Table 1. Results of multiple linear regressions (Type III sums of squares) investigating the influence of forest edge configuration and conspecific group size on three response variables describing aspects of the paths made by Black-capped Chickadees in detour trials (n = 151; see Fig. 2 for variable definitions).

Response variable

Independent variablea

Beta

1 SE (Beta)

Partial r

Partial F

P

Proportion of path length traveled in the open (m)
EpsilonDOa/(EpsilonDOa+EpsilonDFa)b 		DOi / DFi
-126.4
12.5
-0.64
101.8
<0.0001
DOi (m)
1.04
0.21
0.39
25.3
<0.0001
DOi x DOi/ DFi
-2.04
0.29
-0.50
49.1
<0.0001
Group size
0.33
0.44
0.06
0.6
0.5
Constant
90.1
14.8
-
37.3
<0.0001
Risk area (m2)
(square-root transformed)		 DOi / DFi
-68.4
6.3
-0.67
116.7
<0.0001
DOi (m)
0.09
0.03
0.25
9.9
0.002
DOi x DOi/ DFi
-0.14
0.16
-0.07
0.8
0.4
Group size
-0.32
0.27
-0.10
1.4
0.2
Constant
22.1
2.1
-
109.7
<0.0001
Forest edge distance (m) DOi / DFi
-35.5
4.04
-0.59
77.2
<0.0001
DOi (m)
0.07
0.02
0.33
17.4
0.0001
DOi x DOi/ DFi
-0.26
0.10
-0.21
6.8
0.01
Group size
-0.17
0.17
-0.08
1.0
0.3
Constant
13.3
1.3
-
98.1
<0.0001

a DOi (80.4 + 33.3 m, mean + 1 SD) and DOi/ DFi (0.67 + 0.17) were centered on their means to reduce collinearity. Group size averaged 7.3 + 3.3 individuals and ranged between 1 and 18 individuals.

b To avoid confounding multiple response variables in the analysis, the response variable was, in fact, Epsilon DOa and the model included the total path length (Epsilon DOa + Epsilon DFa) as a covariable. Because the effects of the covariable are held constant in the analysis, the model can be interpreted as investigating the influence of the other independent variables on the proportion of path length traveled in the open.