The Islington Crime Survey 1986

The Islington crime survey (ICS) was conducted over twenty years ago. There have been many such local victimisation surveys but this was one of the first and it had a big political impact on the way we now think about things such as fear of crime, the impact of crime on women and ethnic minorities and on police-community relations in poor communities. It remains a key example of how social research is driven by both theoretical and political concerns

The theoretical and political context

People want to do social research because it helps answer theoretical questions, because it helps to provide material and data which can be used in political or policy debates, or both. The ICS was certainly not conducted in a vacuum. Let me give you a little background to the ICS

The mid 1980s were the high point of the approach in criminology which has come to be known as Left Realism. In 1984 Jock Young and I had published What Is To Be Done About Law and Order ? which set out basic Left Realist postion. This book was being widely read by Labour politicians of the time and by forward thinking police officers. For our purposes here, two key elements of the Left Realist approach can be stressed

• the poor, women and ethnic minorities, suffer from high levels of victimisation. Politically, the Left had tended to shy away from these issues and had left the field clear for the Right. We were determined to get crime back on a Left political agenda

• the police had a reputation of being oppressive and out of touch with key communities, especially the poor and ethnic minorities. We argued that the way to bring the police close to communities was to increase the level of democratic control by local communities over the police. This, we argued, was crucially linked to police efficiency. The Tory government at the time (under Margaret Thatcher) was arguing that the police needed all sorts of 'performance indicators.' and targets to make them efficient.

We argued that the key thing was trust: if the community trusted the police then information about crime would flow from the community to the police. Information is the lifeblood of solving crime. An important part of the basis of that trust is that the police are accountable to the needs of the community as regards crime:  that the police prioritise the same sorts of crime as the community. Of course, many police officers responded to such arguments by claiming that:

• 'our job is to enforce the law.' Quite true, but there are too many laws and too much crime to proceed against every infringement of every law. So the police as an organisation needs to prioritise the distribution of its resources by place and by concentrating on certain types of crime and maybe less on others. Various sections of the community will have their priorities - shop keepers or bar owners in the central city will want plenty of police around at night; women walking home from the tube late at night would like to see more officers on the streets at that time etc. 

• 'we know what the public want: they phone us up.' But remember there may be sections of the public who don't phone the police when they become victims of crime: women who are victims of domestic violence or rape may worry about whether the police will take them seriously or whether they will face reprisals from their attacker. Also many crimes reported to the police are for insurance purposes. Thus the police may feel that the 'main problems in this area are car crime and burglary. These may be high profile because of insurance. Other crimes such as racial attacks or domestic violence may feature much lower as regards phone calls to the police but be felt more important by the community.

So as far as Left Realism was concerned police accountability to community needs means efficiency. That indeed is the definition of an efficient police force: one that is responsive to community needs. If the measure of efficiency is simply how many arrests are made this may in fact alienate the community if, for example, many arrests are of young people involved in minor crimes (such as possession of soft drugs)

So those of use at Middlesex Polytechnic (as it then was) who were to carry out the survey, had a number of key questions we wanted to answer. What were the levels of victimisation in poor communities, particularly for women and ethnic minorities? How alienated were poor communities from the police? Did these communities want more accountability of the police?

The funding organisation

Who pays for your research? The funding organisation may have its own priorities and one condition of your funding may be that your research addresses some of your funder's concerns. The organisation that funded the ICS was the London Borough of Islington. They had their agenda. Most of the questions they wanted answered were the same as ours. But there was a more general political agenda. At that time, although the Tories were in government, Labour was very strong in local government. Islington was a Labour held Borough that, like us, wanted a strong orientation to crime but one which was different from the Tories. 

central government

The Tories were not uncritical of the police. Unlike previous Tory governments which tended to view the police as rather like the Monarchy -- as untouchable -- the Tories under Thatcher saw the police as one of a number of bloated inefficient bureaucracies. They, the police, needed a dose of business management from the private sector and some performance indicators to measure their efficiency.

One simple way to measure police efficiency was, as noted above, in terms of numbers of arrests and clear up rates. But of course, as every criminologist knows, the police themselves have considerable control over these figures. You can always arrest more people to meet your targets and you can concentrate on those crimes which are more easily solved in order to increase your clear-up rate. The result is not necessarily, as argued above, in accordance with public need.

One solution is to have a measure of crime which is independent of police activity. Social surveys provide just such a method. Social surveys of crime and victimisation were, by the mid 1980s, widely used in North America. They were seen as a good way of getting at the number of crimes not reported to the police. Police might report incidents to social survey interviewers that they had not bothered to report to the police, possibly because they did not think the police would bother to do anything or they would themselves be the subject of police harassment when reporting the crime.

So the Tory government had passed the 1984 Police and Criminal Evidence Act which, among other things provided for the establishment of local police-community Liaison Committees. These have now, following the 1998 Crime and Disorder Act, been absorbed into the Crime and Disorder Reduction Partnerships.

The British Crime Survey

The government also sponsored the British Crime Survey (BCS). This was a national victimisation survey which began in 1981. It made criminal victimisaion surveys 'respectable' but methodologically had, at least in its early versions, a number of limitations:

• as a national survey it had a very small sample (10,000 basic sample) and the first sweep of the survey achieved notoriety for only picking up one rape

• some commentators came to make ludicrous generalisations about crime on the basis of the results. For example it could be calculated that the 'average person' statistically stands a chance of suffereing a street robbery every 500 years, an assault every hundred years, and a burglary every 40 years. In this context much reported fear of crime appeared an irrational effect of media moral panics rather than real fears based on experience. If the public were worried about crime, they were just wrong! 

Critics of the BCS, including in particular Left Realists argued that, on the contrary, fear of crime by poor communities and in particular women was indeed rational and this would be better illustrated by local victimisation surveys which would pick up the focused nation of crime in poor communities. Generalisations such as 'the average citizen faces an assault once every 100 years' are completely meaningless. Some people may have next to no likelihood of victimisation, others are very likely to become victims. These contrasts may disappear when averages are constructed. The average is a purely statistical construct. It may not correspond to the actual experience of anyone at all! 

Social researchers are well aware of such phenomena as the ecological fallacy whereby the attributes of two populations with contrasting characteristics cancel out one another. In a London Borough such as Islington in the mid-1980s poor communities often lived quite close more affluent middle class areas. 

Above all the notion that fear of crime was 'irrational' expressed an elitist contempt for the public and functioned, in effect, as a cover for police inefficiency.

Interest groups involved

Apart from the general political background, which is very important for a piece of research such as ICS, there is of course a local background. A survey sponsored by a local Borough Council on a controversial topic such as crime and policing is not just another item of market research, it is going to attract attention from local media and interest groups. So you have to put in a bit of groundwork to persuade the local communities, the local police and the press that it is in their interest that such a survey goes ahead. In that way youy help to increase the response rate. If people know about the survey and approve of it, then when interviewers come knocking at the door, they will not be sent away.

Significant groups in the locality, besides the Council itself, were

• Tenants and community groups. Many local people felt isolated and needed a voice for talking both to the council and to the police. There already existing of course local Councillor's surgeries and also the police-community liaison committee. These groups could be helped by the information provided by a local survey. Women and ethnic minority organisations  groups and ethnic minority groups often felt their specific needs as regards criminal victimisation (domestic violence, sexual assault, racial attacks etc., fear of the streets at night) were not being heard strongly enough by either the Council or the Police. The possession of information is an important resource in negotiation and bargaining. If a group which claims the police are not taking its problems seriously enough comes armed with information that demonstrates this, the group's concerns are harder to dismiss as 'irrational fear of crime' So local meetings were organised at which the purposes of the survey and the extent to which questions would elicit useful information, were debated.

• Police. Some local commanders were in favour of the survey because they reasoned it could be a useful source of information and if they could be seen to be responding to local needs then this would improve the state of police-community relations. Also it would help relations between the police and the local Council as both bodies could discuss how best to improve particular areas: for example by having more police officers on patrol, better street lighting, or some combination of both. The Metropolitan Police as an organisation was sufficiently concerned about its general standing with the public to have commissioned a major London-wide survey of police-public relations. This, known as Police and People in London by David Smith and others was undertaken by an independent research institution, the Policy Studies Institute (PSI) and published in 1983. The report showed high levels of dissatisfaction with police racism, and alienation from police particularly of young Afro-Carribeans. Thus Left Realists were not the only group arguing that the police were out of touch with many of the communities they policed. The local police in Islington reflected some of these concerns though, by and large, most officers probably felt that new what the public wanted and were achieving it to the best of their ability. While some far-sighted officers saw that sharing and discussing policing strategy with the public and debating with criminologists was a step forward, and inevitable in the future, many still saw themselves as having a monopoly of wisdom when it came to matters of crime.

• The Local Authority, the Council of the London Borough of Islington was very interested in funding a local victimisation survey. As already noted, their interest were not at all in conflict with ours. The particular advantages for the Council were that the survey could be used toshow public that they, the Labour party, who were then in office in the were serious about crime. Also the survey could  strengthen position of Islington in relation to the police. The Borough would now have its own independent database of crime and public perceptions of crime and policing. This could be used for local negotiations in the Liasison Committee to pressure the police to meet public needs.

Doing the Survey

the questionnaire

All the things discussed so far will have an influence on the construction of the survey questionnaire itself.

Your sponsors may want particular technical questions included (e.g. the Council may want to know which particular type of street lighting is best for making areas appear safe etc.) Community groups will want particular questions included. You have to strike a balance between these demands and your own knowledge of what sort of information about crime and policing is necessary to make the survey useful.

You need your data to be comparable to that from other surveys. Something like the ICS was going to be quite high profile. People were going to start quoting the results and how they contrasted with other local surveys such as the PSI London survey and also with the BCS national survey. Data compatibility is an important consideration. You have to strike a balance between including 'other people's questions' in your survey even if you think you know a better way to phrase the question. For example if the PSI measured alienation of the public from the police in terms of categories like 'willingness to give information' 'willingness to appear as a witness in court'  'willingness to report an incident to the police' and then tabulate by age, gender and ethnicity, then there is considerable pressure on you to include those questions, worded in the same way, in your survey so you can make comparisons.  

To give another example: the rationality of the fear of crime. Most crime surveys asked how worried about crime you are and whether this influences your behaviour (e.g. avoiding going out after dark. This is then correlated with whether you have been a victim of crime. Now the BCS and similar surveys often asked only whether you had been a victim of crime during the previous twelve months. But the ICS was interested in a much longer period. Obviously we could compare our results with the other surveys for the behaviour of people who had been victims of crime over the previous twelve months but there could be no comparative data for those people in our survey who had been a victim of crime over, say, the previous three years. The reason for this longer period in the BCS will be discussed below.

There is also another basic methodological decision to be made:  are you interested simply in descriptive statistics (e.g. what percentage of a particular group fears what form of victimization) or do you want to do multi-variate analysis; degrees of correlation between things like neighbourhood structure, age and say, fear of crime e.g. what is the correlation between age and fear of crime. In the ICS we were interested in the latter. This, as you will know from research methods classes, influences the type and precision of the questions you need to ask. You need responses that can be ranked in some way. Age is obviously rankable if you ask respondents their actual age, or ask them to choose an age range which is no more than, say five years. This gives us a range of age groups which we can then correlate with fear of crime as measured by qualifiers such as 'very', 'somewhat', 'a little','not at all'. You won't get as good a result in terms of correlation if you just ask people to answer 'yes or no' to a question on fear of crime. You won't be able to find out if fear increases in intensity with age or with some other variable or a combination. 

Above all, the questionnaire must not be too long otherwise respondents will grow bored. We used the technique of having a basic questionnaire and then if the respondent had been a victim of crime, they were asked to complete a subsequent interview.

sampling

The ICS sample size was about 2,000 with response rate of 60-70% This included booster samples for ethnic groups. It is very important in a survey of this type to be able to say things with confidence about all sections of the local population. Even with a large initial sample, if it is a random sample of the general population of the locality,  then when you get down to, for example, 'young non-while males aged under 21 who have had at least two contacts with the police over the previous twelve months' you will be down to very few people. Anyone can trot out percentages but if n=5 then there is no great reliability. 

For this reason, as is usual in such situations, we included booster samples for particular sections of the population. This enables you to have a larger number of cases for particular groups of people such as in the example above but it also distorts you total sample. You can no longer make any statements like '60 percent of the people of Islington think...' Because your sample now over-represents certain categories of the population. So you have to weight you samples statistically so that the profile of the population as a whole is reasonably correct. How do you know what this profile is: e.g. what percentage of the population are White, Black, Asian etc.? You have to go to the last official Census data and weight your sample so that it reflects the official government Census enumeration. Fortunately for us the last official census had been in 1981 so the data was reasonably up to date. But, the census only recorded country of birth so it didn't really give an accurate picture of the makeup of enumeration districts. I won't go into any more techicalities as I'm sure you have covered this in your research methods courses.

fieldwork

The interviews were done during March-August 1985. We employed our own interviewers, 58 of them. We used local students mainly. If you use non-professional interviewers, who have not necessarily any familiarity with this type of work (professionals from a market research company would have been prohibitively expensive and not necessarily an advantage in other respects) then you have to train them and monitor them. We had a survey manager to organise training, and also to conduct monitoring. This involved visiting a sample of households that had been recorded as interviewed and checking that they had in fact been interviewed.

Results - what the survey showed

The findings are now history and they have been replicated many times by other surveys. Briefly

attitudes to crime and police

• 70% saw crime as a problem second only to unemployment

• fear of crime clearly corrrelated to risk of victimisation

• 50% of women saw walking the streets as a problem. Indeed the survey concluded that there was a virtual curfew for women in many parts of the Borough during the hours of darkness

• 60% saw street robbery as increasing

• 50% households had suffered repeat victimisation

all these were tabulated by age, race and gender

priorities for what the the public wanted police to concentrate on

• 70% prioritised street robbery

• 70% prioritised sexual assault

• 60% drugs

• 60% burglary

this differed from police perceptions which, for example included things like vehicle crime which have a high rate of reporting to the police because of insurance requirements but didn't score that high as far as the public are concerned.

public perceptions of things police spent too much time on

these included things like prostitution 30%, cannabis 25%. Also important was the correlation between likelihood of victimisation and likelihood of seeing police as acting in unsatisfactory ways. This was also true for issues like 'police understanding of the area'. The more contact people had with the police, the more likely they were to feel the police didn't understand the community they were policing.

Finally significant was the remedy for breakdown in police-community relations: 55% wanted a Police Authority for London. It was 1999 before the establishment of the Metropolitan Police Authority in which elected civilian members have a measure of oversight over general policing policy and efficiency.

comparisons with other surveys

We could compare the ICS with the PSI survey for London as a whole. For example 40% of ICS respondents were dissatisfied with police stop and search (only 19 for PSI).

There was a particular conflict with BCS was over the issue of fear of crime. The BCS in their 1981 sweep found little correlation between fear of crime and actual victimisation. e.g. older women had more fear but less likelihood of victimisation, younger women less fear though greater likelihood of victimisaton. This fuelled ideas that the public, or women in particular, were 'irrational'. 

Firstly the ICS as a local survey focused in a poor areas found much higher levels of victimisation for all women in Islington than for the general population and indeed around one third of all women avoided going out after dark. Secondly we pointed out that the BCS assumed fear would last a year and then be forgotten! In other words it only correlated fear with victimisation incidents over the previous twelve months prior to the survey. The ICS observed that older women remembered their victimisation even though it was less now they were older. But such memories would influence their behaviour. Younger women, by contrast, though a greater chance of victimisation,  would have less lifetime memory of victimisation and so be less worried. No-one was being irrational at all

Results - policy implications

In such a political situation the results are largely out of your hands. You can hold your own press conference and announce the results, but the various interested parties will have better publicity machines than you, and they will be unequal among themselves

As far as we were concerned, some of the main claims of Left Realism had been vindicated. We had established that poor areas like Islington suffered high levels of victimisation and in particular that women's concerns about sexual assault were not irrational. We had also corroborated our perceptions of widespread alientation between public and police and the existence among the public of a need for greater democratic input to policing. 

The particular interests in the locality responded to the survey in various ways. Different Council departments for example saw the results as  establishing the need for more expenditure in their particular areas. Highways argued the survey showed the need for better street lighting and estate improvement, Youth Services used the results to establish they need for more investment in youth projects and community services. These sorts of issues could only be sorted out by the Council's internal budgeting processes.

As regards the police, while some responded negatively and felt 'we knew all that anyway' more far sighted local police managers understood that the ICS had highlighted the conflict between their perceptions of the crime and the public's perceptions and that therefore a more sophisticated process of dialogue and planning was required. Years later, during the 1990s, these issues would find their way into new Community Crime Reduction Partnerships which have proliferated. Whether these have made any different to levels of victimisation or to police community relations I leave you to judge.

Finally the ICS certainly had an influence on subsequent surveys, including the further sweeps of the BCS itself, especially in areas such as the fefar of crime.

bibliography

Koffman, L. (1996) Crime Surveys and Victims of Crime. Cardiff: University of Wales Press
Anderson, S., Grove Smith, C., Kinsey, R. and Wood, J. (1990) The Edinburgh Crime Survey: First Report. Edinburgh, Scottish Office (Central Research Unit Papers).
Jones, T., Maclean, B. and Young, T. (1986) The Islington Crime Survey. Aldershot: Gower.
Kinsey, R. (1984) First Report of the Merseyside Crime Survey. Liverpool: Merseyside County Council.
Lea, J. and Young, J. (1984) What is to be Done About Law and Order? (2nd edition) London: Pluto Press.
Kinsey, R. Lea, J. Young, J (1986) Losing the Fight Against Crime. Oxford: Blackwell
Pantazis, C. and Gordon, D. (1998) Do the poor experience more crime and greater fear than the rich?, In Dorling, D. and Simpson, L. (eds.) Statistics in Society, London: Arnold.
Smith, D. et al. (1984) Police and People in London. London: Policy Studies Institute