ABSTRACT
It has been intimated that you can make statistics show anything you want them to; politicians from opposing parties often use the same data to provide
Introduction ........................................................................................................ 277 Statistical Approaches ....................................................................................... 278
Comparative Statistics ............................................................................. 278 Systematic and Random Error ................................................... 279 Type 1 and Type 2 Errors ........................................................... 279 Statistical Significance ................................................................ 281
Multivariate Methods ........................................................................................ 281 PCA............................................................................................................. 282
Normality and Transformations ............................................... 287 Mean Centring to Unit Variance .............................................. 288 Using Proportions and Closure ................................................. 288 Zeros .............................................................................................. 291 Summary: Checklist for PCA .................................................... 292
PLS .............................................................................................................. 292 Overlap between Signatures....................................................... 295 Summary: Checklist for PLS ...................................................... 296
Geostatistics ........................................................................................................ 297 Variogram/Semivariogram Analysis ..................................................... 298
Kriging .......................................................................................... 301 Anisotropy .................................................................................... 302 Summary: Checklist for Geostatistics ...................................... 304
References ............................................................................................................ 304
support for opposite standpoints. In an environmental forensic context, however, statistics can lend considerable support to other evidence and lead to the interpretation of complex data. The use of statistical methods must be conducted with utmost rigor and be applied only where relevant and necessary. Statistics for the sake of statistics may lead to doubt being raised and imply obfuscation. It has also been suggested that a well planned and executed analytical program should not need statistics to prove a point. This may be true when simple comparisons are all that is required; however, when examining complex signatures, statistical methods might be the only ones that can identify the underlying truth.