FISHERY BULLETIN: VOL. 84, NO. 3 



resulting from the unrepresentative areal sample, 

 precluded testing for an area effect during the 

 1973-76 period (Table 3); however, after 1978 mean 

 kills were shown to differ among the three areas, 

 as noted earlier. Therefore, to minimize the amount 

 of bias which might be introduced into the estimates 

 from a sample which was unrepresentative of area, 

 I retained the three-area stratification. 



Small sample sizes during 1973-76 dictated pool- 

 ing over all quarters within both the North Outside 

 and South areas and over quarters 3 and 4 within 

 the North Inside area (Table 3). The test result for 

 quarter 1 and 2 differences in the North Inside area 

 was inconclusive (Table 6). Based on bias considera- 

 tions similar to those above, I did not pool data from 

 quarters 1 and 2 in the North Inside area in case 

 their means did indeed differ. 



After pooling over year, vessel capacity, and 

 quarter as indicated above, five strata remained for 

 the 1973-76 data: 1) North Inside, quarter 1, 2) 

 North Inside, quarter 2, 3) North Inside, quarters 

 3 and 4 pooled, 4) North Outside, all quarters 

 pooled, and 5) South, all quarters pooled. 



Tests for Differences Within 

 the 1977-78 Period 



Interpretation of the four-way ANOVA (Test 3), 

 restricted by data sparseness to the North Inside 

 and North Outside areas during the second and third 

 quarters of 1977-78 (Table 3), was complicated by 

 significant interaction between quarters and each 

 of the other three factors (P < 0.0159, for total kill 

 and all transformations of total kill). Therefore, the 

 four-way table was partitioned into two, three-way 

 tables, one for each level of quarter (Tables 7, 8), 

 and decisions about between-strata differences dur- 

 ing these years were based on results obtained 

 separately for each quarter. 



For second quarter data, interactions were not 

 significant (Table 7). However, for third quarter 

 data, interaction between year and vessel capacity 

 was significant (P < 0.001) primarily because of the 

 two large means recorded in the North Inside and 

 North Outside areas by small vessels during 1978 

 (Table 8). Omitting the data from one extraordi- 

 narily large kill set in each of these two cells reduces 

 their means from 11.27 to 4.68 for the North Inside 

 and from 9.77 to 5.46 for the North Outside. 



Results from the second and third quarter data 

 were inconsistent for both year and vessel capacity. 

 Neither effect was significant during the second 

 quarter (Table 7), but during the third quarter (Table 

 8) the large means in the two cells noted above pro- 



vided some evidence of both a year and vessel 

 capacity effect. Since the evidence for both a year 

 and vessel capacity effect was confined to two, third 

 quarter cells whose means were each strongly in- 

 fluenced by only one set, I concluded that year and 

 vessel capacity were not significant effects during 

 1977-78. Hence, I pooled over year and vessel 

 capacity within this period. The evidence regarding 

 an area effect during 1977-78 was also inconsistent 

 between quarters (Tables 7, 8); however, I retained 

 area as a stratification factor since it was shown to 

 be significant after 1978. 



Beginning in 1976 when the sampling program 

 became mandatory the sample data became more 

 representative of quarter. Thus, for the 1977-78 

 data, bias considerations were of lesser importance 



Table 7. — Mean of total number of dolphins killed (k), standard 

 deviation (s), and number of dolphin sets (d) for sampled trips, by 

 area, vessel capacity (tons), and year for all sets made in the North 

 Inside and North Outside areas during quarter two of 1977-78. 

 Significance probabilities (P) obtained for a 3-way ANOVA on total 

 kill: interactions (P > 0.2095), area (P = 0.0060), vessel capacity 

 (P = 0.8857), and year (P = 0.6050). 



Table 8.— Mean of total number of dolphins killed (k), standard 

 deviation (s), and number of dolphin sets (d) for sampled trips, by 

 area, vessel capacity (tons), and year for all sets made in the North 

 Inside and North Outside areas during quarter three of 1977-78. 



564 



