After performing a multiple regression analysis, however, the results indicated that the relationship between smoking and hypertension had all but disappeared and only age was a statistically significant predictor of blood pressure. The researcher then decided to see if using logistic regression, using a dichotomous version of the outcome of hypertension (Yes/No) and found a marginally significant association (p = .051) with cigarettes smoked adjusting for the same variables. 1. Why did the relationship between smoking and systolic blood pressure observed in simple linear regression likely disappear when using multiple linear regression analysis? 2. When presenting the results, should the researcher continue to emphasize that smoking has an independent effect on blood pressure based on the observed crude association using simple linear regression? 3. What are some potential problems in changing the analysis plan to use a dichotomous outcome for hypertension status (ves/no) after already analyzing the data using multiple linear regressi 4. When assessing the relationship between a lifestylerisk factor and a chronic disease outcome, why is considering both effect modification and confounding important