This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Victim surveys circumvent the problems of non-reporting and non-recording inherent in police recorded crime statistics. Broadly, victim surveys are concerned with the personal experiences of victimisation - in respect of property and violent offences - of those living in private households (Mayhew 2000). Victim surveys hold importance for both researchers and those who have been the victims of crime. Unsurprisingly many countries, in addition to the International Crime Victimization Survey, have their own national and sometimes even local victim surveys (Mayhew 2000). Given the sheer number of victim surveys it is difficult to account for the different methodological approaches that are adopted, and it is even more difficult to generalise the extent to which the research methods are vulnerable to bias. As such, the focus will be on a specific example of a victim survey, namely the British Crime Survey (BCS) and the extent to which the research methods of the BCS are vulnerable to gender, race and age biases.
The Home Office identifies bias as the effect that would arise if a survey design was implemented repeatedly and the "resulting estimates were systematically different from the true value" (Lynn and Elliot 2000). Where bias relating to gender, race and age is present one would expect an account of victimisation that is both unfair and unrepresentative of the "true value" of victimisation experienced by women, Black and Ethnic Minorities groups (BMEG), the young and the elderly. There are two BCS research methods that are particularly vulnerable to gender, race and age bias. Firstly, the BCS uses a very narrow sampling framework and secondly, the BCS relies on interview techniques that women, BEMG and the old may be less responsive to.
It needs however to be noted that victim surveys are not, and do not profess to be complete representations of the victimisation experienced by all sections of the population. They are intended to provide a generally representative picture of peoples' experiences of victimisation and perceptions of crime. Even research designs with the most carefully selected sample can be compromised by events outside the control of the interviewer (Lynn and Elliot 2000: 10). It is crucial therefore to be realistic in the analysis of the methodological underpinnings of the BCS and to accept a degree of bias. Bias in research data only becomes problematic where the extent of bias is unknown and secondly, the degree of bias is so substantial that it may compromise the validity of the data.
From the outset the BCS is not simply vulnerable to, but is actually biased against those under the age of sixteen as well as the very old. The BCS adopts the Small Users Postcode Address File (PAF) as its sampling framework capturing men and women aged sixteen and above living in private households. Those under the age of sixteen therefore have a zero probability that they will be selected for the survey and as such their experiences of victimisation cannot be accounted for. Since January 2009 attempts have been made to remedy youth bias by interviewing children aged between 10 and 15 years about their experiences of victimisation and perceptions of crime in the last 12 months (Millard and Flatly 2009). However, the research into youth victimisation is largely experimental and no steps have been taken to include the statistics in the routine BCS publication (Millard and Flatly 2009). The research methods of the BCS thus remain not only vulnerable to but unavoidably biased against people under the age of sixteen.
Similarly the BCS sampling framework excludes people living in institutions or communal establishments, people who are homeless, and businesses. The exclusion of institutions and communal establishments is likely to disproportionately affect the very elderly who are permanently resident in care homes. The vulnerability of the research methods to age bias is thus not confined to youth bias but includes bias against the very old. This is problematic as children and the elderly are already marginalised in society and their absence from the BCS compounds this. Moreover the experimental Home Office consultation revealed that 24% of people aged between 10 and 15 years had been victims of crime in the twelve months since January 2009 (Millard and Flatly 2009). As the BCS has not been extended to care homes it is difficult to discern the extent of bias in respect of the very old.
Nevertheless the BCS can account for the victimisation experienced by the majority age groups. The very elderly not resident in care homes constituted 7.6% of those who have been victims of crime in the last BCS (Flatly et al 2009: 57). The very elderly therefore constitute a minority group. Whilst this does not diminish the importance of the victimisation experienced by the very elderly, the extent of age bias is minimal. Moreover there is clear evidence that the Government intends to widen the sampling framework of the BCS to include children aged between 10 and 15 years (Millard and Flatly 2009). If the BCS is extended to those under sixteen, the vulnerability of the BCS research methods to age bias will be significantly reduced.
Interestingly the BCS sample framework is also vulnerable to racial bias despite the incorporation of "boosted samples". As Mayhew (2000: 107) notes, normal sampling methods will not produce sufficient numbers of Black and Ethnic Minorities for reliable analysis so "special sampling techniques need to be employed". Whilst boosted samples have provided a better insight into the experience of victimisation by BEMG, the BCS sampling framework remains vulnerable to race bias. The BCS explores only the effects of victimisation on the larger BEMG and fails to make nuanced cultural and historical distinctions between different ethnic groups. For example, the BCS includes people of African Caribbean origin and African origin under the generic title of 'black'. Failing to disaggregate the groups into their respected ethnic origins makes the methodology of the BCS vulnerable to race bias because the survey cannot capture the experiences of victimisation of individual ethnic groups such as African Caribbeans or Filipinos. Rather the BCS provides a general indication of the experience of victimisation of broadly composed BEMG.
One however must be pragmatic when considering the vulnerability of the BCS research methods to race bias. As Clancy et al (2000: 18) argue "in an ideal world the BCS would have a much larger booster sample, allowing greater disaggregation of the groupsâ€¦and coverage of the many smaller minority ethnic groups." Given the size of BEMG and their geographical distribution there is simply no cost-effective way of accommodating a larger booster sample (Clancy et al 2000). More importantly the current break down of BEMG does allow researchers to identify important trends in the victimisation of non-Whites. For example, the overall risk of victimisation was found to be highest amongst people from mixed ethnic groups (Home Office, 2005). Although the exact ethnic origins of the people whom comprise the mixed ethnic group are unknown, this does not prevent policy makers from implementing strategies to reduce the risk of victimisation for those people who broadly fall into the mixed ethnic category. In essence the BCS sampling framework is vulnerable to racial bias however this does not compromise its validity because the bias is not only known, but the extent of the bias is minimal allowing researchers and policy-makers alike to use the findings constructively.
In addition to the age and race biases that may arise from an unrepresentative sampling framework, the BCS's use of face-to-face interviews may generate further bias. Women whom have been victims of sexual victimisation or domestic violence may be reluctant to relay their personal experiences to an interviewer who is essentially a stranger; a figure they may be disinclined to trust (Smith 1994). Moreover the perpetrator may be present in the household at the time of the interview and so the female respondent may be unwilling to talk about her experiences perhaps "for fear of reprisal by her abuser" (Smith 1994). In refusing to communicate experiences of sexual victimisation and violence to the interviewer, the portrayal of females as victims of crime cannot be anything but biased against the "true value" of victimisation. Put crudely, the researcher has very limited, and in some cases no data, to interpret and to present.
The extent of gender bias may be minimised by the use of Computer Assisted Self-Interviewing (CASI). CASI is used to obtain information about sensitive topics such as the "use of illicit drugs, andâ€¦experiences of domestic violence, sexual attacks or 'stalking'" (Maguire 2008: 268). CASI reduces the possibility of bias arising from non-disclosure by affording the respondent a greater sense of privacy because the survey is completed electronically and is not seen by the interviewer. As the female respondent is in complete control and away from the gaze of the interviewer and the abuser who may overhear, she may feel more confident answering questions relating to sexual victimisation and domestic violence. Whilst CASI will not reveal the "true value" of female victimisation, it does provide an important insight into an otherwise unknown area. In this way the researcher is made aware of the prevalence of gender bias in the BCS and thus can validate the research by accounting for bias as well as flagging up an area requiring further research.
Most problematically race, age and gender bias may stem from the way in which the survey questions are worded. Mooney (2000) offers a very clear demonstration of the gender, race and age biases arising from the variation in women's definitions of domestic violence. Mooney (2000:156-57) observes that "African-Caribbean womenâ€¦are most likely to define all behaviours presented to them as domestic violence" where as women aged 55 and over "are less likely than other [age] groups to define domestic behaviours, including physical violence" as domestic violence. In the absence of a consensus definition, women will interpret a question relating to domestic violence consistent with their own understanding of what is means to be abused by their partner. Older white women may, for example, withhold experiences of victimisation that younger black women may include. The disparity between accounts means that the research findings will be biased; such that they are unfair and unrepresentative of the "true value" of victimisation experienced by women of different age groups and ethnic origins.
Whilst however the wording of the BCS questions may be vulnerable to gender and age bias, the bias can be controlled for. In the North London Domestic Violence Survey Mooney (2000) employs a consensus definition to minimise discrepancies and thus bias. As Hough (1986: 45) notes "the use of a consensus definition enables the argument that violence rates areâ€¦a reflection of definitional variations between different parts of the population to be countered." Again, when considering the extent to which research methods are vulnerable to gender, race and age bias one must be minded to the extent of bias and more importantly, what measures can be used to limit the minimise bias.
The BCS is vulnerable to gender, race and age bias. The BCS findings deviate from the "true value" of victimisation largely because the BCS employs a very narrow sampling framework which excludes those under sixteen and the very old as well as using BEMG which are in themselves very culturally diverse. Moreover, face-to-face interview techniques and the restrictive wording of BCS survey questions are unable to capture fully the experiences of victimisation. Nevertheless the BCS is still able to provide a general overview of the experience of victimisation within Britain in the last twelve months. The vulnerability of the BCS to gender, race and age bias is well-known and the methodology can be modified to minimise bias. Furthermore, as exemplified by Left-Realists and Feminist academics alike the under-representation of certain groups gives rise to the opportunity for 'bespoke' surveys (Mayhew, 2000: 97). These surveys can use a more representative sample and formulate relevant questions which are arguably more useful in the development of policy aiming to address problems of victimisation.