A death row inmate who challenges either his conviction or sentence in postconviction proceedings can be said to succeed if he obtains either a new guilt-phase trial, a new sentencing-phase trial, or a commutation of his death sentence. This Article reports on the success rates of death row inmates in Texas for those who arrived on death row on or after January 1, 2000, up until December 31, 2019. The data set (N=282) covers the status of all these arrivals through February 25, 2020.
A decade ago, Eric M. Freedman and David R. Dow published an analysis of the effect of the Anti-Terrorism and Effective Death Penalty Act of 1996 (AEDPA) on the granting of relief to death row inmates in federal habeas corpus litigation. In that analysis, they began their discussion by summarizing a study conducted by James S. Liebman and his colleagues, which examined every capital case in the United States from 1973 through 1995. This period covered the entire modern era of the death penalty, from its brief demise and almost immediate return in the 1970s up until the enactment of AEDPA.
Liebman’s team accounted for more than 4500 habeas applications raising either guilt- or punishment-phase challenges (or both) to state-imposed death sentences. Liebman observed an overall error rate of some 68 percent during the study period. Specifically, for every 100 cases involving death-sentenced prisoners, forty-seven inmates obtained relief during state court appeals. Of the remaining fifty-three, twenty-one obtained relief during federal habeas proceedings (for an overall error rate of 68 percent (47 + 21)). These data, of course, reflect a staggering overall success rate and a specific success rate in federal habeas proceedings of almost 40 percent. Liebman inferred from his data that the number of inmates entitled to relief was in fact even higher.
Dow and Freedman were interested in determining how AEDPA affected this success rate. AEDPA was designed specifically to accelerate federal habeas review of state death penalty cases and shorten the time between sentencing and execution; its principal means to achieve that aim is to impose both substantive and procedural limitations on the authority of federal courts to grant relief in habeas proceedings. Dow and Freedman were interested in measuring the apparent magnitude of those limitations.
The period from 1996 through 1999 fell outside the focus of the study, on the assumption that most of the cases resolved during that period had been filed before AEDPA’s effective date and would therefore not be significantly affected by its passage.
Dow and Freedman therefore examined all federal habeas applications filed between January 2000 and January 2007. The data revealed that, in the aggregate, death row inmates had a nationwide success rate of around 12 percent. This aggregate success rate, however, disguised significant variations among jurisdictions and federal circuits. For example, the success rate in the Fourth Circuit was less than 2 percent; in the Fifth and Eleventh Circuits, the success rate was less than 4 percent; in the Ninth Circuit, in contrast, the success rate was around 35 percent.
Dow and Freedman’s dataset is now more than twelve years old, and there appear to be no published data examining success rates since the period Dow and Freedman covered. In this Article, therefore, we intend to take a first step in making the data more current. We provide a new batch of statistics, focused exclusively on Texas cases.
We have examined all Texas death penalty cases from January 1, 2000, through February 25, 2020. Our data set comprises all those inmates who arrived on death row on or after the start date, and includes inmates who were previously sentenced to death, but who, following retrials, were sentenced to death again and arrived at death row again on or after the start date.
Our decision to focus specifically on Texas was influenced by two primary factors.
First and foremost, the sheer volume of completely adjudicated postconviction cases in Texas dwarfs the volume in any other jurisdiction. With the exception of so-called volunteers, who may have surrendered their opportunity to obtain postconviction review of their trial proceedings, the occurrence of an execution means that an inmate’s postconviction proceedings have run their course. The largest number of such inmates, by far, is in Texas. Specifically, from 2000 through the end of 2019, Texas accounted for more than 40 percent of executions in the United States. Moreover, Texas accounted for 282 of the 2091 defendants initially sentenced to death nationwide during that period, around 13.5 percent. As a result, Texas provides the most robust dataset for examining success rates in the twenty-first century.
Second, insofar as the success rate in Texas was already among the lowest for the period analyzed by Dow and Freedman, it would be reasonable to assume that any further decline would be quite small or that success rates would modestly climb. By way of contrast, the Ninth Circuit, with its 35 percent success rate in the period examined by Dow and Freedman, and the Seventh and Tenth Circuits, with their 23 percent success rates, provided much greater opportunities for decline. Put differently, the data from the period covered by Dow and Freedman do not leave much room for the success rates of death-row inmates to decline any further, and our new data set therefore permits us to examine whether there are any surprises—in particular, whether, despite the very narrow possibility for further declines, death row inmates have in fact succeeded even less often over the past twenty years.
To be sure, regardless of whether success rates in Texas were to rise or fall in comparison to the period studied by Dow and Freeman, there may be reasons to believe that those success rates are not representative of success rates elsewhere. The reason is that, during the two-decade period we analyze, the U.S. Supreme Court has strongly rebuked both the Texas Court of Criminal Appeals (the highest state court to review death penalty cases in Texas), as well as the U.S. Court of Appeals for the Fifth Circuit (the U.S. circuit that reviews cases from Texas) for interpreting or applying its decisions pertaining to death penalty jurisprudence wrongly or in too narrow a manner. This fact could suggest either that Texas success rates were unduly low before these rebukes, or, that in response to these corrections, the success rates have become unrepresentatively high. If we were to identify lumpiness in the data over the period we study, that lumpiness would be consistent with either of the two foregoing possibilities. In contrast, if the results are reasonably smooth over the entire period, the doctrinal corrections implicit in those rebukes would appear not to have had an impact on success rates.
We stress, of course, that the success rates observed by Dow and Freedman varied widely across jurisdictions, and that same variation may well still exist. For that reason, some caution may be called for in inferring that the data from Texas will be replicated in other death penalty jurisdictions.
II. Introduction to the Data
Two brief doctrinal observations may prove useful in fully appreciating the impact of the data we present below.
First, in 2002, the Supreme Court decided Atkins v. Virginia, and held that the Eighth Amendment categorically prohibits the execution of offenders who are intellectually disabled. When Atkins was decided, Justice Scalia predicted the ruling would lead to a profusion of inmates claiming to be intellectually disabled. That prediction proved to be incorrect. The percentage of inmates raising so-called Atkins claims has remained stable at around 7 percent. In Texas, the success rate for inmates raising Atkins claims from 2002 to 2013 was around 18 percent (8/45)—four-and-a-half times the success rate in the period studied by Dow and Freedman. (For reasons of timing, relatively few Atkins claims had been fully litigated in the period studied by Dow and Freedman.) Of particular importance to our analysis of the data in Texas is that the success rate for those raising Atkins claims has declined in the cases we examined, but it remains comparatively high nationwide.
Second, in 2005, in Roper v. Simmons, the Court held the Eighth Amendment does not allow the states to execute offenders who are under the age of eighteen at the time they commit homicide. At the time Simmons was decided, the seventy-one juveniles then on death row in the United States accounted for approximately 2 percent of the nation’s total death row population. Texas held twenty-nine juveniles on death row (41 percent of the national total). The juveniles whose habeas petitions were pending at the time Simmons was decided would have all succeeded in challenging their death sentences, so any spike in success rates in the immediate aftermath of Simmons would represent a distortion of the pattern, albeit a temporary one.
Both Atkins and Simmons are eligibility decisions. That is, both make a death sentence impermissible for a wrongdoer who has certain characteristics, irrespective of the adequacy of the legal proceedings. There is, of course, an important distinction between these two eligibility decisions—namely, Atkins allowed the states to develop procedures for determining whether a particular inmate (or defendant) is intellectually disabled, and this determination is often subject to disagreement among competing experts. In contrast, whether an individual was eighteen on the day a crime occurred is largely beyond dispute. Notwithstanding this salient difference, what both cases have in common is to shrink the universe of offenders eligible for execution. Defendants not included in the universe of either the intellectually disabled or those under the age of eighteen, however, can succeed in their postconviction litigation only on the basis of less categorical claims; generally, such claims are subject to a much broader range of judicial discretion. All of which is to say that a nuanced view of our data should take into account whether a successful death row inmate prevails on an eligibility claim, or whether that success implicates a more elastic doctrinal principle.
In order to permit eligibility claims to be easily accounted for, we indicate in the data collection whether a successful appeal depended on either Atkins or Simmons.
We now turn to the data. We began by identifying every person sentenced to death (death row arrivals) from January 1, 2000 through December 31, 2019. By statutory mandate, every one of these new arrivals must have his case reviewed in state direct appeal proceedings. (Death row inmates in Texas may waive their right to counsel and proceed pro se, but they may not waive direct appeal review.)
As of the end of 2019, 264 direct appeals in death penalty cases in our data set had been reviewed by the Texas Court of Criminal Appeals (CCA). Thus, approximately 94 percent (264/282) of the inmates in our data set have had their direct appeal completed.
Following direct appeal proceedings, a death-sentenced inmate may seek state habeas relief in accordance with article 11.071 of the Texas Code of Criminal Procedure. State habeas relief is subject to waiver by an inmate deemed competent to waive his appeals.
In the period studied, the CCA adjudicated 214 cases brought by inmates seeking relief in state habeas proceedings. This number, representing more than three-quarters of the inmates in our data set (214/282 = 75.9 percent), excludes inmates who either obtained relief during direct appeal proceedings or elected to waive state habeas proceedings.
Finally, an inmate who obtains relief in neither state direct appeal nor state habeas litigation may seek federal habeas relief pursuant to 28 U.S.C. §§ 2244, 2254. Again, if deemed competent, death row inmates may waive federal habeas appeals. But if the district court identifies bona fide questions regarding the inmate’s competency, due process requires adversarial proceedings to test the inmate’s competency and assess whether his ostensible waiver of collateral review is knowing and voluntary.
In the period studied, death row inmates in Texas completed 151 federal habeas proceedings, meaning slightly more than half our data set have gone through at least one round of federal habeas proceedings (151/282 = 53.5 percent). This number therefore excludes inmates who prevailed in either direct appeal or state habeas proceedings, or who waived federal habeas proceedings, or who, in a small number of cases, died during the midst of federal habeas proceeding.
The complete data set, showing which inmates were successful, in what forum, and on which issue, is contained in can be downloaded at the link in the footnote below. In the text that follows, we summarize the results:
Of 262 direct appeals adjudicated by the CCA, the death row inmate succeeded in fifteen cases, or 5.7 percent. Of these fifteen successful cases, four involved eligibility determinations, and eleven implicated what we refer to as more elastic doctrinal restrictions.
Of the 214 inmates whose state habeas applications were adjudicated by the CCA, twelve, or 5.6 percent were successful. Of these twelve successful cases, two involved eligibility determinations, and ten implicated what we refer to as more elastic doctrinal restrictions.
Finally of the 151 completed federal habeas proceedings, inmates were ultimately successful in a single case. (In one additional case, the inmate was successful in the Fifth Circuit, but the court of appeals subsequently granted the government’s petition for en banc review, and the full court has not yet resolved the case.)
In general, therefore, these statistics reveal a pattern consistent with the earlier period examined by Dow and Freedman. Moreover, we now have a comprehensive study of the two-decade period from the resumption of the death penalty through the enactment of AEDPA, as well as two studies that, together, represent a granular examination of the two-decade period since AEDPA. Even if we cannot state with certainty that AEDPA itself has caused this precipitous decline in the success rate of death row inmates in their appeals, we can say that AEDPA represents the boundary between two eras. In the first, death-sentenced inmates prevailed two-thirds of the time; in the second, their success rate percentage is in the single-digits. If, therefore, a principal objective of AEDPA was to insulate state-imposed death sentences from constitutional attack, the data strongly imply that objective has been achieved. Death row inmates challenging their convictions or sentences in Texas prevail dramatically less often than they did before the enactment of AEDPA. The occasional rebukes by the Supreme Court of the lower courts with jurisdiction over Texas capital cases suggests the best explanation for these statistics is not that constitutional norms are being vigorously enforced in the trial courts, but instead that constitutional rights are increasingly difficult to vindicate.Download data here.
. Because women represent only around 2 percent of the total death row population in the United States, it would be precious to say “he or she” when referring to these inmates; we therefore refer to death row inmates using masculine pronouns. See Death Penalty Info. Ctr., Facts About the Death Penalty 3 (2020), https://files.deathpenaltyinfo.org/documents/pdf/FactSheet.f1585003454.pdf [https://perma.cc/N5ZM-4C3V] (reporting that, as of July 1, 2019, women made up fifty-five out of 2656 (2.07 percent) inmates on death row).
. See Appendix, Texas Capital Rates, 68 UCLA L. Rev. app. (2020), https://www.uclalawreview.org/wp-content/uploads/2020/04/texas-capital-rates.xlsx [https://perma.cc/9VAE-HVTG].
. Siggi B. Wilzig Distinguished Professor of Constitutional Rights at the Maurice A. Deane School of Law at Hofstra University.
. See David R. Dow & Eric M. Freedman, The Effects of AEDPA on Justice, in The Future of America’s Death Penalty 261 (Charles S. Lanier et al. eds., 2009). Our focus was specifically on 28 U.S.C. §§ 2244, 2254 (2018).
. James S. Liebman et al., A Broken System: Error Rates in Capital Cases, 1973-1995 (Columbia Law Sch. Pub. Law & Legal Theory Working Paper Grp., Paper No. 15, 2000), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=232712.
. Dow & Freedman, supra note 4, at 265 (citing Liebman et al., supra note 5, at i).
. Pub. L. No. 104-132, 110 Stat. 1214 (codified as amended in scattered sections of the U.S. Code). See also Furman v. Georgia, 408 U.S. 238, 239–40 (1972) (per curiam) (invalidating all then-existing death penalty statutes); Gregg v. Georgia, 428 U.S. 153, 154–56 (1976) (holding that the death penalty does not violate the Eighth and Fourteenth Amendments under all circumstances, and overturning Furman v. Georgia to the extent that case’s holding conflicted with this principle). In their study of the effect of AEDPA on the granting of relief to death row inmates in federal habeas corpus litigation, Dow and Freedman identify the relevant provisions of AEDPA as § 2254 and, to a lesser extent, § 2244. Dow & Freedman, supra note 4, at 265–67.
. Dow & Freedman, supra note 4, at 265 (citing Liebman et al., supra note 5, at 6).
. Id. (citing Liebman et al., supra note 5, at i–ii). Liebman’s assumption was reinforced by the rather astonishing fact that in 82 percent of the cases in which an inmate received punishment phase relief, the subsequent trial resulted in a sentence less than death (and in 7 percent of the cases, the defendant was found innocent of the capital offense altogether). Id. at 265 (citing Liebman et al., supra note 5, at ii).
. Id. at 266; see also Woodford v. Garceau, 538 U.S. 202, 206 (2003) (citing Williams v. Taylor, 529 U.S. 362, 386 (2000) (opinion of Stevens, J.); id. at 404 (majority opinion)). But cf. Margaret A. Upshaw, The Unappealing State of Certificates of Appealability, 82 U. Chi. L. Rev. 1609, 1614–15 (2015) (noting, appropriately, the muddy legislative history and textual clarity of AEDPA); Note, The Avoidance of Constitutional Questions and the Preservation of Judicial Review: Federal Court Treatment of the New Habeas Provisions, 111 Harv. L. Rev. 1578, 1593–95 (1998) (noting the same, and cautioning against relying solely on legislative intent as a guide to ascertaining AEDPA’s objectives).
. Dow & Freedman, supra note 4, at 266.
. See Lindh v. Murphy, 521 U.S. 320, 326–32 (1997) (holding AEDPA's restrictions apply prospectively only and not to cases already filed at the time of its enactment).
. Dow & Freedman, supra note 4, at 267. The data are reported fully in id. at 270–93.
. We have compiled this data set using a variety of public and private sources, including the Texas Department of Criminal Justice (TDCJ) and the Death Penalty Information Center (DPIC). See Death Row Information, Tex. Dep’t Crim. Just., https://www.tdcj.texas.gov/death_row/index.html [https://perma.cc/M83Y-RDAU] (last visited Mar. 26, 2020); Execution Database, Death Penalty Info. Ctr., https://deathpenaltyinfo.org/executions/execution-database [https://perma.cc/8VZ9-46CE] (last visited Mar. 26, 2020). Both databases maintain robust datasets, and much of DPIC’s dataset is searchable. Our information for arrivals on Texas’s death row comes from our analysis of the TDCJ website. To evaluate dispositions of direct appeals, state habeas applications, and federal habeas proceedings, we searched the dockets for the relevant court and reviewed the final judgments. As all death penalty lawyers are well-aware, the small community teems with goodwill, and we called on that goodwill to help us fill in gaps. We invite anyone with corrections or updates to our data to contact us.
Although not the focus of our present analysis, our dataset also permits an examination of a variety of other statistical data. For example, we reveal the race of the inmate, as well as the race of the victim(s) in each inmate’s case. Finally, in the case of those inmates who have been executed as of February 25, 2020, we calculate the number of days that inmate resided on death row, and arrive at a mean and median calculation of the time spent on death row between arrival and execution.
. See Appendix, supra note 2.
. Volunteers are inmates who waive their right to appeal their conviction and sentence. Meredith Martin Rountree, Volunteers for Execution: Directions for Further Research Into Grief, Culpability, and Legal Structures, 82 UMKC L. Rev. 295, 295 (2014); see also John H. Blume, Killing the Willing: “Volunteers,” Suicide and Competency, 103 Mich. L. Rev. 939, 940 (2005). The number of volunteers is, by some estimates, as high as 11 percent. See Rountree, supra.
. Texas had 369 of the 912 nationwide executions during that period, or 40.46 percent. Execution Database, supra note 14. After Texas, the state with the largest number of executions is Oklahoma, with ninety-four. Id.
. Death Sentences in the United States Since 1977, Death Penalty Info. Ctr., https://deathpenaltyinfo.org/facts-and-research/sentencing-data/death-sentences-in-the-united-states-from-1977-by-state-and-by-year [https://perma.cc/3YDC-77RY] (last visited Mar. 28, 2020).
. See Dow & Freedman, supra note 4, at 267.
. See, e.g., Moore v. Texas (Moore I), 137 S. Ct. 1039, 1051–53 (2017) (rebuking the Texas Court of Criminal Appeals (CCA) for its use of the so-called Briseno factors, Ex parte Briseno, 135 S.W.3d 1, 4–19 (Tex. Crim. App. 2004), abrogated by Moore I, 137 S. Ct. 1039 (applying the Supreme Court’s decision in Atkins v. Virginia, 536 U.S. 304, 321 (2002), which prohibited states from executing individuals with intellectual disabilities)); Moore v. Texas (Moore II), 139 S. Ct. 666, 667 (2019) (per curiam) (reversing the CCA’s remand decision which reached the same conclusion that the Supreme Court had vacated in Moore I); Smith v. Texas (Smith I), 543 U.S. 37, 38, 48 (2004) (per curiam) (rebuking the CCA for its evasion of the Supreme Court’s decision in Penry v. Lynaugh (Penry I), 492 U.S. 302, 303, 317–19 (1989) (holding that “Texas juries must, upon request, be given jury instructions that make it possible for them to give effect to . . . mitigating evidence” of intellectual disabilities at the punishment phase of a capital trial, as required by Lockett v. Ohio, 438 U.S. 586, 607 (1978) and Eddings v. Oklahoma, 455 U.S. 104, 113–14 (1982)) and Penry v. Johnson (Penry II), 532 U.S. 782, 797 (2001) (holding that a so-called nullification instruction did not cure the error identified in Penry I)); Smith v. Texas (Smith II), 550 U.S. 297, 300 (2007) (again reversing the CCA’s remand decision which reached the same result that the Supreme Court had reversed in Smith I); Tennard v. Dretke, 542 U.S. 274, 283, 287 (2004) (criticizing the Fifth Circuit’s similar evasion of Penry I and noting that the lower court was merely “paying lipservice to principles guiding issuance of a [certificate of appealability (COA)]”); Miller-El v. Cockrell (Miller-El I), 537 U.S. 322, 341 (2003) (rejecting in general the Fifth Circuit’s standard for issuing a COA, a prerequisite for appealing from a denial of habeas relief); Miller-El v. Dretke (Miller-El II), 545 U.S. 231, 236–37 (2005) (granting merits-based relief on habeas petitioner’s claim under Batson v. Kentucky, 476 U.S. 79 (1986), after Fifth Circuit denied relief following Miller-El I).
The Roman numerals included in the case citations in the preceding paragraph themselves tell the story in brief of the hostility to claims brought by death row inmates in Texas and thereby accurately adumbrate the data we report below.
. 536 U.S. 304 (2002).
. See id. at 321. Atkins used the phrase “mentally retarded,” a term that was subsequently superseded by the term “intellectually disabled.” See Hall v. Florida, 572 U.S. 701, 704–05 (2014). As a matter of clinical definition, both terms mean the same thing. Id.
. See John H. Blume et al., A Tale of Two (and Possibly Three) Atkins: Intellectual Disability and Capital Punishment Twelve Years After the Supreme Court’s Creation of a Categorical Bar, 23 Wm. & Mary Bill Rts. J. 393, 396–99 (2014).
. Id. at 412–13.
. See id. at 396–99, 396–98 nn.19–29. Blume and his colleagues found a success rate of 63 percent during the period from 2002 to 2008, and a success rate of 43 percent from 2009 to 2013. Id. at 398.
. 543 U.S. 551 (2005).
. Victor L. Streib, The Juvenile Death Penalty Today: Death Sentences and Executions for Juvenile Crimes, January 1, 1973–February 28, 2005, at 11 (2005); Mary Berkheiser, Capitalizing Adolescence: Juvenile Offenders on Death Row, 59 U. Miami L. Rev. 135, 160 (2005).
. Atkins v. Virginia, 536 U.S. 304, 317 (2002).
. See Hall v. Florida, 572 U.S. 701, 710 (2014). There are limits, however, on a state’s ability to define intellectual disability. Such limits are established by accepted medical clinical criteria. See, e.g., id.; Moore v. Texas (Moore II), 139 S. Ct. 666, 668, 670 (2019). Moreover, the differing arduousness of the tests for establishing intellectual disability adopted by the states appears to correlate with the success rates of inmates raising Atkins claims in those states. See Blume et al., supra note 23, at 412–14.
. See, e.g., Falk v. Texas, No. AP-77,071, 2018 WL 3570596, at *1 n.2 (Tex. Crim. App. July 25, 2018).
. Tex. Code Crim. Proc. Ann. art. 11.071 (West 2019).
. See, e.g., Ex parte Lopez, Nos. WR-77,157-01 & WR-77,157-02, 2015 WL 4644657, at *1 (Tex. Crim. App. Aug. 4, 2015) (for the sake of full disclosure, Lopez was represented by Dow and Newberry); Ex parte Reynoso, 257 S.W.3d 715, 720–21 & n.2 (Tex. Crim. App. 2008). To be effective, any such purported waiver must be intelligent and voluntary. See, e.g., Ex parte Gonzales, 463 S.W.3d 508, 510 (Tex. Crim. App. 2015) (Yeary, J., dissenting); see also id. at 511–13 (suggesting state court use same standard for knowing and voluntary waiver as is used by federal courts for purposes of waiver of federal collateral review proceedings).
Our data set contains at least seventeen so-called volunteers, or roughly 6 percent, including the following:
999174, Michael Gonzales (waived state habeas);
999358, Larry Hayes (waived federal habeas);
999370, Danielle Simpson (waived federal habeas appeal);
999378, James Porter (waived federal habeas);
999387, Ynobe Matthews (waived state and federal habeas);
999413, Michael Rodriguez (waived federal habeas);
999438, Alexander Martinez (waived state and federal habeas);
999481, Barney Fuller (waived federal habeas);
999496, Christopher Swift (waived state and federal habeas);
999523, Richard Tabler (waived state habeas);
999552, Jerry Martin (waived state and federal habeas);
999555, Daniel Lopez (waived state and federal habeas);
999563, Travis Mullis (waived state habeas).
In addition, as the accompanying chart indicates, five inmates died of apparently natural causes, and another six died by suicide before the completion of their habeas proceedings—making a total of twenty-eight inmates who died on death row before being executed.
. See Lopez v. Stephens, 783 F.3d 524, 525 (5th Cir. 2015); Mata v. Johnson, 210 F.3d 324, 329 (5th Cir. 2000).
. “Completed” is defined as the resolution of district court proceedings as well as proceedings in the United States Court of Appeals for the Fifth Circuit and the Supreme Court. See supra note 32. The Fifth Circuit also reviews habeas applications from death-sentenced inmates in Louisiana and Mississippi, but those cases are not included in our database.
. See Appendix, supra note 2.
. “Ultimate success” means a grant of relief was sustained; if an inmate obtained relief that was then set aside by either the Fifth Circuit or the Supreme Court, we do not count it as successful.
. Dow & Freedman, supra note 4, at 293.
. See supra notes 20, 32.