HW 3 Question 2

HW 3 Question 2

by Nicholas Arger -
Number of replies: 3

For this question:

A key question for this debate is: what is the trade-off between "true positives" and "false positives"?  That is, how much do you increase your chance of convicting someone who is innocent in order to convict more people who are guilty? This trade-off can be visualized with ROC curves.  Draw two hypothetical ROC curves[1] for this problem, labeling the axes and the points "10" and "12" on the curves for the number of jurors needed to convict. Make the first ROC curve one that would lead you unequivocally to support convictions with only 10 jurors voting guilty, and the other ROC curve one that would lead you unequivocally to oppose convictions with only 10 jurors voting guilty.  (Label the curves "Support" and "Oppose.") Explain your answer.

I am confused about what is being asked in this question. The question asks for two ROC curves for support the 10 juror system and one that is for opposing 10 system. How does these relate to the "10" and "12" curves? Are there really four curves here?


In reply to Nicholas Arger

Re: HW 3 Question 2

by Kerstin Kolodzie -

The question stem states:

"Simplify this problem by ignoring mistrials and considering only two possible verdicts: guilty and not guilty. In this analogy, a truly guilty defendant is like a patient with the disease, and an innocent defendant is like a patient without the disease, and a conviction by the jury is like a positive test."
 
To stay with this analogy, the number of jurors would be like different levels of a multilevel test.  Each ROC curve will have a point for 12 jurors and a point for 10 jurors.
The question asks for 2 ROC curves, one for supporting the initiative and one for opposing it.

I hope that helps.

Kerstin
In reply to Kerstin Kolodzie

Re: HW 3 Question 2

by Nicholas Arger -

Thank you, this makes sense now.


I am also confused with part 2.d.:

One reason why rational people might disagree on whether to support nonunanimous convictions is that their estimates of the slope of the ROC curve between the 10 and 12 juror  points differ.  Suppose two people agree completely on that.  What are at least two additional reasons why they might still disagree on whether to change the law?  


Is the question asking whether two people differ in opinion as to whether to change the law from the 10 juror system to the 12 juror system, or whether to change the law from requiring unananimous convictions to non-unanimous convictions in the setting of a 10 juror or 12 juror system? I do not understand where the non-unanimous part comes into play.

Thanks,

Nikko

In reply to Nicholas Arger

Re: HW 3 Question 2

by Thomas Newman -

Dear Nikko,

Thanks for asking.  In Oregon and Louisiana there are still 12 jurors.  It's just that only 10 of the 12 need to vote guilty to convict a defendant.

I've added "of the 12"  in the sentence below in the problem stem to try to clarify this.

"But in Oregon and Louisiana (as of 2018), only 10 of the 12 jurors are needed to convict for noncapital cases."

Tom