For example, using the **mpg** data file, say we wish to test whether the average highway mpg (**hwy**) significantly differs from 25. We can do this as shown below. Note all my codes have been test under SAS version 9.3.

proc ttest data=mpg h0=25; var hwy; run;

The mean of the variable **hwy** for this particular sample of cars is 23.44, which is statistically significantly different from the test value of 25. We would conclude that this group of cars has a significantly different mean on the highway mpg from 25.

If you believe that a one-sided test is more reasonable, you can specifies an lower or upper one-sided tests by adding **sides=L(U) option**, in which the alternative hypothesis indicates a mean smaller or greater than the null value.

proc ttest data=mpg h0=23 sides=U; var hwy; run;

Since the P-value for this one-side test is not <.05, we would conclude that this group of cars doesn’t have a significantly higher mean on the highway mpg than 23.

*See Also:*