Consider the associations between exercise (a binary variable, where 0 = exercise less than 3 times per week, and 1 = exercise 3 times per week or more) and glucose levels in mg/dL among women without diabetes. What are the general interpretations of the coefficient for exercise: in a simple linear model for glucose levels? in a multivariate linear regression model for glucose adjusting for all known confounders of the exercise association?