Lab Agenda, January 14, 2004

  1. Make sure you know how to set up "Chapters" in Splus corresponding to each dataset you use for homework.

  2. TAs will assist with Splus commands related to Homework 1. Note that all directions are in the homework itself.

  3. Finding quantiles and probabilities in Splus using the command line. For homeworks, you need to use Splus to get exact quantiles and probabilities of the t and Normal distributions. For exams, you should know how to use the tables at the back of Statistical Sleuth.

  4. Open this plot of a regression line with residual vs. fitted plot. Use this applet to draw a round cloud of points near the origin that produces a slope close to zero. Now add a single point somewhere else on the graph. How can you add a single point that does *not* change the slope of the regression line? (a point with low leverage)

  5. What is the difference between the SD line and the regression line? This applet will help explain the "regression effect" or "regression fallacy". Select Data: "gmat"to plot Y=verbal scores vs. X=quantitative scores. Click to plot the SD line (click "No SD Line") and the regression line (click "No Regression Line"). The SD line is the line passing through the point (35, 35.09) (which is the (mean quant, mean verbal) point) with slope equal to sy/sx. The regression line also passes through the point.

    Why is the SD line different from the regression line? Why doesn't the regression line go through the middle of the football shaped cloud of points like the SD line? (Hint: Recall the formula for the slope as a function of the correlation coefficient and explain.) If I score one SD above the mean on the quantitative part, is it likely that I will also score one SD above the mean on the verbal part? Why or why not?