Description
Question Set 5
Imagine a “Sword in the Stone” has been placed under the Washington Square Arch, with an invitation to one who is worthy to pull it and take ownership (or something to that effect).
Ignoring the magical part, you set out to develop a measure of strength that could be used to evaluate the strength of incoming NYU students, so that they might attempt to pull the sword (bear with me here).
You begin with a dumbbell that has not been validated as a measure of strength and so we assume Sensitivity (Se) and Specificity (Sp) are both 0.50. You will assess strength among NYU students but have no basis to estimate the distribution of dumbbell lifting strength so we assume prevalence is 0.50.
Instructions: Adjust the R web-app according to the instructions below. [https://nyupsychometrics.shinyapps.io/RShinyApp]
a) You want to identify individuals who are exceptionally strong, but you’re at NYU and you estimate the prevalence of exceptional strength is only about 10% within the NYU student research pool.
1) Do you care more about the specificity or the sensitivity of the dumbbell, and why?
2) Adjust prevalence in the R-app to match the % of strong students. How is predictive power affected?
b) Your dumbbell is very heavy and it ends up very few students can lift it at all.
1) Do you think this is a useful outcome measure for research on strength? Why?
2) What adjustment(s) would you need for this to be more useful as an outcome?
