You can
put this solution on YOUR website! Using s, this is a t-statistic with df=18
t=(x bar-mean)/s/sqrt(n)
=-3/6.4/sqrt(19)
=-3*sqrt(19)/6.4
=-2.04. But this is doubled since a two way test, and the answer will be
0.05
with n=24 and df=23
t=-2.30
this will make the p-value 0.025
A one way test uses the difference only on one side to the distribution, but a two way requires both sides to be considered.