Search   Memberlist   Usergroups
 Page 1 of 1 [2 Posts]
Author Message
Bill H
science forum beginner

Joined: 11 May 2006
Posts: 9

Posted: Thu May 11, 2006 2:24 pm    Post subject: probability of a false confession

http://www.stltoday.com/stltoday/news/stories.nsf/editorialcommentary/story/846ABB93AB66981A8625716400137951?OpenDocument

The local paper has an editorial by the lawyer Alan Hirsch on the
Moussaoui verdict, see link above, that contains the following
statement:

"Thanks to DNA testing, we have learned that innocent people are
convicted with greater frequency than anyone imagined. Since 1989, DNA
testing has exonerated 175 people convicted of crimes. Intriguingly,
one-fifth of them had confessed, including the five teenagers convicted

of assaulting the 'Central Park jogger.'

Spurred by the DNA exonerations, several scholars have sought to show
the prevalence of false confessions. Professors Richard Leo and Steven
Drizin recently published a study in the North Carolina Law Review
documenting 125 false confessions. Although it is impossible to
estimate the number or percentage of false confessions, experts believe

that known cases represent only the tip of the iceberg."

The phrase "impossible to estimate" grabbed my attention, and made me
wonder what would be required to calculate the probability of a false
confession, say among all those convicted of a crime. It seems like
one could use Bayes Thm as follows:

Out of all convictions, let
A = the event of exoneration by DNA testing
B = the event of confession to the crime

Corresponding probabilities,

Pr(B|A) = pr of confession given exoneration by DNA = .2 (from above
statement, setting aside sampling variability for the moment)

Pr(A) = ? (obtained if we know the number of total convictions during
time period, and given the number of exonerations by DNA from the
statement
above)

Pr(B) = ? (seems like somebody should have this data! out of total
convictions, how many had a confession, true or not?)

Pr(A|B) = pr of exoneration given a confession, ie. the confession was
false, which is the pr we want, and obtained from Bayes Thm as

Pr(A|B) = Pr(B|A)Pr(A)/ Pr(B)

Surely this is too naive. Comments?

posted this in the another newsgroup with no response, maybe it will
get some play here, note I changed some of the details after further
thought . . . probably need to account for the Pr(conviction is
challenged by DNA testing)
Dan Akers

Joined: 19 Jul 2005
Posts: 56

Posted: Fri May 12, 2006 2:44 am    Post subject: Re: probability of a false confession

Bill wrote;
"The local paper has an editorial by the lawyer Alan Hirsch on the
Moussaoui verdict, see link above, that contains the following
statement:
"Thanks to DNA testing, we have learned that innocent people are
convicted with greater frequency than anyone imagined. Since 1989, DNA
testing has exonerated 175 people convicted of crimes. Intriguingly,
one-fifth of them had confessed, including the five teenagers convicted
of assaulting the 'Central Park jogger.'
Spurred by the DNA exonerations, several scholars have sought to show
the prevalence of false confessions. Professors Richard Leo and Steven
Drizin recently published a study in the North Carolina Law Review
documenting 125 false confessions. Although it is impossible to estimate
the number or percentage of false confessions, experts believe
that known cases represent only the tip of the iceberg."
The phrase "impossible to estimate" grabbed my attention, and made me
wonder what would be required to calculate the probability of a false
confession, say among all those convicted of a crime."
______________________________________
Re:
It would seem to me, that one statistic you could glean from the article
is that about 20% of people who are wrongly convicted seem to have
inexplicably (although the article does site several probable reasons)
confessed to the crimes they apparently did not commit. Does this mean
that about 20% of all confessions are insincere? I don't think so; it's
simply a statement of how powerful a self-confession is in the
conviction process. More troubling, to some extent, is the 80% who
apparently pled not-guilty or no-contest and were none-the-less
convicted on what must have been unsubstantial or doubtful evidence. I
think to go further, you need more data; like how many total cases were
looked at to find the 175 errant convictions. The elusive statistic
seems to be an estimate of the portion of ALL convictions that are
errant. A statistic that would, no doubt, turn our justice system on
it's head, if it could be known with some certainty...

Dan Akers

 Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First
 Page 1 of 1 [2 Posts]
 The time now is Sat Sep 22, 2018 7:37 pm | All times are GMT
 Jump to: Select a forum-------------------Forum index|___Science and Technology    |___Math    |   |___Research    |   |___num-analysis    |   |___Symbolic    |   |___Combinatorics    |   |___Probability    |   |   |___Prediction    |   |       |   |___Undergraduate    |   |___Recreational    |       |___Physics    |   |___Research    |   |___New Theories    |   |___Acoustics    |   |___Electromagnetics    |   |___Strings    |   |___Particle    |   |___Fusion    |   |___Relativity    |       |___Chem    |   |___Analytical    |   |___Electrochem    |   |   |___Battery    |   |       |   |___Coatings    |       |___Engineering        |___Control        |___Mechanics        |___Chemical

 Topic Author Forum Replies Last Post Similar Topics Probability Question dumont Probability 0 Mon Oct 23, 2006 3:38 pm probability gorbag Probability 0 Mon Aug 14, 2006 11:06 pm Probability of attaining a minimum value when rolling dice nick@blackmarble.co.uk Math 16 Tue Jul 18, 2006 2:57 pm A mathematical game - probability questions Alun Math 8 Sun Jul 16, 2006 12:25 pm convergence of probability measures pkg Math 0 Sat Jul 15, 2006 7:15 pm