Note: the original hard copy of this report is
NSW Legal Needs Survey in disadvantaged areas: Fairfield, Justice issues paper 5
Justice made to measure: NSW legal needs survey in disadvantaged areas (2006) is the report of a large-scale quantitative study of the legal needs of disadvantaged people in six local government areas of New South Wales. More than 2400 residents across the regions were interviewed about their legal needs. This report was preceded by an initial study Quantitative legal needs survey: Bega Valley (pilot) (2003). There now follows a series of papers in the Justice Issues imprint. Six individual papers will describe how disadvantaged people deal with legal problems, detailing the responses from one of the regions surveyed: Campbelltown, Fairfield, Nambucca, Newcastle, South Sydney and Walgett....
The main statistical technique used to test differences among regions was logistic regression. Logistic regression is an appropriate form of multivariate analysis when the outcome variable is discrete rather than continuous. Like other forms of regression, it examines the relationship of an outcome variable (e.g. whether someone has experienced a legal event or not) to one or more potential predictor variables (e.g. geographical region). In the regressions performed for this report, deviation contrasts were used to determine whether this region was different from the average of all regions to a statistically significant degree. Standard logistic regressions were used for all data where respondent was the unit of analysis (up to Figure 2 in the main body of the report). For the analyses where legal event was the unit of analysis (Figure 2 onwards in the main body of the report), mixed effects binary logistic regression was used (Hedeker 1999). While standard logistic regression assumes the independence of observations, mixed effects logistics regression allows for observations to be correlated. In detail, in the current study, where participants were the unit of analysis, there was only one observation per participant. However, where legal event was the unit of analysis, legal events were clustered within participants. That is, the one participant could have multiple legal events and therefore legal events were not independent of each other. The mixed effects logistic regression technique adjusts the statistical analyses appropriately for this clustering effect.