Hi Kaitlyn,
Margin of error is always critical value*standard error, just so you know that for future reference. This is another close call on whether to use z or t--we've got an approximate normal distribution, but low n, no known population standard deviation. The calculation isn't that cumbersome, so I'll provide both types.
First step is to compute standard error:
SE=s/sqrt(n)
s=standard deviation
n=sample size
Here:
s=9
n=22
SE=9/sqrt(22)
SE=1.92
Now, the z-critical value for 90% confidence is 1.645, which might be worth memorizing, just for future reference. Therefore,
z*=1.645
SE=1.92
MoEz=3.16
Standard error remains the same regardless of which critical value type we use, so:
SE=1.92
For t, we need degrees of freedom, which can be calculated by:
df=n-1
df=22-1
df=21
Go to the t-table, look up 21 degrees of freedom in the row, 90% confidence in the column. This gives:
t*=1.721
Now, we have:
t*=1.721
SE=1.92
MoEt= (1.721*1.92)
MoEt=3.30
Not too much difference in critical values. I hope this helps.