weighted sums. This method was preferred because weighting factors 
and weighted values were more easily understood and computations 
were easier than they would have been if ratios had been calculated 
for each State separately and these ratios weighted. The weighted 
State total is the number which would have been expected had request 
letters to hunters in each State been in proportion to the number of 
hunters in the State. Our ratios are the same as ratios of weighted 
mean numbers of wings per contacted hunter (weighted in proportion 
to stamp sales). 
The calculation of "weighting factors" is shown in Table 5. 
The proportion of stamps sold in a State was divided by the pro- 
portion of all hunter contacts that were made in the same State. 
The stamp sales in 1958-59 are believed to be a good measure of 
relative numbers of hunters from State to State, because the relative 
size of sales remain nearly constant from year to year even though 
the total numbers of stamps sold may fluctuate. This is shown in 
Table 29, where almost identical averages resulted from weighting 
the species-composition data by stamp sales records from two dif- 
ferent years. 
The next step in calculating the weighted ratio was to multiply 
the weighting factor from Table 5 by the number of wings received 
in the category concerned. For example, for Minnesota, the number 
of male mallard wings was multiplied by 0.70, and for Indiana, by 
2.68. These weighted numbers for the different States were added 
together and used as the numerator in calculating the weighted 
ratio: mallard males per female. 
An important assumption, however, is that the mean number of 
wings received per contact is related in the same manner in all 
States to the mean kill per stamp buyer. It is not necessary, nor 
is it necessarily to be expected, that the mean number of wings 
received per contact equals the mean kill, but it is necessary that 
bias be constant from State to State. There is a little information 
that bears on this point: 
(a) Response rates apparently are not uniform from State to 
State. The over-all response rate was remarkably uniform 
for the four States where hunters were identified by. number 
(Wisconsin, Michigan, Illinois, and Arkansas; see Table 6) 
even though response rate seemed correlated with success 
in a previous year. Later, however, the response rates 
were found to be much lower in certain southern States; 
this fact appeared when we assembled names of 1959-60 
respondents for use in 1960-61. To help understand response 
rates further, each hunter was assigned an individual number 
in the 1960-61 wing collection. 
ay 
