Reference no: EM132847898

The athletic director at a large university asked her two assistants, Abbott and Costello, to assess the amount of sleep her student athletes are getting in a typical night relative to the sleep that non-athletes at the university are getting in a typical night. She provides Abbott and Costello with raw data of sleep patterns from a small sample (n=45) of student athletes and a small sample (n=48) of non-athletes.

Both Abbott and Costello receive the exact same raw data, but they decide to use different analytic strategies to make the comparison.

Abbott does a two-sample hypothesis test and finds that the average number of hours slept by student athletes is lower than the average number of hours slept by non-athletes, and the difference is statistically significant.

Costello, in contrast, uses a chi-square analysis to look at the association between athlete status (student-athlete versus non-athlete) and a four-category variable that he creates to contrast those getting no sleep on a typical night, those getting between 1 and 12 hours of sleep, and those getting more than 12 hours of sleep on a typical night. The association between these two variables turns out to be fairly weak in the sample and is not statistically significant.

-Why might the two different approaches for addressing the same question result in different answers?

-Which of the approaches do you prefer and why?