If you plot this model, you will see that there is one state with relatively large Cook's distance, Alaska. To exclude outliers from the analysis, tell S-PLUS to perform the model on only a subset of the data.
> summary(lm(SE70 ~ PI68 + Y69 + Locale, subset=-50))The above will produce a summary of a regression excluding Alaska. Another way to do the exact same thing is:
> summary(lm(SE70 ~ PI68 + Y69 + Locale, data=education[-50,]))The above means use
education, without row 50 but with all columns.
I think the first way is the ``proper'' thing to do. My understanding
is that the data field exists to allow modeling without
attaching to a data frame.