In addition to using variable importance measures as a merely descriptive means of data exploration, different significance tests and schemes for variable selection have been suggested: On the official random forests website, a simple statistical test based on the supposed normality of the z-score is proposed by Breiman and Cutler (2008, date of access), that has been applied in a variety of studies – ranging from the investigation of predictors of attempted suicide (Baca-Garcia et al. 2007) to the monitoring of a large area space telescope on board of a satellite (Paneque, Borgland, Bovier, Bloom, Edmonds, Funk, Godfrey, Rando, Wai, and Wang 2007).
