Friday, July 23, 2010

On The Rate of Wrongful Conviction: Chapter 0.8

This is a bit embarrassing. I must break the monotonically increasing sequence of my posts regarding wrongful conviction rates. I have so far posted Chapter 0.027, Chapter 0.5, Chapter 1.0, and Chapter 1.4. I realized recently that I have carelessly lost track of a substantial survey conducted by Robert Ramsey. That survey allowed me to determine wrongful conviction rates based on the estimates of prosecutors, police, judges, and defense attorneys; four separate rates, four separate chapters, two of which will be numbered lower than the Poveda Number.

Nothing to do now but face up to it. On the upside, for the first time I present both a graph and a table in one of my posts. The excitement is almost too much to bear.

Chapter 0.8
The Prosecutor Number

What is it with those folks in Ohio who are interested in wrongful convictions. They don’t count exonerations or wrongful convictions. They don’t count total convictions, or have someone else do so for them. They don’t divide to determine wrongful conviction rates. They don’t make their own guess as to what the rate might be. Those people in Ohio prefer to ask other people in Ohio to guess for them.

Ronald Huff led the way with this approach in 1986, as discussed in Chapter 0.5. Robert Ramsey followed Huff’s footsteps in 2003. While attending the University of Cincinnati, he wrote a PhD dissertation entitled False Positives in the Criminal Justice Process -- An Analysis of Factors Associated with the Wrongful Conviction of the Innocent. I suggest he could have shortened the title by eliminating the last three words.

Ramsey later joined forces with James Frank to publish the results of his work in a 2007 issue of Crime Delinquency. He removed “of the Innocent” from the title, but included additional words to make up for the loss: “Wrongful Conviction: Perceptions of Criminal Justice Professionals Regarding the Frequency of Wrongful Conviction and the Extent of System Errors.”

Ramsey’s work was Huff’s work done large. Instead of asking 353 people to guess for him, Ramsey asked 1,500. Instead of receiving 229 responses, Ramsey received 798. Instead of having the guessers guess just once, he had them guess twice: once for their own jurisdiction, always in Ohio, and once for the U.S. as a whole. Instead of forcing the guessers to select from one of just four categories of wrongful conviction rates, Ramsey allowed them to select from any of the following 10 categories.

Less than 0.5%
5% to 1%
1% to 3%
4% to 5%
6% to 10%
11% to 15%
16% to 20%
21% to 25%
More than 25%

Ramsey also had an advantage in that DNA exonerations had become part of the public consciousness by the time of his survey. That would tend to increase the rates estimated by the Ohio law enforcement professionals.

Ramsey had sufficient responses to aggregate them into four groups: police, prosecutors, defense attorneys, and judges. Each group had a different estimate of the wrongful conviction rate, and each group will be discussed in a separate chapter. Not surprisingly, the prosecutors were the most conservative. They will therefore be discussed in this first of four chapters based on Ramsey’s survey.


Ramsey received relevant responses from 96 prosecutors. Those responses are categorized below:

Ramsey did not use his survey results to identify a single wrongful conviction rate. I will do that for him.

I will define the Prosecutor Number to be the median wrongful conviction rate estimated by the prosecutors in Ramsey’s survey. The median rate divides the prosecutor survey into two evenly-divided groups, those estimating rates higher than the median and those estimating rates lower than the median.  It is, I believe, the single number that best represents the prosecutors’ responses.

An easy way to determine the median is by graphing the data. I plotted the upper limit of the wrongful conviction rate category along the horizontal axis. I plotted the running percentage total of prosecutor responses along the vertical axis. I then examined the graph to determine the wrongful conviction percentage that split the prosecutors into evenly-divided “higher than” and “lower than” groups. Based on the resulting chart, presented below, I claim the single best number to represent the wrongful conviction rate estimated by the Ohio prosecutors responding to the Ramsey survey is 0.8%.

This 0.8% number is 60% higher than the Huff number of 0.5%, even though the Huff number included police, judges, and defense attorneys, all of whom tend to be less conservative in their estimates than the prosecutors. Had Huff isolated just the prosecutor number from his survey, I suspect it would have been around 0.25%.  I attribute the less conservative estimate of the Ramsey survey to two phenomenon. First, as mentioned previously, the Ramsey survey took place in a DNA exoneration world. Second, Ramsey was clever enough to ask for two responses: one for the respondent’s jurisdiction, and one for the U.S. as a whole. I believe that caused the respondents to be less defensive when estimating for the U.S. as a whole. 

To be clear, I used the estimate rates for the U.S. in the work above.

Consider finally the significance of the Prosecutor Number, conservative as it may be. If it is applicable to all people convicted, not just those who go to trial, it suggests that even prosecutors believe we have 20,000 people wrongfully incarcerated in this country today.

1. As of the publication of his article, Robert J. Ramsey was the Director of the Criminal Justice Program at Indiana University East. He was by that time Dr. Robert J. Ramsey. His dissertation apparently did the trick.

2. As of the publication of his article, James Frank had been the principal investigator for a number of policing-related research projects focusing on the understanding of police behavior at the street level. He received a JD from Ohio Northern University in 1977 and a PhD from the School of Criminal Justice at Michigan State University.

No comments:

Post a Comment