:3.  First-born children are less likely to become alcoholic
than are later-born children.  A psychologist wonders whether
this fact is also true of birth order within pairs of twins.  To
study this, he finds 8 pairs of twins (of suitable age) who
volunteer to provide him with data on their drinking habits. He
gathers this data for a month, and then computes the average
daily ingestion of alcohol for each twin.  Here are the results <p>

Alcohol ingested (in ounces)  <p>

1st born		2nd born  <p>

1		   4   		   5 
 	
      		 2		   0   		   3 
 	
      		 3		   1              2 
Twin pair numbers   4 	!    	   0              1 !I 
 	
 			 5   	   5              4 
 	
 			 6   	   3 		   5 
 	
 			 7   	   4 	        5 
 	
      		 8           1              0  <p>

(For instance, the 1st born member of twin pair 1 had a average
daily intake of 4 ounces of alcohol, while the 2nd born had an
average daily intake of 5 ounces.)  The null hypothesis is that,
on the average, birth order does not affect alcohol intake.  The
alternative hypothesis is that, on the average, the 1st born
member of a pair of twins tends to drink less than the 2nd born
member.  Suppose it is not true that the differences in alcohol
intake have a histogram which follows the normal curve.  Then a
statistical test based on Student's t-distribution is invalid. 
There is, however, a statistical test you can use. Its P-value in
this case is
 
 	
 a. the chance that in 8 tosses of a fair coin you get no     
 heads  <p>

b. the chance that in 8 tosses of a fair coin you get 1 or   
fewer heads  <p>

c. the chance that in 8 tosses of a fair coin you get 2 or   
fewer heads  <p>

d. the chance that in 8 tosses of a fair coin you get 3 or   
   fewer heads  <p>

e. the chance that in 8 tosses of a fair coin you get 4 or   
    fewer heads <p>