Dr. Charles Ely 
March 2 k , 1965 
(k) Each stratification should approach ten-day counts per month and 
fifty indiYiduala. 5^ 
O) Wfctt data you can then test nesting birds against non-nesters, 
8°°d weather against bad and different periods of the year against 
each other. Any other supposed variable can be stratified out in a 
like manner. 
On the page labeled No. 3 I've run a test on the percentage of 
Individurla avflying on the island each day that were not there at the 
earliest hour for which a count was made that day. The test then 
determines the homogeneity of this proportion for the six dayB sampled. 
The results showed a significant difference for this proportion between 
the various days, with the possible exception of the 11th and 23 rd 
which irere not tested per se. 
' 
I have not fooled with the data any further, but this test can be 
run for any particular characteristic in the population. The following 
formula will apply to teat the proportion of individuals exhibiting a 
character1st1c for any two samples (i.e. two different days)* 
^7°* 
m 
P 
I ' v. 
■ ,V\. 
*rts 
<1 * 
n l* 
«2* 
pq. TJT ♦ iff 
ffl nf; 
total percentage of occurrence 
(i.e* percentage of occurrence In 
combined sample) 
1-p. 
number in first sample 
number in second sample 
If the actual difference in percentages is greater than 2 0 / 0 , 
they are significantly different* 
This may not be very clear. If this does not tell you what you want 
to know, or if you want a more detailed explanation or analysis of any 
part of the data, please let me know. 
■ V 
f Sincerely yours. 
V'. v» 
V’ 1 1 
,fT; 
.. * h r Tc.\ •*-> i -• -/ii**". 1 
ip 
I V*. . k > 
' 
■ S’ltt.. 
• r 4 >; 
’v- *. '■ 
Hiilip N. Lehnex 
Research Assistant, Pacific Project 
Division of Birds 
