(d) Second-stage sampling consisted of drawing a stratified 

 random sample of 2^1 vessels for New England, 225 for the 

 GuLf Area and 288 for California from the first-stage sample. 

 This was done as follows: First, the vessel popialation in 

 each area was stratified on the basis of four variables: (a) 

 active or lost vessels; (b) material, i.e., wood or steel 

 vessels; (c) age of vessels; and (d) gross tonnage. Second, 

 the stratified quotas of each sample were determined by 

 multiplying the size of each sample by the ratio of each 

 stratum to the xmiverse. Finally, a sample larger in size than 

 the one required in each quota was drawn at random from each 

 first-stage sample (For determination of stratified quotas of 

 the second-stage sample of vessels see tables B-U of this 

 appendix.) The larger sample size than the one required v?as 

 dra>m in order to allow substitution of vessels because of no 

 response fix)m the vessel owner, change of ownership, or for 

 vessels which did not qualify for interviewing because of 

 service other than commercial fishing. Trained personnel 

 interviewed the vessel owner of each vessel included in the 

 second-stage sample. 



(e) Third-stage sampling . The revised sample design provided 

 for the study of no less than SO percent of the total number 

 of years that the vessels in the second-stage samples had been 

 insured from 19^0 through 195ii. The study of insurance files 

 was conducted on the basis of availability and accessibility of 

 insuTfince records among leading agents and firms in each area. 

 The number of insurance policy-years studied in each area by 

 kind of insurance appear in table B-5 of this appendix. The 

 $0 percent minimum quotas were filled for all cases except for 

 protection and inderoiity policy years in the Gulf Area. In the 

 latter case insurance records were not available or, if available, 

 were not easily accessible. Trained personnel stadied the 

 insurance files and recorded the information on a detailed 

 schedule, 



3. Interview Schedules, Interview Instructions and Field V/ork . 



A considerable amount of preliminary work- was requii'ed for the 

 construction of the interview schedules because of lack of information on the 

 subject. Several revisions of the inteiview schedules were made after each 

 pretesting. Schedule A: Face Sheet was designed to record information 

 readily obtainable from the listing of Merchant Vessels of the United States 

 on the physical characteristics of the vessel. Information on the vessel's 

 fishing operations and other data which were expected to be related to 

 insurance experience were incorporated in Schedule B: Oimers Data. Ex- 

 perience on hull insurance and protection and indemnity insurance v;ere recorded 

 on Schedules C and D, respectively. 



320 



