|
Article Excerpt
That would make the error rate [in felony convictions] .027 percent--or, to put it another way, a success rate of 99.973 percent.
--Justice Antonin Scalia. concurring in Kansas v. Marsh, June 26, 2006 (quoting Joshua Marquis). (1)
I. INTRODUCTION
The news about the astounding accuracy of felony convictions in the United States, delivered by Justice Scalia and Joshua Marquis in the passage set out epigrammatically above, would be cause for rejoicing if it were true. Imagine. Only 27 factually wrong felony convictions out of every 100,000! Unfortunately, it is not true, as the empirical data analyzed in Part III of this article show. Part IV then deals with the impact of an empirically derived factual innocence rate (2) for at least one important category of real world criminal convictions (capital rape-murders in the 1980s) on various participants in current debates on both the death penalty and the problem of convicting the factually innocent generally. Part V examines the problem of generalizing the factual innocence rate from capital rape-murder convictions in the 1980s to other times and other crimes. Part VI provides a lens through which to view the moral and policy implications of such a factual innocence rate. But first, some historical context:
II. PALEYITES AND ROMILLISTS
People who think about the problem of wrongful conviction often fall into two camps, which we might label Paleyites and Romillists. Paleyites, whom I have named after the early exponent of this position, the 18th-century proto-utilitarian the Rev. William Paley, believe that, even though it is wrong to convict an innocent person, such convictions not only are inevitable in a human system, but represent the necessary social price of maintaining sufficient criminal law enforcement to provide an appropriate level of security for the public in general. Hence, one should not be moved by the prospect of wrongful conviction to take actions that would reduce such convictions, no matter how common, at the cost of reducing convictions of the guilty to a dysfunctional level. (3) Paleyites tend to be conservative, in the sense that any changes to current ways of conducting the criminal justice process, proposed for their supposed effect on protecting the innocent, will be presumed so counterproductive in their effect on convicting the guilty that they will be opposed.
Romillists, whom I have named after the early 19th-century reformist Sir Samuel Romilly, have such a horror of convicting the innocent that they are willing to propose many changes to whatever system exists, on the ground that such changes in our way of criminal law enforcement will better protect the innocent. (4) In so doing, it may be that some of the proposals might make the conviction of the truly guilty more difficult, perhaps significantly so. Whatever the actual effect, the Paleyites can be counted on to find the potential effect abhorrent, and to label the proponents "soft-headed sentimentalists" or some similar characterization, while the Romillists in turn will label the Paleyites hard-hearted troglodytes, indifferent to the plight of the convicted innocent, with knee jerk opposition to reform.
What neither side has a good handle on, however, is the magnitude of the problem of factually wrongful conviction and wrongful acquittal. Partly this is due to the inherent difficulty of establishing the ground truth of factual guilt or innocence better than the trials (or plea bargains) that resulted originally in acquittals or convictions. But, at least with regard to convictions, it is also partly due to the fact the legal system is structured to operate as if it were controlled by Paleyites, whatever the personal beliefs of individual participants. This is the result of rather extreme doctrines intended to uphold the integrity and finality of the results of criminal trials. (5) Nevertheless, both post-conviction legal doctrines and those who administer them, prosecutors and judges alike, resist new evidence of innocence to such a degree that it often passes the bounds of rationality. (6) And what but the word "concealment," albeit in the name of protecting the public legitimacy of the system, can explain the efforts undertaken to oppose DNA testing in regard to those already executed, where such DNA testing would conclusively establish guilt or innocence in fact. (7)
Traditionally, a certain stripe of Paleyite has also denied that wrongful convictions happen at all, or, that if they happen, they happen so rarely that worrying about them is like worrying about being struck by a meteorite. (8) The reasons assigned for this assumed near-perfection in regard to false-positive error have generally been the numerous layers of filtration involved in the pre-trial system, and the general fairness of the adversary trial itself, with its formal requirement that the prosecution prove guilt beyond a reasonable doubt. (9)
Such a position is very difficult to take in the era of DNA exonerations. Difficult--but not impossible. As one can see from the epigram at the beginning of this Article, Paleyites such as Justice Scalia and Joshua Marquis still speculate about, and embrace, ludicrously low wrongful conviction rates. However, such speculation has become both obsolete and untenable, since, as I propose to demonstrate, the data and the elementary statistical tools necessary to arrive at a reliable minimum rate of factually wrongful conviction, at least in a certain significant subset of cases, are actually to hand. And from this specific minimum innocence rate other inferences may defensibly be drawn about the problem of factually wrongful conviction. Once Paleyites and Romillists are forced to agree on at least a partial description of the problem of factually wrongful conviction, they can then proceed to develop and set out informed normative responses to the empirical reality.
III. AN EMPIRICALLY JUSTIFIED FACTUAL WRONGFUL CONVICTION RATE: THE CASE OF CAPITAL RAPE-MURDERS IN THE 1980S
In order to derive a minimum factual wrongful conviction rate (a factual innocence rate), we must, of course, have a numerator and a denominator. The denominator would represent a certain reference set of convictions, and the numerator would represent the number of factually wrongful convictions in the reference set. We might look for our numerator in the number of exonerations that have taken place over a certain period of time, whether based on DNA evidence or not. (10) I have chosen, however, to include only DNA exonerations as part of a numerator, in order to avoid the epistemic problems that could arise in regard to any rationally debatable exonerations, (11) since it is easiest to establish DNA exonerations as being close to indisputable cases of factually wrongful conviction. (12)
So let us look for our numerator somewhere in the statistical pool provided by the DNA exonerations, and then define the boundaries of the universe of cases these exonerations represent, in order to find a denominator and establish a minimum rate of factual innocence for that universe of cases. Then we can examine the question of what the DNA cases can tell us in general about rates of wrongful conviction and factual innocence.
To obtain a proper sample of DNA exonerations to work with, one must understand that the cases in which DNA exonerations occur are by definition not a random sample of all cases of criminal conviction. Virtually all such exonerations occur in cases of serious felony, often capital felony, in which a trial resulted in a conviction. The DNA exonerations can usefully be divided into four groups: capital cases, (13) noncapital homicide cases, non-capital rape/sexual assault cases, and others. The most obvious group to examine in searching for a denominator is the capital cases. This group consists of an externally defined set of capital cases of finite and known number in the United States during the period of time since the reestablishment of the death penalty in 1976 from which such exonerations are drawn. These would be specifically the capital sentences imposed from the date of the first such conviction that finally culminated in a DNA exoneration, to the date of the latest trial of the case finally culminating in the capital DNA exoneration, roughly 1977 to 1999. (14) There are fourteen capital-case DNA exonerations so far in cases tried from 1977 to 1999. During that same period of time, 5968 capital sentences were imposed. (15) These figures give an absolute minimum factual error rate for capital sentences imposed during that period of .23%.
Whether the imposition of a death sentence on a factually innocent person two or three times out of every thousand impositions of capital punishment is too high a rate is a heavy question of morality and policy. But of course this percentage does not represent the actual rate of factually wrongful conviction. In fact, it is clearly grossly understated, because we are using the wrong denominator. (16) The choice of the right denominator is what makes it empirically defensible to derive a factual innocence rate from the DNA exonerations. We must therefore carefully define the boundaries of the universe of cases represented by the group of DNA exonerations chosen. (17)
The DNA exonerations can only occur in the subset of capital convictions in which it is reasonable to believe that bodily sources of DNA might have been left in such a way as to provide the basis for including or excluding a defendant as the possible perpetrator. Generally, in capital case exonerations, this has meant what can be called "rape-murders," generally homicides where the victim is raped, then killed. In fact, thirteen of the fourteen DNA exonerations in capital cases involved rape-murders. (18)
Looking at these thirteen cases, two important points emerge about the window that the DNA exonerations open on the problem of wrongful conviction rates in general. First, that window is closing. As DNA technology has become more sensitive, more accurate, and more generally available and understood, the number of cases in which such testing is not done for the original trial shrinks. This is, of course, a great net benefit for the criminal justice system. Those who are guilty in the relatively small percentage of cases where DNA evidence is available will be convicted with much greater confidence, and those who can be exonerated by DNA will be exonerated before or at trial. But it is extremely important to remember that the conditions that cause wrongful conviction in non-DNA cases--the vast majority of cases--remain unaffected by this development. (19) We must use the post-conviction DNA exonerations wisely to throw light on the more general problem. Second, the closing window has statistical implications for our study. Our choice of denominator must be chosen with care, both with respect to the kind of defendants we are examining and with respect to the time period chosen for examination.
The twelve trials of the thirteen capital rape-murder defendants that resulted in their factually wrongful convictions took place between 1979 and 1996. (20) Two of the twelve trials are clearly outliers--the 1979 trial of Dennis Williams took place three years before the next later trial, and the first trial of Ray Krone (1992) occurred three years after the next earlier trial. The Williams case was unusually early for usable DNA evidence to have been preserved and discovered, but this prescience is perhaps accounted for by the fact that the state in that case was still looking to prosecute a co-defendant, which they did not manage to do until 1985 (Verneal Jimerson, also later exonerated by DNA). The Ray Krone case in 1992/1996 is remarkably belated for DNA not to have been utilized originally. At any rate, it seems clear that it is neither required nor justified statistically to retain these two outlier examples in the numerator set. (21) So for purposes of looking at the wrongful conviction rate, we will limit ourselves to the eleven cases that were tried from 1982 to 1989 inclusive. In addition, we will reduce the number by half an exoneration, in order to give some cushion against the criticism that it is not beyond every doubt that every person exonerated by DNA was factually innocent. As noted earlier, there are the Paleyites of the world, such as Joshua Marquis, who will claim that these exonerations are not sufficiently absolute because it is possible to imagine (usually exceedingly unlikely) scenarios in which this or that exoneree might still be guilty. Nevertheless, even the most aggressive of these Paleyites would probably not argue that such exercises in creative imagination mean that none of the DNA exonerees is factually innocent. (22) If we give an exceedingly generous probability of one in twenty to the factual guilt of an apparently exonerated defendant, then a statistical exclusion of one-half an exoneration covers it. (23)
So we start with a numerator of 10.5. What, then, is the denominator? If we choose all death penalties imposed from 1982 to 1989 inclusive, we get a denominator of 2235. (24) That denominator would yield a minimum factual innocence rate of .47%, or nearly five in a thousand (and more than double the figure arrived at when we used all capital DNA exonerations and all death sentences). (25)
But that denominator is still understating the factual innocence rate, because it is still incorrect. The number of all death penalties imposed from 1982 to 1989 inclusive includes all sorts of capital cases that were not rape-murders. (26) The proper denominator is the number of capital rape-murder cases. An analysis of a sample of 406 capital convictions imposed in the period 1982-1989 inclusive indicates that only 21.45% of capital sentences involve a rape-murder. (27) Thus, the proper denominator is 479, (28) and thus the factual innocence rate for capital rape-murder convictions in the period 1982-1989 inclusive is at least 2.2%.
We have not finished yet with our denominator, however. It is still overstated. DNA exonerations can only occur in those rape-murder cases where usable DNA connected to the perpetrator is found to be available when requested for testing. The universe represented by the DNA exonerations, therefore, is defined by that condition. In what percentage of cases involving trials in the reference period is that condition present? (29)
So far as I know, there is only one organization in the country, governmental or non-governmental, with records of sufficient experience to give a defensible answer to this question: the Innocence Project at Cardozo Law School. Established by Barry Scheck and Peter Neufeld in 1992, the Innocence Project has concentrated from the beginning on the exoneration of the convicted innocent through DNA. As the undisputed leader in pursuing that goal, there is little doubt that it has records of more requests for DNA evidence in more cases than any other entity. In the summer of 2006 I contacted the Innocence Project and asked them to use their records to determine in what percentage of cases tried during the reference time period for which requests for the discovery of DNA evidence had been made in which it was subsequently established that, either because it was never collected, because it was discarded or destroyed, or because it was degraded, no usable DNA survived. The Innocence Project itself has dealt with a limited number of capital cases, since those cases usually have other sources of post-conviction representation. However, the Innocence Project has dealt with many non-capital rape-murders, and even more non-murder rapes, and there seems no reason to believe that the percentage of cases in which no usable DNA survived would be significantly different for either of those categories than that in capital rape-murders.
By a lucky coincidence, at the same time that I contacted them, the Innocence Project had just begun a comprehensive file review in order to collect various data across cases, therefore isolating the data on "no usable DNA" in regard to their cases in the reference period was not too burdensome, and they graciously agreed to do it. The results were that 77 of the 212 cases tried during the reference time period for which the Innocence Project made requests for DNA testing did not yield usable DNA, a rate of 36.3%. In that set of 212, there were 15 rape-murders, of which 5 (33.3%) yielded no usable DNA. Although the rape-murder set is smaller, I have elected to use the 33.3% rate in an abundance of caution, since it is lower than the rate associated with total number of cases for which requests were made. (30)
The denominator of 479 for the number of capital rape-murder convictions in the reference period, which we previously derived, must therefore be reduced by 33.3% to account for cases with no usable DNA available for testing, thereby yielding a denominator of 319. Using 10.5 as the numerator, as previously explained, and 319 as the denominator yields a true minimum innocence rate for rape-murder from 1982-1989 inclusive of (pace Joshua Marquis and Justice Scalia) 3.3%. (31)
So there we have it. A conservative minimum factual innocence rate, derived from a real, not insignificant, set of serious criminal convictions, and capital convictions to boot. The question immediately comes to mind: What can this specific rate tell us about wrongful conviction rates in general?
Before addressing this important question, however, we must examine one more issue in regard to our initial reference set: capital rape-murder convictions in the 1980s. We have derived a minimum factual innocence rate. What, if anything, can be said about the maximum factual innocence rate? We have a floor. What can we say about a ceiling?
As it turns out, I think we can say some closely defensible things. We start off being reasonably sure that there are around 319 capital rape-murder cases (more or less) with potentially usable DNA evidence during that period. We also know that many of these cases have had the DNA requested and analyzed. We do not know how many cases exactly, because there is no central database of such information, (32) but capital cases generally attract post-conviction aid from anti-death penalty advocates. Among those 319 cases, to be sure, there are a few that are so clear on factual guilt that DNA analysis might not be requested. For instance, Charles McDowell broke into a house in his own neighborhood and attacked the maid, who was alone in the house at the time. Because of the victim's screams, a neighbor who knew him confronted him as he was leaving the premises covered in blood. He stabbed the neighbor, but police were called by other neighbors. The police followed the blood trail, and within a short period of time, found him hiding in some bushes at the other end of the blood trail. There was semen on the victim's panties, but DNA evidence was not likely to help the defense. (33)
However, most cases are not so clear as the McDowell case with regard to factual guilt independent of DNA. Thus, DNA exclusion is the greatest post-conviction hope, and would generally be worth requesting. It would be quite surprising, perhaps even shocking, if capital post-conviction counsel had failed to request DNA testing in anything close to half the 319 capital rape-murder cases in the reference set. Additionally, even in these cases, it is likely that the requests, if not universally made, would be skewed toward being made in the otherwise more factually questionable cases. So I believe we can conclude without much doubt that the ceiling is not double the floor (which would give a maximum ceiling figure for actual innocence of 6.4%), but is in fact substantially less. I believe it is fair to put a reasonable maximum under these circumstances at around 5%.
Thus, we have an empirical minimum of 3.3% and a fairly generous likely maximum of 5% for factually wrongful convictions in capital rape-murders in the 1980s.
IV. IMPLICATIONS OF A 3-5% FACTUAL WRONGFUL CONVICTION RATE FOR BOTH PALEYITES AND ROMILLISTS
These figures are guaranteed not to make many people happy. Whatever the depth (or shallowness) of one's emotional or moral response to a 3-5% factual innocence error rate in a significant set of real-world capital cases, it is hard to characterize it as de minimis, or to fairly say that it represents a "remote" possibility of conviction of the innocent. Paleyites often depend on the tenability of such assertions either to make themselves feel better, or to convince the general mass of people that there is no systemic problem of wrongful conviction to be considered, or both. When real data carefully derived destroy the tenability of such claims, one can depend on the Joshua Marquises and the Justice Scalias not to be happy with that result.
In addition, Paleyites will find little to comfort them regarding claims that such exonerations are demonstrations of "the system working," or that reversals through the ordinary appellate process take care of the problem of wrongful conviction. (34) In general, over two-thirds of all capital convictions are reversed, and more than half of defendants who initially receive capital sentences are ultimately removed from death row (although fewer than four in one thousand...
|
