The best way to determine the contamination of a soil is to know the natural contents of that soil, or a soil equivalent in terms of climate and geological origin. Depending on the country, you may find these natural values for different soil types. Then you just have to calculate the geoacumulation index, or better the enrichment factor. Normalization with Ti or other low-labile metal is practical to reduce the effect of pedogenesis (climate) on metal concentration, especially if you use reference soil a little different from the studied soil.

Usually you would collect samples from the area you thought or knew was contaminated and then sample from a neighbouring property/site that had not been exposed to the same pollution. Ideally it should be on the same soil type / underlying geology so that you dont have large differences that are just because of the difference in location of the samples and not the pollution source. As Thomas said, you can use normalisation (ie looking at the ratio of your contaminant to a base standard like Fe, Li or in your polluted and background sample as Thomas suggested Ti rather than the just the difference in concentrations of the contaminant between samples).