202 
Fishery Bulletin 96(2), 1998 
11-12 13-14 15-16 17-18 
Area 
■ Spring HI Summer □ Fall 
Figure 5 
Probabilities of observing a set with high bycatch (> 0.033%) for the sea- 
son-area combinations. 
Table 4 
Estimated odds ratios and Wald chi-square (% 2 ) values for contrasts (in 
parentheses) for observing high bycatch between areas, given a set was 
made in a certain season. For example, the odds of observing high bycatch 
are 1.628 times greater in area 11-12 than in area 13-14. * indicates a 
ratio significantly different from 1. Wald y 2 values are in parenthesis. For 
all contrasts, y 2 df=l. 
Season 
Area 
Area 
11-12 
13-14 
15-16 
Spring 
13-14 
1.628 (0.33) 
15-16 
3.877 (6.64) 
2.380 (1.23) 
17-18 
10.857* (9.29) 
6.666(3.79) 
2.799 (2.13) 
Summer 
13-14 
0.791 (0.04) 
15-16 
1.407 (0.42) 
1.777 (0.23) 
17-18 
2.671 (2.33) 
3.374(0.94) 
1.898 (1.22) 
Fall 
13-14 
0.615 (0.27) 
15-16 
0.230 (3.63) 
0.375 (1.50) 
17-18 
0.288 (2.43) 
0.468 (0.84) 
1.250 (0.12) 
cant differences between these two sea- 
sons existed for area 11-12 (Wald 
% 2 =10.03, df=l,P>% 2 =0.0015). When the 
summer and fall seasons were con- 
trasted, a significant difference between 
seasons was observed only for area 11— 
12 (Wald % 2 =9.25, df=l, P>% 2 =0.0024). 
For contrasts of the odds of high 
bycatch between areas for a given sea- 
son, significant differences between ar- 
eas 11-12 and 17—18 in the spring sea- 
son were observed (Wald % 2 =9.29, df=l, 
P>^ 2 =0.0023). All other contrasts be- 
tween areas for the spring, summer, and 
fall seasons were not significant. Wald 
values for these contrasts are pre- 
sented in Table 4. 
The estimated odds ratios for the con- 
ditions associated with the hypotheses 
are also given in Tables 3 and 4. Only 
estimated odds ratios of contrasts for 
the rejected hypotheses of no difference 
are examined in detail. 
If a fishing boat was in area 11-12, 
the odds of observing a set with high 
bycatch are about 11.8 times higher if 
a boat was fishing in the spring rather 
than the fall. Also, if a fishing boat was 
in area 11-12, the odds of observing a 
set with high bycatch are about 0.1 
times (or approximately 10.3 times 
lower) in the fall than in the summer. 
It appears that for area 11-12, the odds 
of observing a set with high bycatch in 
the fall are significantly lower than in 
the spring or summer. 
The third significant contrast indi- 
cated that for a vessel fishing in the 
spring, the odds of observing a set with 
high bycatch are about 10.8 times higher 
in area 11-12 than in area 17-18. 
Refining the final model We were in- 
terested, on the basis of our contrasts, 
in determining if a model with simpler 
dichotomous classes for areas and sea- 
sons would provide as good a statistical fit as this 
“full model” (Eq. 1). We compared three potential 
models that had one or both of these variables with 
reduced classes against the full model (Table 5). For 
these models we classified area into two groups: 1 ) east 
of the Mississippi River; and 2) west of the Mississippi 
River. Season was also classified into two groups; 1) 
early — sets sampled April through August; and 2) late — 
sets sampled September through October. 
All three reduced models provided a fit as good as 
that of our full model (Table 5); therefore we chose 
the model that had reduced classes for season and 
area because it was the simplest. Four contrasts of 
interest were examined: 
1 Test for seasonal differences in the odds of ob- 
serving a set with high bycatch given the set was 
sampled east of the Mississippi River; 
