Question

For the following below you should decide

a) What test would you use?

b) Why would you use this test?

1. You wonder if capital punishment has an effect on the number of murders committed. You obtain data from 23 states that have re-instated capital punishment. Your data list the ranking of each state with regard to number of murders per 100 people for the year prior to re-instating capital punishment and again 3 years later.

2. Denver Colorado claims that it has a significantly thinner population than the rest of the country (this is true I saw it in an add). If the average person in the U.S. has 23% body fat and the average person in Denver has 18% body fat (st dev 5%). Is their claim true?

3. You want to know which type of discipline works best on problem children. You select 45 children who have been identified as having behavior problems and select 15 for your control group, 15 to get positive reinforcement and 15 to get negative reinforcement. After 6 months you measure the number of behavioral problems each child exhibits in one week and calculate averages for each group. In analyzing your date you notice that all three group have slightly different variances (control, 23.3; positive reinforcement, 21.1; negative reinforcement, 19.9).

Answer 1

1. Spearman's correlation test because we have to find the effect of ranked data on each other.

2. One proportion z-test because we are comparing the proportions.

3. Kruskal Wallis test because we have to check if all three groups have slightly different medians.