Reader Comments

Post a new comment on this article

Colinearity

Posted by jebyrnes on 21 Jun 2007 at 21:16 GMT

Time-dependent variation in resighting rates was better explained by local rainfall (45.1% variance explained) than by the NAO (33.5% variance explained, Figure 1a).
http://plosone.org/article/info:doi/10.1371/journal.pone.0000539#article1.body1.sec2.p1

I'm curious, are these two variables colinear? Would a model that incorporated this actually provide better explanatory power (i.e., a structural equation modelling approach where you had NAO->Rainfall->Resighting Rate).

RE: Colinearity

jordif replied to jebyrnes on 23 Jun 2007 at 10:45 GMT

This is a very interesting question I will try to solve with SEM. However, what I can say at this moment is that although NAO is negatively related to rain in Spain and in western Mediterranean in general, the correlation between NAO and local rainfall is not very high (R2=0.07, N=26 years, p=0.17)