To investigate whether a certain medication increases the reaction time, 20 people take a reaction test, where 10 people receive the medication beforehand and the remaining 10 form a control group. The following reaction times are measured (in seconds):

Use a U-test of level ˛ D 0:05 to test the null hypothesis that the reaction time is not influenced against the alternative of a prolonged reaction time. Apply

(a) the exact U-distribution,

(b) the normal approximation.

Experimental group Control group .83 .66 .94 .78 .81 .60 .88 .90 .79 .86 .64 .70 .69 .80 .71 .82 .62 .91 .59 .63