topic Re: difference between the parameters cibasic and ciquantnormal in proc univariate in Statistical Procedures

difference between the parameters cibasic and ciquantnormal in proc univariate

ipqnojug — Mon, 17 Sep 2018 06:41:41 GMT

In the sas example,

proc univariate data=Heights ciquantnormal(alpha=.1);
   var Height;
run;

calcualtes the confidence interval for quantiles, while

proc univariate data=Heights cibasic(alpha=.1);
   var Height;
run;

calcualtes the confidence interval for mean.

I find that the confidence interval for 50% quantile is the same as that of mean. I am a little confused by this.

Re: difference between the parameters cibasic and ciquantnormal in proc univariate

FreelanceReinh — Mon, 17 Sep 2018 10:29:59 GMT

Hello @ipqnojug and welcome to the SAS Support Communities!

Both CIBASIC and CIQUANTNORMAL (unlike CIQUANTDF) calculate confidence intervals based on the assumption that the data are normally distributed. Since mean and median (=50% quantile) of a normal distribution coincide, a confidence interval (CI) for the mean is also a CI for the median.