3.2 Data Collection, Sample Selection and Variables
3.2.1 Self-reported or official data, and data collection
Prior to identifying the most appropriate sample and method by which to collect data, existing literature was reviewed to determine the techniques previously employed by researchers. Although a number of studies have used self-report data (e.g., Dhami, Ayton & Lowenstein, 2007; Gover, Perez & Jennings, 2008; Heubner, 2003; Lee & Edens, 2005; Varano, Heubner & Bynum, 2011; Zamble, 1992; Zamble & Porporino, 1990), the majority of studies were found to have used official data (e.g., Arbach-Lucioni,
Martinez-Garcia, Andres-Pueyo, 2012; Craddock, 1996; Cunningham, Sorensen, Vigen & Woods, 2010; Edens, Poythress, Lilienfeld & Patrick, 2008; Griffin, & Hepburn, 2013; Innes, 1997; Ruback & Carr, 1993; Steiner, 2009).
Self-report methods of data collection have a number of benefits, including that they allow participants to describe their own experiences, and can examine a large number of variables (Junger-Tas & Marshall, 1999). However, self-report data has its limitations. For example, prisoners may exaggerate their involvement in prison
offending to bolster their bravado with researchers or peers, or may underreport their involvement in prison offending for fear of repercussions from management (Steiner & Wooldredge, 2012). In addition, prisoners’ interpretation of what is considered a prison offence according to the Prisons Act 1981 may differ from the official interpretation of
the legislation. This may result in underreporting of potential offences. This could be particularly the case in regard to some minor prison offences which require a degree of construal by prison officers, such as “behaves in a disorderly manner”, “swears or uses indecent language” or “does any act or omission of insubordination or misconduct subversive of the order and good government of the prison” (Prisons Act 1981). For example, verbal abuse by a prisoner towards another prisoner is arguably an offence according to the legislation. However, the prisoner who verbally abused another may think of it as a meaningless act, rather than an offence.
Alternatively, the use of official data, or data collected by agencies for official purposes, has been noted to have benefits. The use of official data allows researchers to access large samples as well as representative samples, and trends over time are
possible (Junger-Tas & Marshall, 1999). However, the use of official data has been criticised by some scholars as possibly underestimating the total number of incidents of offending within prisons, due to some offences going undetected or indeed unreported (Lee & Edens, 2005; Steiner, 2009). Many reasons for offences going undetected, including crowding or understaffing, both of which may affect the level of direct
supervision of prisoners by prison officers, which may result lower figures represented in official data (Hewitt, Poole & Regoli, 1984; Poole & Regoli, 1980; Steiner & Wooldredge, 2012; Van Voorhis, 1994). Similarly, studies have revealed that official prison offending data can be influenced by officers’ use of discretion, or their inconsistent application of procedures, which may result in some, often minor, prison offences going unreported (Cunningham, Reidy & Sorensen, 2008; Poole and Regoli, 1980). Based on prior
literature, both official misconduct data and self-report measures have been determined to be valid indicators of the rates of actual prison misconduct (Simon, 1993; Van Voorhis, 1994).
Prior to determining the method by which data was collected, consideration was given to a study period, and possible sample size. Firstly, in order to ensure that each prisoner had the same opportunities for offending as others within the sample by virtue of the length of their imprisonment (DeLisi, Berg, & Hochstetler, 2004), and to enable the recording of situational variables across the seasons and months of the year
Remster, 2012; Freeman, 2003; Lahm, 2008; Lahm, 2009; Tasca, Griffin & Rodriguez, 2010; Wright, Salisbury & Van Voorhis, 2007), a 12-month study period was decided upon. Next, publicly available prisoner population records were assessed to determine a possible sample size. The assessment of available records indicated that from January 2012, the prisoner population in Western Australia reached, and remained above 4,800 prisoners, from January 2012. Discussions were then held with appropriately authorised staff members within the Department, in line with the Departmental research and ethics approval received, to further determine which prisoners from the initial 4,800 prisoners could be included in the sample. These discussions revealed that approximately 2,200 of the-then approximately 4,800 prisoners in Western Australian prisons had been
imprisoned for at least 12 months on any give date in 2012, either sentenced to a period of imprisonment or remanded in custody by a court of law. Prisoners imprisoned for longer than 12 months reduced considerably in number, which suggested that a 12- month study period was appropriate to ensure the largest sample size possible, while enabling the inclusion of seasonal situational variables. In addition, this methodology ensured that all male, female, non-Aboriginal and Aboriginal prisoners who had been imprisoned in Western Australian prisons for 12 months to the study date could be included in the sample, reducing selection bias (Cuddeback, Wilson, Orme & Combs- Orme, 2004).
Consideration was then given as to whether to collect self-report data from as many as possible of the approximately 2,200 prisoners who had spent 12 months or more in prison in Western Australia, or to request the Department to provide official data. Collecting self-report data from each of the possible 2,200 prisoners had obvious security, safety and resource implications for the Department and the researcher. Firstly, if semi-structured interviews were employed to collect data, the researcher, or
associates, would have been required to visit each of the 16 prisons across the state of Western Australia, and although a serving prison officer at the commencement of the research, necessary supervision would have been required while visiting and
interviewing each prisoner, to maintain the good order and security of each facility. In addition, some prisoners would have not been able to be accessed for the purpose of interview due to logistical or safety issues such as prisoners on work release in the community and those housed in high security areas of prisons. Also, participation of
each prisoner would have been voluntary, and many would likely refuse. If surveys were administered for collection at a later date, responses may not have been complete due to literacy issues, which may have precluded a number of prisoners from participating. In addition, the researcher would have been reliant to some extent on staff to assist prisoners in their understanding of what was being requested from them, in addition to the distribution and collection of study materials, which would have imposed an
unacceptable burden on staff. Finally, if surveys were distributed or collected by prison officers, some prisoners may have felt compelled to participate due to the power differential between prisoners and supervising staff. This method would therefore have an impact on the voluntariness of participation. For these reasons, the decision was made to request the Department to provide official data. The study date of 30
September 2013, shortly after Departmental research and ethics approval was received by the researcher, was agreed upon in conjunction with the Department, with data then being provided in line with Departmental research and ethics protocols, to the 12- months prior to this date.