SPSS File Creation
First, I allocated a case number and pseudonym for each of the 185 individuals in the sample. I cross-checked names and dates of offences on the police file with the court files to ensure I had consistently assigned pseudonyms and case numbers in the files. Then, using IBM SPSS Statistics 20, I created a single electronic file with more than 100 variables from the three court files. I cross-referenced names and dates of birth to identify the number of distinct victims, and individuals who were named as the respondent and as the aggrieved on DVOs. I also cross-referenced dates of DVO applications on the respondent and the aggrieved file to identify if the individual had been first the respondent or the aggrieved; and names, dates of birth and dates of DVO applications to identify cross-applications, and whether cross-applications were made simultaneously or consecutively. In some cases, I saw that the respondent was also an aggrieved but I was unable to tell if this involved a cross-application because the other party was not included in the sample. As variables were created, I recorded them in a data code book for the electronic management of the combined data to facilitate later data analysis.
I also created two variables from the written police reports: level of violence used and whether or not there had been charges under the Criminal Code Act 1899. Using the case numbers I had allocated to each of the 185 people in the sample, I added coded data from the police files to the SPSS data file and added the variables and coding details to the code book. The police reports varied in the detail provided, and I could not use some variables I created from the reports for the statistical analysis (e.g. jealousy and presence of alcohol or other drugs), but they are included in the analysis of the police reports in Chapter 6.
I identified and eliminated the eight cases involving data entry errors at the point of collection (i.e. cases entered twice, and those with no data entered). As data for each case were entered, I checked for accuracy and I checked again when I had entered all the data for all cases. I then ran frequencies for all data and, through a variety of procedures, I identified and corrected errors.
Some of the variables in the data set had a very large number of attributes. For example, the court data file identified 43 court locations in which a magistrate made any DVO related to the sample as respondents, and 28 court locations where a breached DVO related to the sample as respondents had been made. This level of detail may be useful for understanding the mobility of those in the sample, but it is less helpful in comparing four groups. I recoded data for these variables as remote or other court locations, creating a dichotomous variable for analysis.
Some of the variables I created from the police reports were also re-coded. For example, I created a variable from the reports on the level of violence used in breach offences with the following five attributes:
No actual violence—usually a breach of a no contact condition or entering prohibited premises, but a small number of cases involved a threat of violence.
Low level violence—includes damage to property; a slap or shove without injury.
Medium level violence– physical abuse of a limited nature (e.g. one punch to the shoulder; a slap) not causing injury. It may also include significant property damage or other behaviour that would likely cause considerable fear.
High level violence—more substantial violence or exacerbating circumstances including pregnancy and violence towards children; and in a few cases the respondent’s violence towards self.
Extremely high level violence—stabbing/multiple blows/kicks causing injury; assault with a weapon; strangulation, rape/attempted rape; suffocation/assault while victim restrained/trapped; risk of death or permanent injury.
There were few cases with no actual violence and few cases with extremely high level violence. Thus, I created three categories: I combined categories one and two as low level violence, I retained the category medium level violence, and I combined categories four and five into high level violence. I later re-coded this variable as a dichotomous variable for conducting statistical tests to compare the four groups. I also dichotomised other variables such as type of penalty and number of distinct DVO breaches.
Police Reports and Interviews
I copied the police reports from the encrypted compact disc to an encrypted electronic excel spreadsheet file on my password protected laptop computer. I had assigned case numbers and pseudonyms to link the police reports with the court files, which was also important for the content analysis to follow.
I transferred the audio-recorded interviews to encrypted files on my laptop and deleted the digital recordings. In the following weeks I had the recorded interviews transcribed and I assigned each participant a coded identity. The coding protocol first identifies the participant as a service provider (SP) or a police prosecutor (PP), followed by a number representing the order in which they were interviewed to distinguish them from their counterparts. For example, PP3 is the third police prosecutor to have been interviewed. Five of the service providers I interviewed identified as Indigenous. To add further context to the reflections from service providers, the coding for them includes an additional element to represent Indigenous (I) or non-Indigenous (NI) status. SP/11/I, for example, is an Indigenous person and the eleventh service provider interviewed. For the sake of anonymity, I do not identify whether the participant was male or female, nor their location (Cairns or Mount Isa), except for the case of Thelma in the opening paragraph of Chapter 1, where I disclose the location as the Cairns Esplanade. This is a very large public area, where many Indigenous people gather, so anonymity is not at risk.