Suppose you are comparing the proportion of people who ...

[tsand2]

Suppose you are comparing the proportion of people who successfully stopped smoking on each of two different smoking cessation plans. You randomly assigned subjects to treatment, but you also collected data on each subject on a variety of measurement variables as well. How can this help you to verify that your randomization procedure worked?

Question 2

Journal articles often report means and standard deviations, and they also often report their results in the form (mean +/ standard deviation). Does this latter form represent a real confidence interval for the population mean? If yes, explain why. If not, tell what it does represent. (Assume the populations are bell-shaped.)

[kristenb95]

Answer to Question 1

YOU CAN LOOK AT THE MEANS AND STANDARD DEVIATIONS OF ALL THESE MEASUREMENT VARIABLES, AND COMPARE THEM FOR EACH OF THE TWO GROUPS STUDIED TO MAKE SURE THAT THESE VARIABLES WERE DISTRIBUTED EVENLY AMONGST THE GROUPS THROUGH YOUR RANDOMIZATION PROCESS.

Answer to Question 2

NO; THIS WOULD REPRESENT A RANGE WHERE 68 OF THE INDIVIDUALS IN THE POPULATION SHOULD FALL, ACCORDING THE EMPIRICAL RULE.

[tsand2]

Excellent

[ktidd]

Gracias!