A letter in a recent (2011) article of Nature doi:10.1038/nature09817 refers in several places to a Fisher's predicted least-square difference test. A google search returns 102 hits for "predicted least square difference" and 3 for "predicted least squares difference". The first two pages seem to mostly come from journal articles and none of the ones I checked included a citation. Those that reported the use of a particular analysis package tended to mention StatView. However, StatView seems to have been swollowed by SAS at some point and then disappeared. In the only version of their manual I found refered to a PLSD but called it a "Protected Least Significance Difference" post-hoc. That term has 9980 results on google, again mostly journal articles. Is it safe to assume this is really just a regular Tukey LSD post-hoc that has been only run when the omnibus ANOVA was judged to be statistically significant?
Asked
Active
Viewed 257 times
2
-
I would try to contact the authors of that Nature article.. – user603 May 25 '13 at 12:52
-
I considered it. However there there are enough other papers that claim to use the same statistic that it seemed more appropriate to discuss the point here if only to provide a web resource for those who have a similar question. – russellpierce May 26 '13 at 04:46