Question 3. Consider a laboratory experiment with 100 sets of triplicate response values for a group of known disease-negative (i.e., control) subjects (e.g., .05, .04, .65 for one of the 100 control subjects). The problem involves developing a cutoff value to classify test results (as either positive, negative, or indeterminate) for 50 sets of triplicate response values (for 50 subjects that need tested) based on the set of controls values.

a. Outline an approach for generating a cutoff value to classify the 50 unknown subjects. Be sure to list any underlying assumptions about the control subjects or test subjects in justifying your approach.

b. Suppose that negative response values are biologically impossible, but suppose several of the control response values are negative. Modify the approach in (1) to account for these negative values.

c. What would you consider the minimum number of control subjects required for generating a reliable cutoff? Why?