Table 1. Multiple logistic regression analysis of simulation results. The response variable is ln(p/(1-p)), where p is the probability that the landscape with woody borders has a smaller pest population (better pest control) than the landscape without woody borders. We conducted 10,000 simulation run pairs (with and without woody borders). Random values for each predictor variable were selected at the beginning of each run, as described in the text. Variables are ranked in decreasing magnitude of effect. All coefficients are significant (alpha = 0.05), except the one for FIELDS.
Predictor variable
|
Sign
of coefficient
|
Wald
chi-square
|
DISPWOOD/DISP
|
-
|
824
|
MAXAGE
|
-
|
671
|
SURVWOOD/SURV
|
-
|
326
|
FIND
|
+
|
71
|
FIELDS
|
|
not
significant
|