R Notes for Topic 29: Inference for Correlation and Regression

plot(UOPgpa$GPA ~ UOPgpa$Hours)

lm(UOPgpa$GPA ~ UOPgpa$Hours)

Let's add it to the scatterplot, too.

abline(lm(UOPgpa$GPA ~ UOPgpa$Hours))

r = cor(UOPgpa$GPA,UOPgpa$Hours)
r^2

I'll admit that I only know how to read the standard error from R, and not how to automatically get it into a variable.

We can get more information than you'll ever need about a linear model by asking for a summary of the linear model.

summary(lm(UOPgpa$GPA ~ UOPgpa$Hours))

The line of interest gives various information about the slope

> summary(lm(UOPgpa$GPA ~ UOPgpa$Hours))
...
             Estimate Std. Error t value Pr(>|t|)    
...
UOPgpa$Hours  0.08938    0.02771   3.226  0.00184 ** 
...

Although R just reported the test statistic, you should calculate it yourself just to make sure that R is correc.t

Although R just reported the p-value, you should see whether you get a similar value from the table.

Remember, you compute the confidence interval with b +/- t^*SE(b). You can look up the critical value in the table.

Recall that we use hist to produce the historgram.

hist(res)

We make normal probability plots and lines through them with

qqnorm(res, datax=T)
qqline(res, datax=T)

You should be able to figure this one out, since we've been doing scatterplots lately.