real p values from ttest
in [SAS]
|
From: Robin R High on 20 Jan 2010 15:15 With ODS there is a very simple way, but as this topic has come up before, my first question (never answered) is Why? and second, With regards to HO, what does a pvalue represent? Robin High UNMC From: df ss <tggsun(a)YAHOO.COM> To: SAS-L(a)LISTSERV.UGA.EDU Date: 01/20/2010 02:09 PM Subject: real p values from ttest Sent by: "SAS(r) Discussion" <SAS-L(a)LISTSERV.UGA.EDU> Hi SASor, I run ttest and get the results. but some of p value from the results are <.0001. Is there a way to show the real p values (such as 0.00000000000546) to substitute for <.0001? Thank you, ss df
From: "Keintz, H. Mark" on 20 Jan 2010 15:33 Why would you believe a p value=0.00000000000546? But maybe what you are trying to do is rank a set of test results by p-value, and you have a lot of tests with p < 0.0001, no? If so, then I ask what sort of data and statistical tests would provide a meaningful ranking of such results? Regards, Mark > -----Original Message----- > From: SAS(r) Discussion [mailto:SAS-L(a)LISTSERV.UGA.EDU] On Behalf Of df > ss > Sent: Wednesday, January 20, 2010 3:07 PM > To: SAS-L(a)LISTSERV.UGA.EDU > Subject: real p values from ttest > > Hi SASor, > I run ttest and get the results. but some of p value from the results > are <.0001. Is there a way to show the real p values (such as > 0.00000000000546) to substitute for <.0001? > Thank you, > ss df > > >
From: Mary on 21 Jan 2010 10:07 This is done in genetics, but significance is then considered to be very conservative, such as applying a bonferoni correction like 10 -7. To answer the user's question, if you save the results using ODS, then you'll have the entire p-value and not just the <.01. Such as: ods output tablename=tablename; proc procedure; {steps for procedure} run; Then your results will be in the tablename to the left of the equal sign. I posted a complete example of how to do this about a year ago; I'll look for it or you can try to find it in the archives by searching mlhoward(a)avalon.net (but I had a lot of posts last year!) -Mary --- rhigh(a)UNMC.EDU wrote: From: Robin R High <rhigh(a)UNMC.EDU> To: SAS-L(a)LISTSERV.UGA.EDU Subject: Re: real p values from ttest Date: Wed, 20 Jan 2010 14:15:55 -0600 With ODS there is a very simple way, but as this topic has come up before, my first question (never answered) is Why? and second, With regards to HO, what does a pvalue represent? Robin High UNMC From: df ss <tggsun(a)YAHOO.COM> To: SAS-L(a)LISTSERV.UGA.EDU Date: 01/20/2010 02:09 PM Subject: real p values from ttest Sent by: "SAS(r) Discussion" <SAS-L(a)LISTSERV.UGA.EDU> Hi SASor, I run ttest and get the results. but some of p value from the results are <.0001. Is there a way to show the real p values (such as 0.00000000000546) to substitute for <.0001? Thank you, ss df
From: Mary on 21 Jan 2010 10:14 I found the post in the archive that saved ttest results to ODS; here's the post again (this was March, 2009) -Mary Eduardo, Here's a more complete example: data test; infile cards missover; input d1 d2 v1 v2; cards; 1 0 30 20 1 0 29 19 1 0 28 18 0 1 40 20 0 1 41 41 0 1 43 33 ; run; data dependent_vars; informat varname $30.; infile cards missover; input varname; obsnum + 1; cards; d1 d2 ; run; ods output position=position; proc contents varnum data=test; run; proc sql noprint; create table independent_vars as select variable as varname from position where variable not in ('d1','d2'); quit; data independent_vars; set independent_vars; obsnum + 1; run; %macro get_results(dependent_var, independent_var); ods trace on; proc datasets library=work; delete ttests equality; quit; ods output TTests=ttests Equality=equality; proc ttest data=test; class &dependent_var; var &independent_var; run; %let probt=.; proc sql noprint; select Probt into :probt from ttests where variances='Equal'; quit; %put probt= &probt; %let probf=.; proc sql noprint; select Probf into :probf from equality where method='Folded F'; quit; %put probf= &probf; data results; dependent_var="&dependent_var"; independent_var="&independent_var"; probt=&probt; probf=&probf; run; data allset; set allset results; run; %mend get_results; %macro do_calls; data allset; stop; run; proc sql noprint; select count(*) as dep_count into :dependant_count from dependent_vars; quit; %put &dependant_count; proc sql noprint; select count(*) as dep_count into :independant_count from independent_vars; quit; %put &independant_count; %do i=1 %to &dependant_count; %do j=1 %to &independant_count; proc sql noprint; select varname into :dependent_var from dependent_vars where obsnum =&i; quit; %put &dependent_var; proc sql noprint; select varname into :independent_var from independent_vars where obsnum =&j; quit; %put &independent_var; %get_results(&dependent_var,&independent_var); %end; %end; %mend do_calls; %do_calls; -Mary ----- Original Message ----- From: "Eduardo Galvan" <EGalvan(a)SURVEYSCIENCES.COM> To: <SAS-L(a)LISTSERV.UGA.EDU> Sent: Monday, March 30, 2009 10:13 AM Subject: Running/Outputting t-test results repeatedly Hi, Sorry for the repost, but I'm hoping somebody can help me with this. I have 10 dependent variables (e.g., dv1, dv2,....dv10) that I want run t-tests on and output the mean, standard deviation, and p-value to a data file. The catch is that I have over 2,000 variables that I want to use as the independent variable when I am running these tests. In other words, I need to run 2,000 t-tests on dv1, 2,000 t-tests on dv2, etc. Suggestions are appreciated. Eduardo --- mlhoward(a)avalon.net wrote: From: Mary <mlhoward(a)avalon.net> To: <tggsun(a)YAHOO.COM> Cc: <SAS-L(a)LISTSERV.UGA.EDU> Subject: Re: real p values from ttest Date: Thu, 21 Jan 2010 07:07:51 -0800 This is done in genetics, but significance is then considered to be very conservative, such as applying a bonferoni correction like 10 -7. To answer the user's question, if you save the results using ODS, then you'll have the entire p-value and not just the <.01. Such as: ods output tablename=tablename; proc procedure; {steps for procedure} run; Then your results will be in the tablename to the left of the equal sign. I posted a complete example of how to do this about a year ago; I'll look for it or you can try to find it in the archives by searching mlhoward(a)avalon.net (but I had a lot of posts last year!) -Mary --- rhigh(a)UNMC.EDU wrote: From: Robin R High <rhigh(a)UNMC.EDU> To: SAS-L(a)LISTSERV.UGA.EDU Subject: Re: real p values from ttest Date: Wed, 20 Jan 2010 14:15:55 -0600 With ODS there is a very simple way, but as this topic has come up before, my first question (never answered) is Why? and second, With regards to HO, what does a pvalue represent? Robin High UNMC From: df ss <tggsun(a)YAHOO.COM> To: SAS-L(a)LISTSERV.UGA.EDU Date: 01/20/2010 02:09 PM Subject: real p values from ttest Sent by: "SAS(r) Discussion" <SAS-L(a)LISTSERV.UGA.EDU> Hi SASor, I run ttest and get the results. but some of p value from the results are <.0001. Is there a way to show the real p values (such as 0.00000000000546) to substitute for <.0001? Thank you, ss df
|
Pages: 1 Prev: ODS Html Style Graph with SGPLOT in Excel Sheet Next: not enough memory |