Chapter Six – Data Collection
6.4 Towards a Central Analytical Framework
Having identified the data available for addressing this thesis’s proposed course of analysis (based on the analytical techniques discussed in Chapter 5 for investigation of the cohort density and age composition expressions of the age-structure crime pattern set out in Chapter 3, and the underlying nature of the age-crime pattern discussed in Chapter 2), this thesis’s central analytical framework is as follows.
The distribution and correlation analyses discussed in Chapter 5, which are conducted to provide insight into the association between age and crime and as a measure of the strength of the relationship between age structure and crime, are performed for only South Australia and Western Australia, utilising the same age groupings and offence categories as those applied in the analyses of the cohort density and age composition expressions. These analyses form Chapter 7. Neither expression of the Easterlin hypothesis is explicitly addressed in this chapter. Rather, the purpose of Chapter 7 is to confirm or deny whether the age-crime and age structure-crime patterns are generally evident for the two regions being analysed. This includes an initial indication of whether the nature of these two associations appears to be variant or invariant.
The cohort analyses discussed in Chapter 5 are conducted as a means of indicating how birth cohort size and relative disadvantage influence cohort-specific apprehension trends (the cohort density expression). Cohort analyses form Chapters 8, 9, and 10, looking firstly at total apprehensions, followed by offence-specific apprehensions, and finally the set of apprehension trends for each cohort, for both males and females. This aspect of the analysis is restricted to Western Australia, being the only one of the two possible regions identified for analysis where age groups in the apprehension data have been grouped equally. The age groups pertain to four year periods; consequently, age-specific ratios will be observed in 1994, 1998, and 2002. These observation intervals allow for the offence trajectories of five birth cohorts to be examined. These are the cohorts born 1957-60, 1961-64, 1965-68, 1969-72, and 1973-76. The sizes of these five Western Australian birth cohorts (males and females) are included in Appendix B (section B.4).
The cohort density expression would be supported if the apprehension ratios for the Western Australian cohorts born 1969-72 and 1973-76 are seen to depart from the classic age-crime pattern between 1994 and 2002. That is, its ratios do not decline consistently as the cohorts have aged, suggesting an extended participation in criminal activity (i.e. beyond that suggested by the age-crime pattern). This is because the Western Australian birth cohorts born 1969-72 and 1973-76 are the most reflective of the period that relates to the Australian baby bust cohort (which was born between 1968 and 1974, and peaked in 1971). In this analysis, the former cohort can thus be regarded as a leading edge cohort, and the latter a lagging edge cohort. Accordingly, should the cohort born 1969-72 be seen to experience the greatest departures from the age-crime pattern, this would indicate that size per se is more influential than relative disadvantage with regard to cohort-specific apprehension trends, but vice versa if it is the cohort born 1973-76 that has experienced the most pronounced extension of involvement in criminal activity. Should these pivotal cohorts be seen to have experienced increasing apprehension rates as they aged, this can be regarded as evidence of variance in the age-crime pattern, as can any difference in such increases by gender or offence category (the small base for many specific offences making analysis of such unviable at this time).
Of the five Western Australian birth cohorts being analysed, those born 1957-60, 1961-64, and 1965-68 are the smallest. Therefore, we would expect that apprehension ratios for these smaller birth cohorts will be the most reflective of the age-crime pattern (i.e. ratios should generally decline as the cohort ages, indicating that the cohort has not experienced an extended period of offending). The cohort density expression would thus be negated if the apprehension ratios of the smaller Western Australian cohorts born 1957-61, 1960-64, and 1965-68 are more likely to depart from the age-crime pattern than those of the large Western Australian cohorts born 1969-72 and 1973-76.
Further, the cohort density expression would be regarded as plausible if either: 1. Both larger and smaller Western Australian birth cohorts have experienced
increasing apprehension ratios as they aged between 1994 and 2002; or 2. Cohort-specific apprehension ratios for the five Western Australian birth
cohorts suggest that a common period effect (such as a change in policing strategies) has impacted on all of their apprehension ratios between 1994 and 2002.
As suggested previously, departures from the age-crime pattern for this aspect of the analysis are categorised as being:
1. Significant if apprehension ratios have increased (by more than five per cent) as a cohort has aged;
2. Substantial if apprehension ratios have been relatively stable as a cohort has aged (either increasing or declining by less than five per cent); and
3. Minor if apprehension ratios show minimal change in the rate of decline as a cohort has aged (the rate of decline decelerating by more than one-quarter).
Importantly, as the sources of apprehension data do not include an unemployment variable (the life event featured in the analyses as a proxy variable for relative disadvantage), cohort-specific apprehension trajectories will be considered in light of Western Australian unemployment rates. These unemployment rates are reorganised from Labour Force Status by Sex, Age, State, Marital Status (ABS 2008a) to reflect
those experienced by male and female birth cohorts when first seeking entry to the workforce. These rates are graphed in Appendix B (section B.7).11
The comparative analyses discussed in Chapter 5 (i.e. standardisation and decomposition analyses) are conducted as a means of measuring the impact of structural ageing on total apprehensions levels, both retrospectively and prospectively (the age composition expression). As the South Australian data are available for a longer period of time than the Western Australian data, the retrospective analyses are restricted to the former region. As suggested previously, the analyses are conducted over two separate periods (1987-1997 and 1998-2004) to account for the difference in counting rules. These analyses, which focus on the past influences of changing age composition on male and female apprehension trends (initially in relation to total apprehensions, and subsequently for offence-specific change), form Chapters 11 and 12. The age compositions of the South Australian population over these two periods are graphed in Appendix B (section B.5). However, to allow for a regional comparison of age compositional change, the prospective analyses will be conducted for both South and Western Australia (males and females), 2004-2051. These analyses form Chapters 13 and 14, and are similarly conducted first for total apprehensions, and subsequently for offence-specific apprehensions. The prospective age compositions of these two state-level populations are also graphed in Appendix B (section B.6).
The age composition expression would be supported if total apprehension levels in South Australia (1987-1997, 1998-2004, and 2004-2051) and Western Australia (2004-2051) have been (or should be) reduced by the ageing population composition. That is, apprehensions levels would have been higher than observed had the population not aged structurally.
11
These figures have been annotated in accordance with the period of cohorts, and thus reflect four year age groupings. However, the original data are organised around five year age groupings.
Conversely, the age composition expression could be negated in one of two ways: 1. The ageing population composition has had no effect on total apprehension
levels; or, more significantly,
2. Total apprehension levels have actually increased as a consequence of the ageing population composition. That is, apprehension levels would have been lower than observed had the population not have aged structurally.
Further to these specific propositions, should there be differences in the impact of age composition effects by gender or offence category, this would be indicative of variance in the underlying association between age and crime. Furthermore, some differences in the ‘age’ of the two regional populations being analysed could accordingly be expected to result in some differences in the regions’ respective experiences of the age structure-crime pattern. South Australia is an older region, while Western Australia is a younger region (see Chapter 3).12 Generally, therefore, it could be expected that:
1. Retrospective age composition effects will be greater for South Australia. This is because it is a substantially older population, and has been ageing structurally at a much greater rate than that of Western Australia; and
2. Prospective age composition effects will be greater for Western Australia. This is because the sub-national population has yet to really start ageing, and thus is also still growing strongly.
The quantitative analyses commence in the next chapter with an exploration of the age-crime and age structure-crime patterns in South Australia (1987-1997 and 1998- 2004) and Western Australia (1994-2002).
12
Assumptions underlying the ABS (2005b) population projections also differ by state. For South Australia the assumptions are: a TFR of 1.71 in 2004, falling to 1.66 in 2016 and then remaining constant; total net migration of 1,108 in 2004, rising to 1,350 by 2006 and then remaining constant; and life expectancy continuing to increase but at a decelerating rate. For Western Australia the assumptions are: a TFR of 1.75 in 2004, falling to 1.67 in 2021 and them remaining constant; total net migration of 14,906 in 2004, rising to 17,108 the following year and remaining constant thereafter; and life expectancy continuing to increase but at a decelerating rate.