BAYLIFF: INTEGRITY OF SKIPJACK TUNA SCHOOLS 



Table 6.— Continued. 



there were sets with more than five tagged fish in 

 them, and yet the results of the Chi-square tests 

 were not significant. This is caused by the require- 

 ment that each category have an expected value of 

 5 or greater. Accordingly, it is likely that if there 

 had been more tag returns, there would have been 

 more categories for many of the tests and signifi- 

 cant results for more of them. The practice of com- 

 bining the results for fish of the same experiment 



released on different days ("Total" lines in Table 

 4) and for fish of different experiments ("Total" 

 lines of Table 5) helps to overcome this weakness. 

 It can be seen in the text table above that the ratios 

 increased with the degrees of freedom. If there were 

 more tag returns, there would be fewer tests with 

 one degree of freedom and more with three or more 

 degrees of freedom, and the ratios of significant 

 tests would almost certainly increase. 



641 



