« Previous | Contents | Next »
Listen
Tracking Homelessness: A Feasibility Study
CHAPTER FIVE: TRACKING PEOPLE
5.1 A crucial element of any survey with a prospective element is the requirement to maintain contact with sample members and this would seem particularly challenging in the case of homeless people. In this section we consider several approaches to tracking homeless people, from using administrative data to maintaining contact with the sample members themselves. Depending on the data collection mode, it is likely that a combination of approaches would be required.
5.2 Where appropriate, evidence of the success (or failure) of using a particular method of maintaining a sample that was uncovered in the literature review (see Chapter 3) is included.
METHODS FOR MAINTAINING CONTACT
Sources of follow-up leads
5.3 For any surveys of homeless people, the most promising source of tracking information is the occasion of the first wave of data collection - particularly if this consists of a face-to-face interview. This interview allows the opportunity to collect contact details relating to a range of potential future sources of information - family, friends, agencies - and lifestyle information that may be predictive of future location. Once the purpose of the survey has been explained and co-operation gained, the respondents themselves are likely to be in the best position to identify ways in which the researcher/interviewer might be most likely to be able to locate them in the future. This suggests that a rather flexible (and hence more expensive) approach to the collection, up-dating and storage of contact information will be needed.
5.4 It is best to get as many stable contact points for homeless people as possible. Relatives, friends, social workers and agency workers are the main sources of follow-up leads for homeless people. Of these, family members and friends with stable addresses are the most productive. Fitzpatrick (2000) suggests that telephone numbers should be obtained as well as names and addresses wherever possible, as a telephone call is more likely to produce results than written correspondence. However, as telephone calls are more intrusive than letters, researchers should be sensitive when phoning a homeless person's family.
5.5 One study by Harocopos et al. (forthcoming) asked for the mobile phone numbers of a sample of crack users, but found that this was not particularly helpful as these young people sold or changed their mobiles so often. It is possible that this would also be the case for homeless people, but is certainly worth trying.
5.6 Professionals with whom homeless people are likely to have continuing contact, such as social workers, are good contact points and staff in agencies which they use periodically, such as day centres, can also be useful in contacting homeless people. Less helpful are agencies they tend to be in contact with for only a set period, such as hostels. Fitzpatrick (1998) found that using agency contacts can be problematic because of concerns about confidentiality and staff time being used to assist researchers. However, she suggests that it is best to arrange follow-up interviews through helping agencies first, before embarking on the more difficult, and potentially sensitive, road of pursuing homeless people's personal contacts. She recommends getting agencies on board from the outset in relation to the tracking exercise so that they expect the researcher to return and (one would hope) will be more willing to assist. It is best to attempt to persuade agencies to co-operate as a network so that they will alert the researcher if they come into contact with any of the sample of homeless people.
5.7 It is a good idea to get participants to sign 'permission to locate' forms as evidence of willingness to be tracked, but informants (particularly from services) will often still want to double check that the participants are still willing to be contacted. Both Harocopos et al. (forthcoming) and Craig et al. (1996) achieved a very high level of co-operation from a wide range of agencies (including prison and probation services) because they were able to produce written consent from their respondents.
Maximising success in pursuing leads
5.8 There were several recommendations from the homelessness studies reviewed for this project (Chapter 3) for maintaining contact with respondents. These are summarised in this section.
5.9 There is considerable evidence in a range of contexts that the proportion of panel members successfully traced at each wave of data collection tends to be higher the more frequently contacts are made. Harocopos et al. (forthcoming) in particular felt that the short period (one month) between the first and second interviews in their study was crucial in building up rapport.
5.10 Informal networking in the respondents' 'environmental niche', such as homelessness services or neighbourhood, has proven a valuable means of re-contacting respondents in a number of studies (e.g. Fitzpatrick, 2000; Crane, forthcoming; Harocopos et al., forthcoming). Maintaining a continuous presence in respondents' current or former environmental niche has sometimes been used to facilitate tracking, but this is likely to be expensive and could probably only be achieved in ethnographic studies and those with restricted geographical areas.
5.11 Most successful studies have used cash or voucher incentives, and have reported that these are crucial in securing the participation of some respondents. Cohen et al. (1993) used very low level incentives, and felt that a higher level might have motivated more participants (although they also note that ethical concerns could be raised by incentives that were high enough to be considered 'coercive'). Rossi (1991) suggests that such incentives could involve a sliding scale of payments that increase with each successive contact, but Harocopos et al. (forthcoming), who used this device, felt it actually had little impact on the retention rate achieved.
5.12 The most common method for encouraging homeless respondents to let the researchers know if they move is pre-paid postcards. These were used by O'Callaghan et al. (1996), but were found to be relatively unproductive. The use of incentives matched to notifications of changes of address has also had little success.
5.13 Wherever possible, the researcher who first interviews a homeless person should also track them for the duration of the project. This allows a one-to-one relationship and rapport to build up between the respondent and researcher, and tends to incentivise both to maintain contact.
5.14 Researchers tracking homeless people need to be creative, persistent and exhaust all leads (both Harocopos et al. (forthcoming) and Bassuk et al. (2001) managed to re-contact respondents they had lost in one wave of a follow-up in a later wave of interviews, thus demonstrating the value of persistence). They also need to be flexible, so they can respond immediately if a participant surfaces unexpectedly. Cohen et al. (1993) recommend that researchers should try to find out when homeless respondents have meetings scheduled with particular services, and be available to interview them at that time. Similarly, Crane (forthcoming) used knowledge of 'habits' of interviewees to time visits. Many studies report that telephone and face-to face contact is usually more effective than written correspondence, although some researchers have experienced success in sending out letters a week in advance of visits.
USING ADMINISTRATIVE DATA TO MAINTAIN A SAMPLE
5.15 Previously in this report it has been stated that it was probably not feasible to use purely administrative data to anwer research questions (see paragraph 4.39). However, this does not imply that administrative data would not be useful when undertaking other research methods. All additional information that can be obtained about panel members would assist when trying to trace them. Both O'Callaghan et al. (1996) and Craig et al. (1996) reported some success with checking administrative records for move-on addresses and other contact details.
5.16 Any administrative databases for which the Data Protection issues have been resolved could potentially be considered. This includes the main datasets related to homelessness described in Chapter 4, but any other administrative datasets maintained by homeless projects and agencies could also provide valuable information. Some datasets not related to homelessness could also add useful information. These include Benefit Records and Scottish Morbidity Records. Benefit Records contain a range of information and also a current address of any individual. Scottish Morbidity Records include information on all inpatient and outpatient hospital activity as well as deaths.
5.17 Matching to any of these datasets could help to build up a picture of the movements of any lost panel members over the period from which the last contact was made. This information could assist in helping to make physical re-contact with a panel member either through refining any search or by giving a recent contact address.
5.18 In theory, matching to these datasets could be achieved using a minimum of information. Just date of birth and full name would be sufficient, although more accurate matches would be achieved to some of the datasets if National Insurance numbers were available. Matching to more than one dataset would be fairly straightforward and, if feasible, it would be recommended to obtain as much information as possible.
5.19 Note that it is not recommended to just use administrative data to attempt to track a sample, but that administrative data should be used as an additional source of follow-up leads.
DATA PROTECTION AND ETHICAL ISSUES
The Data Protection Act 1998
5.20 In this section we outline relevant changes in the Data Protection Act and the impact these would have on undertaking a longitudinal study of homelessness. Please note that this is only an interpretation of the possible impact of the Data Protection Act 1998 - it may be advisable to obtain a legal opinion before undertaking the proposed study. (See wood.ccta.gov.uk/dpr/dpdoc.nsf for a copy of the 1998 Data Protection Act.)
5.21 The Data Protection Act 1998 was brought into force on 1 st March 2000 and replaced the 1984 Data Protection Act. The most relevant changes of this new act were to cover paper records as well as computerised data and to specify certain conditions that need to be satisfied before personal 6 and sensitive 7 personal data can be processed 8. In order for one organisation to pass personal data to another, consent would have to be obtained from those individuals. This consent needs to be expressly given, so cannot be assumed from non-response to a communication. To pass sensitive personal data, explicit consent needs to be given. This means that the subject needs to be given clear information about to whom their data will be passed, which data items will be disclosed and for what purpose.
5.22 Therefore consent would be required to both obtain samples from any administrative source and to track anyone through administrative data. At present no adequate consent is collected for any of the main datasets related to homelessness.
5.23 The consent required would need to be worded depending on the use to which the administrative data are to be put. If sensitive information were to be passed then explicit consent (as defined above) would need to be collected. Information about homelessness is not actually defined as sensitive by the Data Protection Act 1998 - therefore obtaining explicit consent would be more of an ethical consideration.
5.24 Consent would have to be collected for the different sample sources as follows:
- for samples selected from administrative data, consent would have to be obtained at the first point of contact with the organisation from whose administrative database the sample would be selected (e.g. for HL1 returns data, this would be when the application is made to the local authority);
- if following up samples from a survey, a consent question could be added to the survey. (This is common on interview surveys, e.g. the 1995 and 1998 Scottish Health Surveys to monitor hospital records.)
- for people at risk of homelessness released from institutions, permission would need to be obtained from these individuals prior to their release.
5.25 All of the above would require co-operation from the relevant agencies. In addition, the precise wording of the consent form would need to be agreed with the agencies from which the information is required in advance. (A generic consent form might not be acceptable to all organisations and would not be sufficient to obtain sensitive personal data.)
5.26 The impact of satisfying the Data Protection Act 1998 has already been observed from the question requesting permission to re-contact respondents to the SHS (see paragraph 4.12). The Data Protection Act 1998 requires that where it is known in advance that follow-up studies are planned or likely, respondents should be told of this. The question in the SHS has been worded so as to obtain explicit consent to pass on data collected in the SHS to other organisations and, most likely as a result of this wording, 28% of respondents did not agree to be re-contacted.
Ethical Issues
5.27 As well as satisfying the Data Protection Act, other ethical issues need to be considered. This is mainly to protect the individuals included in the study - especially where they are from vulnerable groups, such as homeless people. In addition though, a lack of ethical consideration will undermine the integrity of the research and also the organisations undertaking the study. The Social Research Association's (SRA) Ethical Guidelines (www.the-sra.org.uk) state that:
"Social researchers must strive to be aware of the intrusive potential of their work. They have no special entitlement to study all phenomena. The advancement of knowledge and the pursuit of Information are not themselves sufficient justifications for overriding other social and cultural values."
5.28 In this report, we do not consider all the wide range of ethical issues raised in the SRA guidelines. Instead, we focus on some of the main issues that would be raised by undertaking a panel study of a particularly vulnerable group.
5.29 For any study or survey, assurances of confidentiality are of the utmost importance. Respondents need to be absolutely convinced that their answers will not be passed on to anyone else - except where permission has been expressly obtained to pass the information to other organisations. In addition, informed consent should be obtained - it should be made clear to a respondent what the study is about and also the purpose of the research. In addition, it should be made clear that participation is voluntary and that there would be no detrimental consequences from not taking part in study.
5.30 The fieldwork should be designed so as to protect the interests of the respondents. For example, if the respondent is not happy participating in a particular environment, then an alternative venue for the interview should be sought. In addition, the same applies if the interviewer does not feel safe in a particular environment.
5.31 Running a panel study generates additional consideration of ethics - namely the impact of the research on the respondent. For example the effect of paying incentives to retain a panel. While a certain level of incentive is acceptable, large incentives might be inappropriate for a study of homelessness. These concerns were highlighted by Rossi (1991), who expressed concern about incentives that were high enough to be considered 'coercive'.
5.32 A further consideration is the amount of advice that interviewers should give respondents. It is not the role of the interviewers to offer advice to the respondents. If respondents ask for advice on any issues related to homelessness, they should be referred to agencies that can offer such advice such as the Citizens' Advice Bureau or Shelter. This policy should be made clear to the interviewers during the briefings. (Even this intervention might impact on the validity of the research as this could effect long-term outcomes, as the process of interviewing an individual might encourage some individuals to seek advice when they might not otherwise. However, it would be unacceptable to withhold details of agencies that could offer advice.)
5.33 These and other ethical issues would need to be tackled before a survey is undertaken, and researchers and interviewers made clear of the necessity to comply with any decisions reached. Consideration would be required when undertaking any study on homeless people or people at risk of homelessness that the process of following them is undertaken sensitively. It would certainly be highly unethical to undertake a study that would make the participants feel stigmatised or that would have a detrimental psychological impact.
« Previous | Contents | Next »