British Crime Survey (BCS)
SUBJECTS ON THIS PAGE:
BCS has two main purposes. First, it is the main source for information about people’s perceptions of crime, how much they are worried about crime, etc. This is the so-called ‘non-victims’ part of BCS.
Second, it is the main source of information about the prevalence of crime (the ‘victims’ part of BCS). This is an important purpose because analyse of it tends to result in very different answers than analyses of police-recorded crime statistics, largely because the latter have been affected by changes in recording practice from year to year. However, for reasons discussed below, it is difficult for external researchers to use the dataset for this purpose and it is therefore recommended that external researchers generally restrict themselves to what is available in the Home Office’s annual report (entitled Crime in England and Wales).
Note that, although called the British Crime Survey, it does not cover Scotland. Also note that there are equivalent but less frequent surveys in Scotland and Northern Ireland.
- Available from: UK data archive.
- Registration required: yes.
- First survey available: 1982.
- Frequency: annual.
- Updated: December.
- Scope: England and Wales only, even though it is called the British Crime Survey.
- Format: SPSS, STATA or TAB.
- Files: 2 files per year, one (non-victims) relating to the whole of the population’s views on crime and the other (victims) relating to actual crimes.
- Documentation: questionnaire only.
- Weighted or unweighted: weighted.
- Household income data: yes but unequivalised only.
- The non-victims file is a household-level file but it also contains some information about a ‘random adult’ in that household. As such, it can be used to analyse information about either households or individuals.
- The victims file has a record per crime for each of the households in the non-victim file. So, for example: if the household had been subject to three crimes then there would be three records for that household in the victims file; if, however, the household had been subject to no crimes than there would be no records for that household in the victims file.
In addition to the dataset itself, the Home Office publishes an annual report entitled Crime in England and Wales giving many statistical analyses from the dataset.
Whereas the non-victims file is a standard dataset, the victims file is very difficult to analyse without making mistakes. The reason for this is that, whilst the file in principle contains a record for each crime that has occurred over the previous twelve months, this is not a consistent time frame because the dates when interviews were undertaken varies from household to household. Further complications arise because some crimes are not single incidents, others do not have a clear date, and yet others took place abroad. There are a whole variety of fields which can be used to manage these complications but the lack of documentation means that this cannot be done by external researchers unless they have previously talked to the Home Office people responsible for the dataset.
The complications above also mean that, whilst the two files can in principle be linked together using the case identifier, it is difficult for such linkages to contain the appropriate information from the victims file. So, for example, it is difficult for an external researcher to analyse the proportion of households who have been burgled over the previous twelve months.
In order to produce the graphs on this website, we have, in discussions with the Home Office, worked through these complications. But, unless you have both substantial time available and a strong need, it is not recommended that you try and do so. Rather, it is suggested that you restrict yourselves to analyses regarding peoples’ views on crime (which do not require any data from the victims file) rather than the actual prevalence of crime (which does require data from the victims file). If analyses of prevalence of crime are required, it is suggested that you restrict yourselves to what is available in the Home Office’s annual report.
Which software to use
As the annual dataset is around 48,000 records for non-victims and 18,000 records for victims, it can be exported into Excel.
When to use the victims and non-victims datasets
As discussed above, it is suggested that external researchers restrict themselves to the non-victims dataset.
What weights to use
There are two weights in the non-victims dataset: a household weight and an individual weight for the random adults who were interviewed. Use the household weight when the variables to be analysed relate to households (e.g. whether or not they have home contents) and the individual weight when the variables relate to individuals (e.g. perceptions of crime).
Relevant graphs on this website
Because the dataset is for England and Wales only, the graphs below are also for England and Wales only.
|Problem drug use||all||n/a||Data actually obtained from the Home Office annual report rather than from the dataset itself.|
|Anxiety||all||non-victims||Use the individual weight and exclude those with a null weight.|
|Without home contents insurance||first||victims||Use data from the Home Office annual report for the total number of burglaries.
Use the Living Costs and Food Survey to estimate the proportion of households which have home contents insurance
Use the dataset to estimate the proportion of burglaries where the household had home contents insurance. Use the household weight. Requires a complicated set of decision rules to decide which victims to include/exclude and how many burglaries they have been subject to.
|Victims of crime||first to fourth||n/a||Data actually obtained from the Home Office annual report rather than from the dataset itself.|
|fifth||non-victims||Construct a ‘worry about violence’ variable using the rules set out in the glossary of he Home Office annual report.
Use the individual weight and exclude those with a null weight.
|sixth||n/a||Data actually obtained via a request to the Home Office as the requirement is similar to, but not the same as, the data available in their annual report.|