10172016 11:08 AM  last edited on 10172016 11:28 AM by Reeza
how to calculate Chisquared 2x2 test (Fischer’s exact test, twotailed) to get an p value using sas.
10172016 11:23 AM
Basic
Proc freq data=have;
tables var1*var2 / chisq fisher;
run;
10172016 12:12 PM
how to calculate for below data
(PERCENTAGES)  
A  10  0  0 
b  16  31.16556  29.49382 
c  12.89  3.96325  12.35552 
d  33.42764  37.29058  28.69669 
e  31.25851  23.25707  24.27262 
f  3.353243  4.323545  5.181347 
Total  100  100  100 
10172016 01:47 PM
ChiSquare is for count data, you don't appear to have count data.
10172016 02:04 PM
That is also NOT 2x2 data.
10172016 02:33 PM
2000  2001  2002  
(PERCENTAGES)  
A  0.104789  0  0 
B  29  31  29.9 
c  2.  3  12.4 
d  33  37  28.7 
e  31  23  24.3 
f  3  4  5.2 
Total  100  100  100 
i need to calculate 2000 vs 2001;2001vs 2002 ...
10172016 03:05 PM
Do you have actual counts or possibly even better data with one record per observation?
You can get significantly different results comparing percentages when the percentages are the same but the group counts differ significantly.
10172016 03:08 PM
this is the only data availble ....
10172016 03:26 PM
Then you need to explain your question. You need the N's even if it's just the total and you back calculate the N based on percentages.
If you don't have N's you can calculate the variance and you can't do any tests.
There's a difference between tossing two coins and getting two heads versus tossing a coin 100 times and getting 100 heads.
10192016 10:07 AM

20002005  20062010 
C  277  162 
d  898  448 
for this type of p value, you want to calculate a Chisquared 2x2 test (Fischer’s exact test, twotailed). So for example, for C, you would do…