Question 1003719
You use a t-test because you don't know the population sd and assume the sample sd is a valid estimate of it.
The CI is 1.99 *s/sqrt(n)=1.99*4/9=0.884, and everything else is fine.  
The long form of the answer for a CI is we don't know what the true value is and we will never know, but we are highly confident (95%) that the interval we construct will contain that parameter.