7.9. Using the computer or otherwise, generate a bivariate normal sample of 20 cases with parameters m1 m2 0, s11 s12

7.9. Using the computer or otherwise, generate a bivariate normal sample of 20 cases with parameters m1 ¼ m2 ¼ 0, s11 ¼ s12 ¼ 1, s22 ¼ 2, and delete values of Y2 so that Prðy2 missing j y1, y2) equals 0.2 if y1 < 0 and 0.8 if y1  0.

(a) Construct a test for whether the data are MCAR and carry out the test on your data set.

(b) Compute 95% confidence intervals for m2 using (i) the data before values were deleted; (ii) the complete cases; and (iii) the t-approximation in (2) of Table 7.2. Summarize the properties of these intervals for this missingdata mechanism.