12 FISH AND GAME COMMISSION 



To overcome these defects a new system was tried in 1939 and is 

 now in use. A random sample list of approximately 10 per cent of the 

 anglers was built up throughout 1939 as the license stubs reached the 

 statistical department, and in Jauary of 1940 questionnaires were mailed 

 to these names. The angler was thus enabled to make his report at leisure 

 and reasonably soon after the close of the season ; and inspection of the 

 returns shows great improvement over the earlier ones. Further, the final 

 consolidated report could be completed at least a year sooner than was 

 possible under the old system. 



It should be noted that the voluntary report continues to be basic. 

 This means that not all anglers in the sample make reports, and that 

 estimates of the total catch of any kind of fish by all the anglers have had 

 to be based on the catches reported by those anglers who do make returns. 

 The proportion of report returns, under all systems so far tried, has been 

 in the neighborhood of 30 per cent. Even under the postal card question- 

 naire, this has involved close to 10,000 units, a number which would be 

 considered statistically adequate to serve as a basis of estimate provided 

 it constituted a true random sample, that is, provided the reporting 

 anglers were truly representative of the whole angling population. 

 Unfortunately, there is every reason to believe that this is not the case. 

 The consideration of only one factor serves to demonstrate this : men who 

 catch no fish obviously are less likely to report than those who do, and 

 the proportion of such men is therefore greater among the noureporters 

 than among the reporters. The game records, where the deer tags furnish 

 an absolute check unequaled in the fish records, have clearly demonstrated 

 the truth of this general proposition. 



In an effort to shed light on the darkness surrounding the activities 

 of these noureporters, a ''second call" was sent out in 1938 to a sample 

 of the licensees who had failed to report their 1937 catches on their 1938 

 license applications. In this case only 18 per cent returned catch records, 

 leaving again a large blind spot. The principal piece of information 

 gleaned from this work was that at least 12 per cent of all licensees catch 

 no fish, of which about half do not fish at all, while the other half fish but 

 catch nothing. This figure has been adopted as a basic factor to use in 

 working up all estimates of total catches. 



In spite of all questions as to the validity of the sample and as to 

 the reliability of the individual reports, information of great value is 

 none the less furnished by the catch records. The distribution of the 

 angling licensees by county of residence is factual matter to which no 

 shadow of question attaches; the distribution of their fishing effort by 

 species and by county of catch, while not factual, may be looked upon as 

 liaving considerable reliability. 



As for the estimates of total catch, it should be stated that these are 

 based on two general assumptions; first, that catch reporters and non- 

 reporters fish for each species in similar proportion ; and second, that the 

 average catch of the noureporters, after making allowance for the 12 

 per cent of all anglers who catch no fish, is the same as the average catch 

 of reporters. While there may be and probably are fallacies in these 

 assumptions v/hich make for errors in the absolute values of the estimates, 

 it seems none the less probable that the comparative values of the esti- 

 mates from year to year are highly significant, and that they do present a 

 fairly accurate picture of the yearly fluctuations in the total catch of any 



