Because the relationship between Breathalyzer-determined and selfreported blood alcohol levels in Problem 4.8 appeared to change above a certain level, we reanalyzed the data after throwing out the...

Because the relationship between Breathalyzer-determined and selfreported blood alcohol levels in Problem 4.8 appeared to change above a certain level, we reanalyzed the data after throwing out the points at high levels (Problem 4.10). Another approach to this problem is to consider that the data define two different linear segments: one that describes the region of self-reported alcohol over a range of levels of 0–.11 (determined by examining a plot of the data) and the other over the range of values greater than .11. Using only one regression equation fit these two line segments. (Hint: You will need to use a dummy variable in a new way.) What are the slopes and intercepts of these two segments?

Problem 4.10

In Problem 4.8, the data seem to reflect a threshold of about .10 blood alcohol in which a relatively good correspondence between Breathalyzer-determined blood alcohol and self-reporting of consumption breaks down. Reanalyze these data after omitting all data points with Breathalyzer-determined blood alcohol levels greater than or equal to .10. Now, how well does self-reporting reflect actual blood alcohol?

Problem 4.8

Driving while intoxicated is a pressing problem. Attempts to deal with this problem through prevention programs in which social drinkers learn to recognize levels of alcohol consumption beyond which they should not drive require that people be able to accurately judge their own alcohol consumption immediately prior to deciding whether or not to drive. To see if drinkers could actually judge their alcohol consumption, Meier and coworkers* administered a Breathalyzer test to measure blood alcohol and administered a short questionnaire to record recall of alcohol consumption, which was then used to estimate blood alcohol level based on the respondent’s age, gender, and reported consumption (the data are in Table D-17, Appendix D). How well does drinker recall of alcohol consumption reflect actual blood alcohol level as determined by the Breathalyzer?

Table D-17