Forest list archive: msg00062

[Prev][Next][Index][Thread]

Troubles with statistics...



I am a PhD Student in forest ecology and, these days, I'm having a lot of 
fun analyzing my data.  My objective is to evaluate the effect of many 
factors on the diameter growth of black spruces, including stand age, 
competition and spruce budworm defoliation.  Competition pressure is 
estimated with the relative density, which is the observed density (stems 
per ha) divided by the maximum potential density, according to the
stand density management diagram (developed by Newton and Weetman 1993).

My goal is to quantify the importance of each factor on tree growth. 
However most of these "independent" variables are in fact correlated to 
each other .  For instance, relative density is strongly correlated 
with stand age, since the competition stress increases with tree size.  
The occurence of spruce budworm outbreaks is also correlated to stand 
age, since outbreaks occur during specific periods on large territories 
and most of my stands have the same age.  

This multicorrelation, or colinearity,  seems to be a major obstacle to 
the use of multiple regression (cf Neter, Wasserman & Kutner 1990). When
collinearity is important, the estimation of the partial regression 
coefficients seems to be hazardous.

On the other hand, in their classical handbook, Sokal & Rohlf (1981) suggest 
path analysis as a powerful tool to investigate complex networks of 
inter-correlated variables.  This approach is based on the determination 
of causality relationships; the strengh of these relations is estimated 
with multiple regressions. However, Sokal & Rohlf seemed not to bother 
with colinearity...

Since I obviously  do not fully understand the nuts and bolts of multiple 
regression and path analysis, I need help to answer these questions:

(1) Is path analysis a relevant approach to solve my theorical problem 
(i.e. evaluate the importance of competition on growth)?

(2) If it is relevant, how should I manage the colinearity problem?

(3) If path analysis is not adequate, are there possible alternatives?

I hope the description of my problem is sufficient.  Any comment from my 
colleagues forumites will be welcome!

HELP ME! I'M (ALMOST) DESPERATE!!!

Jean-Martin Lussier
Universite du Quebec a Chicoutimi
E-Mail: jmlussie@uqac.uquebec.ca




[Metla] [Main Index] [Thread Index]

Mail converted by MHonArc 1.1.0