Chapter 9, Problem 16E

Candy bars per serving: calories A student collected nutrition data about candy bars by reading the labels in a supermarket. Because candy bars have different “serving sizes,” the data are given as values per serving. Here is a regression predicting calories from the sugar (g/serving). (Five sugar-free candy bars have been omitted from this analysis.)

Response variable is: Calories/serving

R-squared = 37.4% s = 0.3443

Variable	Coefficient
Intercept	6.92032
Sugar/serving	–3.91491

1. a) What is a proper interpretation of the coefficient of Sugar/serving?

Here is another regression, now with protein (g/serving) as a second predictor:

  Response variable is: Calories/serving

  R-squared = 44.5% s = 0.3312

Variable	Coefficient
Intercept	5.61689
Sugar/serving	–2.18057
Protein (g/serving)	6.04437

1. b) What is a proper interpretation of the coefficient of Sugar/serving in this regression?
2. c) How can you account for the change in Sugar coefficient values?