A second problem with stepwise multiple regression is more widely recognized and yet appears not to have deterred many ecologists from using the technique. The problem is that the algorithm used (forward selection, backward elimination or stepwise), the order of parameter entry (or deletion), and the number of candidate parameters, can all affect the selected model (e.g. Derksen & Keselman 1992). This problem is particularly acute where the predictors are correlated (e.g. see Grafen & Hails 2002 for an example). In addition, the number of candidate parameters has a positive effect on the number of nuisance (or noise) variables that are represented in the selected MAM (Derksen & Keselman 1992).