Survey example 2: Public attitudes to sex offenders in the community

Our second example of sensitive survey research details a project we conducted in 2004/5 aimed at revealing public perceptions of sexual offences and attitudes towards the sentencing and potential reintegration of those convicted (Brown et al., 2007). Two distinct problems were raised by this research. These were:

1. How to approach the emotive topic of sex offender reintegration sensitively.

2. Ensuring that the sample of the general public is representative, and includes some respondents from hard-to-reach groups.

Just under 1, 000 responses were gathered using a range of techniques, including a targeted postal questionnaire, a web-based questionnaire promoted via various Usenet newsgroups, and a community-based questionnaire delivered to community groups, local meetings, health services and local businesses (Brown et al., 2007).

Previous research has shown that gauging public attitudes to offenders and sen- tencing, in general, is fraught with problems (Stalans, 2002). Results have been varied (often contradicting the common perception that people are generally puni- tive). Stalans argues that this variance is caused by methodological issues: ‘researchers have measured attitudes in a variety of ways and the public’s response often depends on the manner in which the question is asked and the context in which it is located’ (Stalans, 2002: 16). Stalans (2002) goes on to argue that research considering public attitudes to sentencing tends to be relatively superficial without fully exploring the depth of public opinion. Clearly, attempting to unravel the layers of attitudes to offenders, sentencing and their management in the community is no easy task.

In our study of public attitudes towards sex offenders, the task was further com- plicated by the highly sensitive and diverse nature of offences that have been termed ‘sex crime’ and the emotive response they provoke. Our decision to conduct a survey of the general public about this sensitive and controversial topic resulted in a range of challenges to be met and problems to be overcome.

Sample selection and data gathering

Our first problem was in ensuring that we constructed a representative sample. The sample needed to include people from different social, demographic, ethnic and geographic groupings, and include those that could be considered ‘hard to reach’ (see Box 9. 2). In order to take account of these differences within the sample, we used several different sampling methods.

First, using social and demographic research data accessed via the Office of National Statistics we identified a range of socio-economic locations. We were then able to select postal areas from within our larger, targeted, geographic area. We included two areas that had high levels of deprivation, two areas with ‘average’ levels of economic activity and one area that was ‘affluent’; we also included one area that was defined as multicultural. Two of the areas identified were rural areas and the remaining geographic locations were close to city centres. Once we had decided on the geographic locations, we began to negotiate access with local com- munity-based groups and health centres; these were across a spectrum of pursuits and concerns so that we could ensure variations of age and interests. We were avail- able to explain the purpose of the questionnaire, to be on hand while respondents completed the questionnaire and to administer the questionnaire verbally or trans- late it if necessary. Through this method, we were able to include some respondents within the ‘hard to reach’ category, and we were able to offer advice and support to this group. However, this is a very time-consuming method of data gathering. Spending time at each of the community groups was a highly labour-intensive means

of data gathering and, at times, the questionnaire return was small due to the num- ber of people attending the meetings. We collected 98 questionnaires from community groups and health centres.

At the same time, we mailed out a total of 5, 000 questionnaires. The question- naires were split across our geographic areas to named respondents, which was an attempt to embed an element of control in the structure of the sample. We also mailed the questionnaire to large workplaces. In total, we received a 15% response rate to our postal questionnaire: an average response for an unsolicited postal questionnaire.

Additionally, we placed a web-based questionnaire on a newsgroup site which consisted of mainly open questions. This was intended to supplement the data gathered through the other, more targeted, sampling methods. The problem with the web-based method is that while you can gain rich data you have no control over the structure of the sample (see also Chapter 18 by Yar in this volume). The very nature of the respondents using newsgroups meant that the data could have been skewed towards those with a greater interest in, and knowledge of, current or topical issues. Additionally, people likely to respond to web-linked questionnaires are often those with the most extreme views. There were 72 web questionnaires returned, which we used separately from the postal and community responses as additional, illustrative data to support (and sometimes confound) the rest of our results (Brown et al., 2007).

Making sense of the data

In total, we had 979 responses to our questionnaire. We had attempted to elicit a range and depth of information from respondents and so had used a number of sur- vey data-gathering methods; for example, we included a range of open questions and vignettes to create space for attitudes to be expressed and definitions to be explored (see Box 9. 6). We asked a number of questions in relation to definitions of sex crime, attitudes towards sex offender sentencing and rehabilitation, views on the effective- ness of current resettlement strategies and the sex offenders register. All respondents were provided with an information sheet about the project that included the contact details of relevant help and support organizations (for a more detailed report of this research, see Brown et al., 2007).

The data generated from the various survey methods was rich and allowed us to analyse attitudes to sex offenders against the variables of gender, age, socio-economic status and type of area of residence, for example urban or rural. The responses were varied and in sufficient volume to enable us to complete a range of statistical tests on the data. In general, our results were consistent with those of previous research stud- ies and highlighted that attitudes towards sex offenders were much more nuanced and complex than the claims made by the tabloid press. Our survey method had effectively uncovered these complexities through a combination of vignettes and open and closed questions directed at revealing attitudes.

The research highlighted the problem that policy-makers face in continually hav- ing to reproduce high levels of trust in the criminal justice system, in order to ensure that the public retain confidence not only in the delivery of justice but also in the management of offenders in the community (Brown et al., 2007). Our research sug- gests that the level of public trust in the criminal justice system and in criminal justice professionals is essentially very fragile. We were able to discover the fragility of pub- lic trust through the use of a survey in what is a very sensitive area.