NOTE: a cribsheet is provided as part of the exhibits with all exams. There is one typo--a missing over-bar on X in the formula for the confidence interval for mean prediction. This will be fixed in the version distributed during the exam.

NOTE: Prior to the Fall of 1997, the course used Version 7.0 of SHAZAM. In that version, the format of the p-values for hypothesis testing was different:

SHAZAM used to report the percentile of the t-distribution represented by the value of the t-test statistic in the p-value column of the regression output. This meant that p-values ranged from zero (if the t-ratio was a large negative number) up to 1.0000 (if the t-ratio was a large positive number). You needed a p-value less than 0.025 or greater than 0.975 in order to reject the zero hypothesis about the corresponding parameter at the 5% level of significance.

SHAZAM now reports something more intuitive, namely the probability of occurrence of a t-test statistic value larger in absolute value than the number appearing in the t-ratio column if the zero hypothesis is true. This is the "area in the two tails" cut off by the t-ratio value and its negative. Now, if the p-value is smaller than 0.05, it means that the zero hypothesis about the corresponding parameter is rejected at the 5% level of significance.