Stat 10, UCLA
Chapter 9 Solutions

    1. False - it's the other way round. Below-average values of the dependent variable tend to be associated with above-average values of the independent variable.
    2. False. For example, height in inches is usually less than weight in pounds, but the correlation coefficient is positive.

    1. Height at 16 and 18; it is easier to predict 2 years ahead than 14.
    2. Height: environment, lifestyle, etc.. influence weight more than height.
    3. Age 4

  3. False - these are cross-sectional data not longitudinal data. Younger people were born more recently and education levels have been going up over time.
    1. False: r=-0.57. The two sections that liked their TA the best did the worst (Sections C and I)
    2. True: there is no pattern in the scatter diagram (r=.12)
    3. False: there is a moderate, positive association (r=0.46)

    1. True: The test takers are a self-selected group, and good students are more likely to take the test. Hence, if a larger percentage of high school graduates take the test, the average skill level goes down.
    2. False: The average in New-York is lower; but the reason may be that in New York, a higher percentage of students take the test.

  6. Quite a bit less because this is an ecological correlation.

    1. Education is measured in whole years.
    2. Some dots correspond to more than one couple.
    3. A (iv), B (iii), C (i).

http://www.stat.ucla.edu/~dinov