*2.5. Statistical Analysis*

Data were analysed using SPSS Statistics IBM version 21 (2012). Soil physicochemical parameters were analysed for descriptive statistics (mean ± standard error). Roadside and control sites were compared using one-way ANOVA, with significance taken as *p* < 0.05. Where necessary, data were square root or log10 transformed to obtain an acceptable normal distribution and to stabilise the variance of the residuals, and significance was based on an analysis of the transformed data. Relationships between the metal of interest and other physicochemical parameters were analysed using a 2-tailed Pearson correlation test; significance was treated as *p* < 0.05. Optimised linear models for soil metal concentrations were determined using stepwise regression with criteria of α = 0.05 for entry. The input parameters used were soil physiochemical properties including pH, EC, ECEC, TOC, Tc, DGT, soil total metals, CaCl2 extractable metals, water extractable metals, and the age of the road as a measure of metal aging in soil. Linear correlations were also developed; *p* < 0.05 was considered significant and 0.05 < *p* < 0.10 was considered marginally significant.
