More Info

As part of the continuing commitment to the development of criminal statistics, the
National Statisticians at the Home Office teamed up with the British Society of
Criminology and the Scottish Criminology Society to organise the second Workshop
on Crime and Justice Statistics. This year’s meeting was held at the Royal Society in
Edinburgh on 14 March 2003 and was hosted by Professor David Smith, University of
Edinburgh. Professor Smith’s welcoming and opening remarks are reproduced in full
below and is followed by a report of the Workshop.

Reflections on Criminal Statistics

David J Smith
Edinburgh University

I am aware that the convener at an event like this is not expected to say anything
substantive in the opening remarks. I won’t break too blatantly with that tradition, but
in preparing for the workshop, I have been struck by the remarkable development of
government statistics on crime and criminal justice over the past 20 odd years. In
fact, transformation would not be too strong a word.

I give three examples.

1. Police recorded crime was the sole measure up to 1982. Since then, the crime
surveys have provided an alternative measure of crime and of trends in crime; and
have provided the basis for many kinds of analysis that were not possible before.

2. Statistics on criminal process have become more complete. For example, statistics
on arrests have been introduced.

3. There has been a co-ordinated effort to monitor the interactions of ethnic minorities
with criminal justice by producing a whole new range of statistics: e.g. ethnic booster
samples in crime surveys, homicide statistics by ethnicity of victim and suspect,
police stops by ethnicity of suspect, arrests by ethnicity, prison receptions and prison
population by ethnicity.

Also, the greatly increased volume of information has become far more easily
available than before, primarily through the government websites. Not only are the
reports easier to find, but much of the statistical data can now be directly accessed by
researchers outside government.

Although the expansion of information over the past 20-odd years seems striking, the
Home Office stated in its review of crime statistics published in July 2000 that ‘the
current methods of producing national crime statistics at the Home Office would still
be recognisable to the nineteenth century officials who invented them’. The review
(which focused just on crime, leaving out the much wider field of criminal justice)
envisaged a further, more radical transformation of crime statistics over the following
five years to meet a range of objectives. Some of these objectives seem utopian. For
example, ‘The new system will be one where there exists a clearly understood and
reliable picture of the performance of government and communities in reducing
crime’. I am not sure that the performance of government or communities in reducing
crime ever can be reliably measured. But it seems piquant and novel to be
complaining that Home Office officials are too starry eyed. My point is that statistics
have already been expanded enormously, and government is still determined to
expand them much further still. It seems worth reflecting on how and why that has
come about: without of course stepping outside my role as convener, or saying
anything opinionated or other than bland.

In part this transformation flows from the growth of information technology, and from
the development of survey and other research technologies initially within the private
sector and for commercial purposes. Also in part the transformation reflects a
fundamental change in the salience of crime and criminal justice in political debate. It
has moved from being a subject of dull cross-party consent prior to 1979 to being a
central, often controversial issue in policy and politics. This is not just a British
phenomenon, it is a cross-national change. For example, we have seen a French
presidential candidate, Lionel Jospin, miss the presidency largely because of the lack
of a policy on what the French call ‘la sécurité’.

Another related change is that political debate - if you like, political rhetoric - and
discussion about actual policy has been brought much closer together. There was a
time not so long ago when the Home Office saw itself as allowing the Home
Secretary to throw red meat to the party conference but meanwhile got on quietly with
developing different policies in another part of the wood. Now that policy on crime
and criminal justice has long ceased to be bipartisan, and has become central to party
platforms and government programmes, all that has changed. We have seen the
development of a bewildering array of initiatives and policies often following one
another in quick succession, all tied in to a set of consciously enunciated ideas and
principles. Some of this is inward-looking: about getting the system to do whatever it
is doing more efficiently: a bigger bang for your shilling. But much of it also is
outward-looking: about getting the system to do different things and finding out if
they work better. Obvious examples are antisocial behaviour orders, parenting orders,
electronic tagging, mediation and restorative justice programmes.

This far more politically salient and managerially proactive criminal justice policy
needs to be serviced by far more effective information systems. It is no accident
therefore that the last 20 years have seen the development of far more extensive
criminal justice statistics. At the same time, this development was not inevitable. It
emerged out of the choices of everyone involved. Another factor has been the
growing commitment to evidence-based policy in government in recent years. A
genuine commitment to evidence-based policy requires a respect for evidence
especially when it conflicts with ideology, pre-conceptions, or existing policy. The
tradition of having an independent and incorruptible statistical service was established
in the nineteenth century. It can be argued that over the recent period of expansion of
statistical services, this tradition has been strengthened. For example, there is now an
explicit and detailed code of practice for statistics in government. This is a very
important element of a mature democracy. Its importance is re-emphasized by the
increased availability of government statistics.
As I have said, the expansion of crime and criminal justice statistics was not
inevitable, and I have often wondered just how the senior Home Office officials of the
day, Sir Brian Cubbon and David Faulkner, managed to sell the idea of the British
Crime Survey to Willie Whitelaw. In fact, I often ask my students to prepare the
brief. How would it go?

‘Home Secretary, every year you go to the House of Commons to explain that police
recorded crime has risen. We think you should have a second measure of crime – so
that you have to go to the House of Commons twice a year to explain that crime has

‘A further advantage, Home Secretary, is that this second measure will reveal that
there is four times as much crime as everyone previously thought…’

I can only speculate that the tactical argument in favour was that the crime survey
would show a gentler slope in the rise of crime than the recorded crime statistics, but
it would have needed deep insight to foresee that result; alternatively, a purely
cynical, tactical argument is that it is better to have two measures rather than one,
since one would be bound to show a more favourable trend than the other at any given
point in time; or that a multitude of measures would make it easier to confuse and

Currently under discussion are moves to make the definitions underlying police-
recorded crime correspond more closely to victims’ perceptions and less closely to an
assessment by the police of the strength of the evidence. That kind of change will
increase the count of police-recorded crime. Again it is hard to imagine the
arguments deployed with ministers in favour of such a change. ‘The advantage,
Home Secretary, is that when we introduce this change, no-one will know whether
crime has gone up or down…’

Whether or not tactical or cynical arguments have been deployed to gain acceptance
for these transformations in crime statistics I do not know. In any case, the fact is that
the expansion and reshaping of these information systems does represent the triumph
of reason even in an age that my modish academic colleagues (not modish in their
dress sense of course) want to call post-modernist. I would draw attention to four
things that are increasingly made possible by these developments.

1. Comparative research. Much of what we know in social science comes from
international comparisons. Take the causes of unemployment. Our understanding has
been enormously enhanced by making comparisons between low unemployment and
high unemployment countries. In order to do that, it is crucial to have standardized
measures of unemployment that support international comparisons, and these have
been developed for example by the International Labour Organization. Or take the
returns to education or the consequences of social mobility. Again international
comparisons are crucial, and they depend on standardized measures of level of
education and social stratification. Similarly our understanding of the causes of crime
and the effects of crime reduction policies will be enhanced by international
comparisons, but these are very insecure at present because measures of crime are not
internationally comparable, for the most part. A crucial benefit from the development
of crime statistics is that this will eventually support much more refined comparisons.
In the first instance, close comparisons between Scotland and England & Wales
would be enormously useful.

2. Monitoring the delivery of programmes and policy and the management of criminal
justice services. There has been increasing emphasis in recent years on monitoring
the results of policy. In practice, and despite government rhetoric, most of this
monitoring is of outputs rather than outcomes: it focuses on what the system or the
programme does, whether it produces the outputs that it was meant to produce. There
is also a place at least occasionally for monitoring outcomes: what has the policy or
programme changed in the wider world. This is the distinction between counting the
number of criminals caught and punished, as against evaluating any reduction in
crime brought about as a result. The large investment in producing information about
crime as opposed to criminal justice process gives some assurance that information
systems are concerned with outcomes as well as outputs.

3. As I already pointed out, improved information about criminal justice process is
supporting the systematic monitoring of the fairness of the system in a plural society,
with particular emphasis on fairness to ethnic minorities.

4. The expansion of statistical information supports open debate about policy, and
hence underpins democracy. As I pointed out earlier, 25 years ago the real debate
about criminal justice and crime reduction took place in private in Queen Anne’s Gate
or in gentlemen’s clubs in St James’s, while a ritualised debate took place in the
public arena which often had little influence on policy. As more information is
collected, and crucially made available and usable to people outside government
through modern information technology, the habit of discussing criminal justice
policy like grown-ups becomes entrenched.

I have mentioned four things that are made possible by the improvement and
expansion of statistics on crime and criminal justice. The four sessions in today’s
programme almost (but not quite) map onto those four objectives.

Report of the Workshop

Roger Tarling
University of Surrey

The day was split into four sessions:

   1.   Recent changes in the measurement of crime in the UK
   2.   Joining up the criminal justice system
   3.   Measurement of diversity within the criminal justice systems in the UK
   4.   International comparison of criminal justice statistics

Speakers from all three criminal justice systems within the UK (England and Wales,
Scotland and Northern Ireland) outlined developments in their systems.
Approximately 100 delegates from throughout the UK attended the meeting.
Crime statistics

Jon Simmons and Tricia Dodd began their session by pointing out that many of the
developments anticipated at last year’s meeting had come to fruition. Police recorded
statistics and findings from the British Crime survey are now published at the same
time in the same publication; Crime in England and Wales 2001/2002 Jon Simmons
and colleagues, Home Office Statistical Bulletin, 07/02. Simultaneous publication
provides a more comprehensive picture of crime – especially as the BCS has been
expanded in size to 40,000 interviews. Recorded crime is now presented at the
regional, BCU (Basic Command Unit) and CDRP (Crime and Disorder Reduction
Partnership) level in addition to the national and police force level.

Recorded crime statistics, however, are still going through a process of change.
Changes brought in 1998 to the counting rules had produced a sharp rise in recorded
crime and a discontinuity in the recorded crime time series. However, additional
disruptions in the series were imminent due to the introduction of the National Crime
Recording Standards (NCRS). More and more interest in measuring crime changes at
the local level, to assess need, local performance and targets, prompted a review of
the accuracy and comparability of recorded crime. In a study of police crime
recording practices, Burrows et al (2000) Review of Police Forces’ Crime Recording
Practices, Home Office Research Study No 204 pointed out that not all allegations
reported to the police were recorded as crimes. In fact only about one half were
’crimed’. There are often good reasons why allegations should not be crimed, but
Burrows et al noted two approaches that the police were pursuing. They termed these
the ‘prima facie model’ and the ‘evidential model’. In the former the details of
alleged crimes are taken at face value, and recorded without scrutiny. In the
evidential model, details of any incident reported to the police are challenged and
validated, in the same manner that might be expected if the case were to be presented
in court in order to charge a suspect.

The NCRS encourages all forces to adopt the prima facie model. Some forces
changing from the evidential model to the prima facie model are likely to see a rise in
recorded crime as many incidents which would not previously be recorded as crime
will be recorded in future. In 2002/2003 it was estimated that the NCRS had an
impact of 10% on all crime - that is, crimes counted in that year were 10% higher than
they would have been under pre-NCRS recording. The effect varied considerably
across different crime types with, for example a fairly small (3%) effect calculated for
burglary in a dwelling and robbery, but a much larger (23%) affect on violence
against the person

Developments in England and Wales are reverberating in Scotland and Northern
Ireland. In October 2002 a Scottish Crime Recording Standards Working Group was
set up in Scotland and decided to adopt an ‘holistic approach’, according to Susan
Provan, Strathclyde Police. Among the issues being addressed were a victim-based
approach to recording crime, a definition of ‘clear up’ and a review of the counting
rules. To implement the changes a common manual on crime recording is being
prepared for use by all eight forces in Scotland. In addition, National training courses
were being set up and auditing practices and procedures were being established.
An interesting development in Scotland is an examination of the crime groupings and
classification of crime used. The plan is to marry the groupings with the codes used
in the ISCJIS (described more fully below). Twelve groupings have been proposed:

       Group 1: Non sexual crimes of violence
       Group 2: Crimes of indecency
       Group 3: Crimes of dishonesty
       Group 4: Damage and recklessness
       Group 5: Proactive policing crimes
       Group 6: Crimes against public justice
       Group 7: Public disorder
       Group 8: Licensing offences
       Group 9: Environmental offences
       Group 10: Community offences
       Group 11: Motor vehicle offences
       Group 12: Miscellaneous offences

The impact of recent changes on the level of recorded crime was illustrated by Tony
Mathewson of the Police Service of Northern Ireland. In 1989 about 60,000 crimes
were recorded – a figure which remained remarkably constant for a decade. The
change in counting rules in 1998 led to an 82% increase in recorded crime. The level
of crime settled down for the next two years until electronic crime recording
procedures were introduced when crime went up a further 17%. (Some of this rise
may also be due to the anticipation of NCRS.) The end result is that recorded crime is
now running at 140,000 per year – more than double a decade or so ago. Little of the
increase is, perhaps, attributable to an increase in the underlying level of crime.

The criminal justice system

Projects are underway in both Northern Ireland and in Scotland to integrate the
various criminal justice agencies’ IT systems. Brett Hannam described the Causeway
Programme in NI and Sandy Taylor the ISCJIS (Integration of Scottish Criminal
Justice Information Systems) in Scotland. The systems they are designed to replace
are inefficient and costly. Many agencies are capturing and recording the same
information, often in manual or paper systems. Brett gave the example of the
sentence awarded by the court. Sentence is recorded separately by six different
agencies, all in slightly different ways. In addition the potential for error is increased
every time the data is captured which can, in turn, lead to different agencies releasing
different statistics.

An objective of the Causeway project is that “information will be captured once at the
point it enters the justice system and then shared and re-used electronically”. Each
agency has the right to design, build and operate their own business systems but must,
in future, meet centrally agreed standards.

The perceived benefits for statistics include:

   •   More accurate – fewer opportunities for human error
   •   More timely and speedier reporting
   •   More detailed
   •   More flexible and greater linkage
   •   More comprehensive coverage of decisions and processes
   •   Cheaper and more efficient to collect and less demands on agencies

Although criminal justice systems are themselves complex structures, Northern
Ireland and Scotland have the advantage of having relatively few organisations to
involve. (There is only one police force in Northern Ireland, for example.)

Pat Dowdeswell reported on developments in England and Wales. The Review of
Administration of Justice Statistics had been completed and a period of consultation
ended in November 2002. An Interagency Statistical Steering Group had been
established to take forward recommendations. These included:
    • Increasing coverage
    • Procedures for data validation and priority to data quality
    • Single, standard and detailed coding systems
    • Greater flexibility
    • One set of published National Statistics on Administration of Justice
    • Outputs more frequently reported than annually
    • More explanation of the counting rules and differences between series

Pat said that attention was also being given to developing a new Probation Index and
a project was in hand to improve the quality of prison data. A new statistical volume
would be produced to cover sentencing statistics and their impact. In developing
outputs consideration was given to potential users. For example, new outputs were
being devised for sentencers.

Measurement of diversity

David Smith introduced this session by saying that “on the one hand, there is evidence
of discrimination against ethnic minorities at various stages of criminal justice. On
the other hand, there is evidence of elevated rates of offending among certain specific
minority groups. The joint result of these factors is that Afro-Caribbeans and black
Africans are greatly over-represented among the prison population. At the same time,
there are puzzling conflicts of evidence. For example victim surveys find that a
highly elevated proportion of offenders as described by victims are black. Yet self-
report studies have found no differences in rates of offending between ethnic groups.
Interpretation of apparently conflicting evidence can be highly contentious especially
when opinion is polarized by events such as the murder of Stephen Lawrence.
For these reasons, monitoring the fairness of the system in a plural society becomes a
very high priority. Far more relevant information is produced now than 20 years ago
when I first started to research the topic. This can be used to investigate not only
direct discrimination, but also structural properties of systems that systematically
work against specific ethnic minorities for example, the criteria for granting bail or
the decision to opt for trial at Crown Court.”

Section 95 of the 1991 Criminal Justice Act includes a duty on the Home Secretary to
publish information to help persons engaged in the administration of justice “to avoid
discrimination against any persons on the grounds of race”. As David Smith said, this
has led to more information being collected and routinely published. Information is
available by ethnic group on stop and search, arrest, caution, conviction and sentence.
But there are gaps in what is felt should be monitored, for example decisions to
charge, grant bail, and to prosecute. Nor do we know the different rates at which
members of different ethnic groups reoffend.

Where information is collected it is not always comparable. Ethnic status is recorded
differently by different agencies. The Home Office is trying to achieve consistency of

Even when data is available and consistent, differences between groups cannot be
taken as evidence of discrimination. All relevant background factors need to be
controlled for in a multivariate statistical analysis. Bonny Mhlanga underlined the
point in describing his research looking at decisions taken by the police and the
Crown Prosecution Service to prosecute or discontinue in over 6,000 cases.

Issues of fairness and equity between the two communities are of paramount
importance in Northern Ireland. Following Section 75 of the Northern Ireland Act
1998, which places an obligation on all public bodies, The Criminal Justice Board has
adopted the aim of making the criminal justice system as open, inclusive and
accessible as possible and to promote confidence in the administration of justice. A
review of the criminal justice system, which reported in 2000, recommended
developing and implementing a strategy for equity monitoring. Siobhan Morgan
reported progress on developing an appropriate methodology for determining
community background. The most obvious way of ascertaining community
background is to ask the direct question of the person. However, there are problems
of non-response and some assessments may require the analysis of administrative
data. In addition census data may be required to provide a base line against which to
judge the outcomes of decisions. Siobhan was assessing the suitability of other
characteristics that could be used as proxies for community background, such as
postcode, name, school attended, qualifications, membership of organisations,
societies or sports clubs. At present approaches are being piloted to determine the
most effective system to put in place.

International comparison of criminal justice statistics

David Smith pointed out in his introduction to this session that “international
comparisons of criminal justice statistics reveal some extraordinary facts. For

   •   The rate of imprisonment is ten times as high in the US and Russia as in a
       number of European countries. It is particularly provocative, I think, that
       towards the end of the Cold War era, the two opposing models of the good
       society were countries having exceptionally high levels of incarceration.

   •   The rate of burglary is higher in England than in the US.

   •   The streets of Belfast are safer than the streets of London, and the citizens of
       Belfast feel safer out at night than the citizens of London.
   •   Recorded crime rose tenfold in England and Wales between 1950 and 1994,
       and recorded robberies increased by 48 times. Over the same period recorded
       crime in Japan declined slightly, and there was no increase in recorded
       robberies there.

   •   There were large increases in recorded crime in all developed countries except
       Japan between 1950 and 1980. In all of these countries, the upward trend
       reversed at some point between 1980 and 1994.”

David Smith felt that explaining these contrasts and trends should be a central task for
criminologists, although they have paid far too little attention to the problem. The
explanation would have far-reaching implications for government policy. For
example, can government activate the mechanisms that have brought about low crime
rates in Japan? Are the causes that reversed the upward postwar crime trend in so
many countries within the reach of government, or beyond it?

Unfortunately it is very hard to tackle such problems today, because the information
base is so poor. For example, trends shown in recorded crime statistics for many
countries are grossly misleading in specific respects, because of changes in definitions
and procedures. For fifty years or more, economists have put tremendous effort,
channelled through various international organizations, into standardising definitions
and producing harmonized statistics on topics such as unemployment and gross
domestic product, and the technology of purchasing power parities has been
developed to facilitate international comparisons. The same period has seen the
development of national surveys using comparable methods and definitions, such as
the Labour Force Survey in Britain. Similar developments are needed in the field of
crime and criminal justice. These come later in history, but may turn out to be equally
important, David surmised.

Gordon Barclay, who has played an active part in improving international statistics,
gave as a sub title to his talk ‘Are they safe?’ Interest in international comparative
statistics is not new. Gordon quoted from the Report of the Committee appointed to
revise the criminal proportion of Judicial Statistics 12 December 1892

       “The editor should draw attention to the returns of foreign countries; and thus
       edited, the reports would gradually prepare the way for a time when
       international comparisons should be made with safety”

More than 100 years has elapsed since that suggestion was made but the time when
international comparisons are ‘safe’ is still someway off. What leads to comparisons
not being safe? Criminal Justice Systems differ between countries, as do definitions
of offences and the methods of collecting and counting crimes. Gordon provided
many examples; at what point are crimes recorded – when they are first reported to
the police or when the ‘file’ is passed to the prosecutor? Are threats and attempts
recorded? If an offence involves more than one victim how many crimes are
counted? What counts as a prison? The concept of different types of theft which are
part of the English system are not present in continental systems. There are
differences in the age of criminal responsibility. And some countries simply do not
collect certain statistics denying the possibility of any comparisons, however
qualified. Of course similar problems exist in most areas of economic and social
policy. The difference, as David Smith pointed out, is that more effort to standardise
statistics has been put in to other areas. Perhaps because internal security falls outside
EU competencies crime statistics have not been accorded priority or the resources to
develop them.

Gordon listed the attempts to date to produce better comparative statistics:

 United Nations Survey. The UN undertakes a survey every four years of member
states. Common definitions are used and countries are expected to reconfigure their
statistics as best they can to fit the definitions. The shortcomings of the survey are
that no country meets these definitions and there has been little quality control
recently. The resources to analyse the data and to publish the results have always
been uncertain and fluctuate depending on UN priorities. Six surveys have been
concluded but the 7th has still to be published. (The 6th appeared as Global Report on
Crime and Justice by Graeme Newman, 1999, United Nations.)

International Crime Victimisation Survey. To get around all the problems associated
with police recorded crime, an initiative was undertaken about 15 years ago to
conduct a victim survey simultaneously in different countries, using a common
questionnaire (translated into the languages of the participating countries). However,
the sample sizes are small (about 2,000 in each country). A varying number of
countries have bought into the four sweeps measuring crime for the years 1989, 1992,
1996, and 2000. Results from the 2000 survey can be found in Criminal Victimisation
in Seventeen Industrialised Countries, van Kesteren et al. The Hague: NSCR.

Council of Europe European Sourcebook. More recently the Council of Europe
originated a survey of criminal justice statistics based upon information provided by
National correspondents in most of the 43 Council of Europe member states. By
working with correspondents it was at least possible to address and ‘iron out’
anomalies rather than simply taking the data at face value. The first survey covered
the period 1990-96 and the second survey 1996-2000 – but the latter has still not been
published. Reference:

The most recent development has been the initiative of the European Union which has
set up a new committee under the European Union Crime Prevention Network. It is
presently carrying out an audit of what data is available on crime in EU countries and
is initially focussing on three crimes: street robbery, domestic burglary and car theft.

Gordon, himself draws on the available sources and, applying judgement based on his
own experience, publishes each year comparative data on 36 countries. To avoid
misleading, or ‘unsafe’ comparisons between countries, the emphasis is on changes
over time within countries. Inferences are thus restricted to whether crime is going up
or down more in one country than another. Reference: International Comparisons of
Criminal Justice Statistics. Home Office Statistical Bulletin 05/02.

To top