About testing, the method and theory are complete in classical statistics for those data and parameters in point form. It is not easy to do the testing work if our sample data are intervals. There are problems unsolved such as (1) What is the distribution of interval-valued random variables. (2) How to test population mean which is an interval? (3) How to test two populations' variance equal or not? In this paper we try to extend classical testing methods such t-test and F-test with extended concept to test population mean and population variance which are intervals.