Just Search page
Note: the original hard copy of this report is 350 pages .

cover image

Justice made to measure: NSW legal needs survey in disadvantaged areas   

, 2006 Six disadvantaged areas were surveyed by telephone interviews: three suburban areas within Sydney (Campbelltown, Fairfield, South Sydney), one major provincial centre (Newcastle) and two rural/remote areas (Nambucca and Walgett)...


Demographic factors related to reporting different types of legal events


To assess whether the types of events experienced were related to the characteristics of participants, a series of standard binary logistic regressions were performed. Each regression examined whether sociodemographic factors were associated with whether or not participants reported experiencing one or more events from a particular legal event group.10 Given that the frequency of reporting some types of events was low, there were insufficient numbers to conduct a separate regression for some legal event groups. As a result, regressions were performed for the 10 most frequently occurring legal event groups. These event groups comprised eight civil legal event groups (i.e. accident/injury, consumer, credit/debt, education, employment, government, housing and will/estates), one criminal legal event group (i.e. general crime) and the family legal event group.

The full results of these 10 logistic regression models are presented in Tables C4 to C13 in Appendix C, and the corresponding descriptive statistics are presented in Tables C14 to C23 in Appendix C. The results of these regressions are discussed in turn below.

Although the relationships of sociodemographic factors with reporting the five least frequent types of legal events11 were not examined via regression analyses, they were examined via chi-square analyses. It is worth noting that, unlike regression analyses, chi-square analyses only examine the bivariate relationship of each sociodemographic factor to reporting each type of event. That is, chi-square analyses do not take into account the interrelationships between sociodemographic factors and their combined effect on reporting each type of event. The chi-square results and the relevant cross-tabulations for the five least frequent legal event groups are presented in Tables C24 to C28 in Appendix C.

Accident/injury events

The logistic regression results revealed that gender, age, country of birth, disability status and personal income were statistically independent predictors of reporting one or more accident/injury legal events. Indigenous status and education were not significant predictors of reporting accident/injury events (see Appendix Table C4).

More specifically the odds of reporting at least one accident/injury event were:


Consumer events

According to the logistic regression model, age, disability status and personal income were statistically independent predictors of reporting at least one consumer event. Gender, Indigenous status, country of birth and education were not significant predictors of reporting consumer events (see Appendix Table C5).

The odds of reporting at least one consumer event were:


Credit/debt events

The logistic regression showed that age, Indigenous status and disability status were statistically significant predictors of reporting credit/debt events. The remaining sociodemographic variables were not significant (see Appendix Table C6).

The odds of reporting at least one credit/debt event were:


Education events

Age and disability status were the only sociodemographic factors that were statistically significant predictors of reporting at least one legal event related to education (see Appendix Table C7).12

Specifically, the odds of reporting at least one education event were:


Employment events

Based on the logistic regression model, age, Indigenous status and disability status were statistically significant predictors of reporting employment events. The remaining sociodemographic variables were not significant (see Appendix Table C8).13

The odds of reporting at least one employment event were:


Although age was also a significant predictor of reporting employment events, none of the specific comparisons tested in the regression were significant.14 The highest incidence of employment events was reported by 45 to 54 year olds (24.9%), followed by 15 to 24 year olds (22.6%) and 25 to 34 year olds (22.2%, see Appendix Table C18).

Government events

Age, disability status and education level were statistically independent predictors in the logistic regression model for reporting government events. The remaining sociodemographic variables examined were not significant (see Appendix Table C9).

More specifically, the odds of reporting at least one government event were:


Housing events

The logistic regression revealed that age, disability status and personal income were statistically independent predictors of reporting housing events (see Appendix Table C10). The odds of reporting at least one housing event were:


Gender, Indigenous status, country of birth, and education level15 were not significant predictors of reporting housing events.

Wills/estates events

Age, Indigenous status, country of birth, personal income and education level were statistically independent predictors in the logistic regression model for reporting wills/estates events. Gender and disability status were not significant predictors (see Appendix Table C11).

More specifically, the odds of reporting at least one wills/estates event were:


General crime events

Age, country of birth, disability status and personal income were statistically independent predictors in the regression model (see Appendix Table C12). The odds of reporting at least one general crime event were:


Gender, Indigenous status and education level were not significant predictors of reporting general crime events.

Family events

Age, Indigenous status, disability status and personal income were statistically independent predictors in the regression model (see Appendix Table C13). The odds of reporting at least one family event were:


Although personal income was a significant predictor of reporting family events, none of the specific comparisons tested, using the highest income bracket as the reference category, were significant. The highest rates of family events were reported by the middle two income groups (9.9% and 9.6%) while the lowest rate was reported by the lowest income group (4.9%). The rate for the highest income group (6.3%) fell in between these other rates (see Appendix Table C23).

Gender, country of birth and education level were not significant predictors of reporting family events.



In each case, a standard rather than mixed-effects logistic regression model was appropriate because there was only one observation for each individual: for example, reporting at least one accident/injury event versus not reporting any accident/injury event.
The five least frequent legal event groups in the present study were the business, health, human rights, domestic violence and traffic offence groups.
Of the 2431 respondents, only 1076 were students or were responsible for students. Thus, only these 1076 respondents had the potential to experience an education event. The regression is based on 913 of these 1076 participants who did not have any missing data on the sociodemographic variables. Because only one person aged 65 years or over reported an education event, this age group was combined with the 55 to 64 year age group, and the combined (55 years or over) age group was used as the reference category in the regression.
Of the 2431 respondents, only 1417 were employed at some time during the 12 months prior to the survey, so only these respondents had the potential to experience an employment event. The regression is based on the 1195 of these 1417 participants who did not have any missing data on the sociodemographic variables. Because only two people aged over 65 years reported an employment event, this age group was combined with the 55 to 64 year age group, and the combined (55 years or over) age group was used as the reference category in the regression.
However, it should be remembered that, for predictors that have three or more categories (such as the age predictor in the present case), regression analyses do not make comparisons between all possible pairs of categories.
Note that even though the overall education variable was not significant in the regression, one of the comparisons for education was significant.

10  In each case, a standard rather than mixed-effects logistic regression model was appropriate because there was only one observation for each individual: for example, reporting at least one accident/injury event versus not reporting any accident/injury event.
11  The five least frequent legal event groups in the present study were the business, health, human rights, domestic violence and traffic offence groups.
12  Of the 2431 respondents, only 1076 were students or were responsible for students. Thus, only these 1076 respondents had the potential to experience an education event. The regression is based on 913 of these 1076 participants who did not have any missing data on the sociodemographic variables. Because only one person aged 65 years or over reported an education event, this age group was combined with the 55 to 64 year age group, and the combined (55 years or over) age group was used as the reference category in the regression.
13  Of the 2431 respondents, only 1417 were employed at some time during the 12 months prior to the survey, so only these respondents had the potential to experience an employment event. The regression is based on the 1195 of these 1417 participants who did not have any missing data on the sociodemographic variables. Because only two people aged over 65 years reported an employment event, this age group was combined with the 55 to 64 year age group, and the combined (55 years or over) age group was used as the reference category in the regression.
14  However, it should be remembered that, for predictors that have three or more categories (such as the age predictor in the present case), regression analyses do not make comparisons between all possible pairs of categories.
15  Note that even though the overall education variable was not significant in the regression, one of the comparisons for education was significant.