a manner that does not affect existing time series.
Substantive investigation
The definition of series crimes is rather arbitrary and fits crime counting purposes rather than criminological theory and empirical research (Farrell and Pease Reference Farrell, Pease, Hough and Maxfield2007). A way forward is arguably an exploratory module for series victims to explore, for instance, how best they can be portrayed / counted within crime statistics; the risk or protective factors and causes of series victimisation in comparison to no, single, repeat or multiple victimisation; the consequences of series crimes on health, quality of life, employment opportunities etc.; whether perhaps series victims should be conceptualised and their needs accommodated in ways akin to those relating to individuals with long-term disability; and recommendations for potential avenues to prevention.
The escalation or not of crime seriousness during the same incident in composite crimes (Tseloni et al. Reference Tseloni, Ntzoufras, Nicolaou and Pease2010a) is another area worth exploring. It is important for crime prevention and harm reduction as demonstrated in the ample literature on situational risk and preventive factors for crime completion (versus attempt), seriousness and likelihood of injury. Thus situational characteristics will be identified that would limit or erase crime seriousness escalation and on that base inform context - specific advice that should be given to potential victims.
CONCLUSION
This paper has reviewed the kinds of data apparently most often used in British criminological research. It has argued that more use might be made of available data sources, rather than depending on the limited data collection resources that are often available to researchers. This is not to criticise published studies, only to suggest that some strong resources seem to be seldom used.
The paper went on to a more detailed discussion of national victimisation surveys, homing in on the Crime Survey for England and Wales (CSEW) as a particular example, which provides a rich array of data collected over some 30 years.
Although we are enthusiastic advocates for greater use of the CSEW in criminological research, the foregoing discussion makes clear a) that there is much devil in the detail and that the unwary therefore risk making misleading use of it and b) that there is substantial scope to improve this already excellent research resource.
To finish we have three recommendations. First, criminologists are advised to broaden their search for data sources on which to devise and test theories. Many potentially fruitful possibilities seem to be widely overlooked. There is scope for more imagination is using existing sources of information. Second, we recommend in particular that crime surveys are used as standard resources for criminological research and that those contemplating research careers should develop a good understanding of them and their possibilities. Third, we urge the custodians of victimisation surveys to facilitate their use and to canvass ways in which they could become ever more informative for criminological research. There is scope to improve an already strong resource.
Acknowledgements
We thank Dr Charlotte Harris, Executive Director of the British Society of Criminology, and the organisers of the National Socio-Legal Training Day: Sources and Methods in Criminology and Criminal Justice for inviting us as well as the Editor of this journal, Mr David Wills, for sustained encouragement. Professor Ken Pease, Professor Graham Farrell, Professor Jim Lynch and the BCS /CSEW statisticians at the Home Office /ONS are gratefully acknowledged for lengthy discussions and tireless clarifications on the crime survey data over the years.
