Significance level at 1% 5% 10%

[ivanpersie]

Hi there, i would like to have the different level of significance reported in my regression results. How do i do this stata? The second question is how do i report my regression results in Excel organised format.

[PaigeMiller]

But these are not Stata forums.

[ballardw]

Having no idea what "Excel organized format" might mean it is hard to answer that.

In SAS most procedures have an option of ALPHA= that control the significance level which usually defaults to 5%. How/where the option should be applied depends on the procedures. Often it goes on the procedure statement.

[ivanpersie]

Yes i would like to report significance levels at 0.01 0.05  and 0.1 in the Proc reg statement. Is that possible

[Rick_SAS · Accepted solution]

In statistics, you decide on the significance level BEFORE you run the analysis, and then you see if the data confirms or rejects various hypotheses at that significance level. It is considered "cheating" to run an analysis and then decide on the significance level after you see the results.

That said, in many procedures the p-values are reported as numbers like 0.0432.  That number is significant at the 0.05 level, but not significant at the 0.01 level.  For example, the parameter estimates table in PROC REG report p-values for the hypothesis that parameters are significantly different from 0.

@ivanpersie wrote:

Hi Rick, because am replicating a previous study and they  used t statistics. So i was wondering in SAS. how i tell the significance level for the t statistics rather than p value

---

Assuming that you know the degrees of freedom (DOF), you can use the QUANTILE function to invert the p-values to get the corresponding t-statistics:

data Signif;
dof = 20;
do alpha = 0.01, 0.05, 0.1;
   t = quantile("t", 1-alpha, dof);
   output;
end;
run;
proc print; run;

[ivanpersie]

Hi Rick, because am replicating a previous study and they  used t statistics. So i was wondering in SAS. how i tell the significance level for the t statistics rather than p value

[Rick_SAS]

Assuming that you know the degrees of freedom (DOF), you can use the QUANTILE function to invert the p-values to get the corresponding t-statistics:

data Signif;

dof = 20;

do alpha = 0.01, 0.05, 0.1;

   t = quantile("t", 1-alpha, dof);

   output;

end;

run;

proc print; run;

[Reeza]

Your significance levels are 0.01, 0.05, and 0.1. Your p-value is what you report.

IN comparing the p-value to a significane level you can determine if a result is significant.

As Rick explained above, the significance level is chosen ahead of time. 0.05 is commonly used in medicine, while 0.2 might be great in marketing.