population was checked against the records of the Fish and 

 Wildlife Service and any vessel which had not landed at least 

 one comnercial catch of fish or shellfish during the year 19$3 

 was excluded. Third, after the samples had been drawn at random 

 the available records were checked and vessels whose ownership 

 changed during 19^0-51; were eliminated. (Substitution for change 

 of ownership was made in the Gulf Area and California samples 

 only.) Vessels from the same stratum were substituted for ones 

 disqualified. Fourth, any vessel which during interviewing was 

 found to be engaged in service other than commercial fishing, 

 such as, sport fishing, was disqualified and a substitute vessel 

 from the same stratum was interviewed. 



Despite these amendations, the statistical ujiiverse may have 

 fallen a little short of a 100 percent coverage of the sample- 

 units. This situation may be attributed to the relative ease 

 with vfhich a vessel may enter or leave commercial fishing for 

 other maritime activities. Field work experience has sub- 

 stantiated the original assurance of the Fish and Wildlife Service 

 that the nm.iber of such vessels was relatively small, 



(b) Sample size and standard error . A sample of 1,^00 commercial 

 fishing vessels vjas distributed among the five geographical areas. 

 New England, Middle Atlantic, Gulf Area, California, and Pacific 

 Northwest on the basis of a minimum of 200 sample-ujiits for each 

 area and the distribution of the remaining 500 sample-units in 

 proportion to the vessel population in each area. (See table B-1 

 in later pages of this appendix). After the interviewing of the 

 New England vessel owners in the sample was almost completed, 

 costs limited the survey to the study of the insurance experience 

 in New England, the Gulf ^rea, and California. The sample size 

 for New England and California remained the same, 251 and 288 

 sample-units respectively. The Gulf Area was redefined to ex- 

 clude the South Atlantic Fisheries, and the original sample size 

 of 389 sainple-Tinits was reduced to 225 vessels. 



Assuming a parameter percentage of 50 percent, i.e., only 50 per- 

 cent of the universe carrying insu rance of e ither kind, and 

 applying the formula E = 100 J 6.65'x p x q the error is 9.1 



N 

 percent for 200, 8.2 for 250 and 7.5 for 300 sample-units within 

 100 to 1 certainty limits, 



(c) First-stage sampling involved the selection of a nuiaber of 

 home ports in each ai'ea (except California) v;ith a combined 

 vessel registration of no less than OU.O percent of the statistical 

 universe. In this manner, fifteen New England home ports and nine 

 in the Gulf Area vjere excluded from final sampling, thus reducing 

 considerably the cost of field work, (For first-stage sampling 

 see tables B-2 and B-3 iJi later pages of this appendix.) 



319 



