> I know from previous research that the RFI for each study is 5, 5, 10 and 4.
Well, the most likely explanation is that I misunderstood what you are asking for. So you need to clarify the problem. Please explain the above statement by constructing an example for Study A that runs Fisher's Exact Test and results in a significant p-value. In my previous response, I interpreted '5' for Study A to mean the following frequency table and analysis:
data A;
input Group Event Count;
datalines;
1 1 5
1 0 146
2 1 1
2 0 148
;
proc freq data=A;
tables Event*Group / norow nocol nopercent;
exact fisher;
weight Count;
run;
Please explain what table you want to analyze for Study A when '5' is used. Also, explain the logic by which you obtained the table. Then do the same explanation for Study C, which uses '10'.
Rick,
This is a method to review already published data. Below are the datalines for this table. The p-value for this dichotomous outcome is >0.05. The goal is to see the minimum number of nonevents that need to change to an event for each outcome measure in order for the nonsignificant p-value, as measured by Fisher's exact test, to become significant. Alpha level 0.05. Based on previous analysis, I know the minimum number of events needed is 5 for this study and want to determine a way to replicate the analysis in SAS.
data want;
input study $ group $ outcome $;
datalines;
A Exp Yes
A Exp Yes
A Exp Yes
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Exp No
A Con Yes
A Con Yes
A Con Yes
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
A Con No
;
run;
proc freq data= want;
table group*outcome /chisq fisher;
run;
Please explain the statement "Based on previous analysis, I know the minimum number of events needed is 5 for this study."
When I run your DATA step, I see it is equivalent to the following shorter program:
data Have;
length Study $1 Group $3 Outcome $3;
input study $ group $ outcome $ Count;
datalines;
A Con No 146
A Con Yes 3
A Exp No 148
A Exp Yes 3
;
proc freq data= Have;
table group*outcome /chisq fisher norow nocol nopercent;
weight count;
run;
You say that you can change 5 nonevents to events and get a significant p-value. Please show how. Here is a new data set in which I have changed 5 nonevents in the experimental group to events. Upon rerunning the analysis, I do not get a significant p-value:
data New;
length Study $1 Group $3 Outcome $3;
input study $ group $ outcome $ Count;
datalines;
A Con No 146
A Con Yes 3
A Exp No 143
A Exp Yes 8
;
proc freq data= New;
table group*outcome /chisq fisher norow nocol nopercent;
weight count;
exact fisher;
run;
If I change 5 observations from Yes to No in the control group, I also do not get a significant p-value. Show me what you do to the original data to get a significant p-value.
SAS Innovate 2025 is scheduled for May 6-9 in Orlando, FL. Sign up to be first to learn about the agenda and registration!
Learn how use the CAT functions in SAS to join values from multiple variables into a single value.
Find more tutorials on the SAS Users YouTube channel.
Ready to level-up your skills? Choose your own adventure.