A. Prior Studies of Habeas Grant Rates.

While the raw figure of 40% was consistent between our study and the opposition, the interpretations assigned to that figure were markedly different. Persons who favor relitigating every federal constitutional question on federal habeas corpus interpreted the figure as a sign of systemic deficiencies in the state courts, indicating that they could not be trusted to protect federal constitutional rights.⁽³⁾ CJLF's detailed analysis of the individual cases, however, indicated that the state courts had invariably reached a reasonable decision on the claims presented to them as of the time of their decision. When relief was granted in the federal courts, it was generally because the law had changed, because the claim was never properly presented to the state courts, or because the courts simply disagreed in a vast and confusing area of law where competent, conscientious jurists frequently disagree.⁽⁴⁾

In late 1994, the National Center for State Courts (NCSC) and the State Justice Institute published a new study of habeas corpus cases. That study, based on more recent data, concluded that the success rate for capital habeas petitioners was "vastly overstated."⁽⁵⁾

Looking at the dates and success rates reported in the various studies, it appears likely that, rather than being overstated, the 40% figure is simply outdated. Each study that is done, with one exception, shows a lower rate than earlier studies. In a brief filed in 1983, the NAACP Legal Defense and Education Fund stated that the rate was 73%.⁽⁶⁾ In 1989, CJLF found a 40% rate as did a 1976-91 study by Professor James Liebman, both noted above.

The National Center for State Courts examined data for two years. The 1990 data showed one grant of relief in twelve cases,⁽⁷⁾ but the resulting 8.5% rate can be dismissed as based on an inadequate sample size. The 1992 data showed 7 grants of relief in 45 final judgments, or 15%.⁽⁸⁾ Combining the two would yield 8 grants in 57 cases, or 14%. The NAACP, CJLF, Liebman, 1990 NCSC, and 1992 NCSC data are graphed in Figure 1, with each study assigned a date of the midpoint of its period.

This remarkable drop in the grant rate over time warranted further investigation. CJLF decided to do another study similar to our 1989 study to see if the results really do vary over time. Stated another way, is the success rate really a declining function of time, as Figure 1 would seem to indicate, or do the different figures result from differences in the studies, with the apparent time dependence being merely coincidental?

B. Method of the Present Study.

Unpublished cases were available from the Fourth and Eighth Circuits; the database was limited to published cases from the rest of the circuits. However, all of the Eighth Circuit capital habeas cases were published. The Fourth Circuit had three unpublished opinions, all of them denials of relief. An earlier study found little difference in grant rates between published appellate opinions and all verified cases.⁽¹²⁾

The cases were counted as granting or denying relief as to the guilt determination. A grant on guilt is necessarily a grant of relief from the death penalty. Cases denying relief on guilt were further classified as granting or denying relief as to penalty. Thus there are three possible outcomes: denial of relief, grant as to penalty only, or grant as to both guilt and penalty.

Upon tallying the result by circuit, it quickly became apparent that the Ninth Circuit was radically different from the others. The results are therefore presented in Table 1 for the overall count, the count for circuits other than the Ninth, and the count for the Ninth alone.

C. Results.

The overall count confirms NCSC's result that the grant rate has dropped dramatically from the earlier studies. The overall rate for any relief, guilt or penalty, is 15%. That figure is slightly over one-fifth of the NAACP's 1976-83 figure and a little over a third of Liebman's 1976-91 figure and CJLF's 1986-89 figure.

D. Discussion.

     1. Overall rate.

Similarly, in the guilt phase of trials we seem to have reached a quiescent period in which the Supreme Court has stopped disapproving standard jury instructions which have been long accepted and widely used.⁽¹⁶⁾ CJLF's previous study had found that one such decision accounted for a major portion of guilt-phase reversals in a state that had used a traditional, but since disapproved, instruction.⁽¹⁷⁾ If turmoil was the cause of the high grant rate, then reduced turmoil naturally reduces the grant rate.

A second possibility, based on the hypothesis that the earlier, high grant rate was caused by dismal state courts and counsel, would be that the state courts and counsel have improved dramatically in just a few short years. Such dramatic improvement in so short a time seems highly unlikely, but even if it were the explanation, it would still remove the argument that state courts cannot be trusted.

A third possibility would be that the Supreme Court's new limitations on habeas relief are responsible for the dramatic decline in grant rates. Since this possibility is likely to be the one endorsed by the opponents of reform, it bears close examination.

The first major limitation is the procedural default rule of Wainwright v. Sykes.⁽¹⁸⁾ This rule has been in place since the second year of the modern capital punishment era, and the major questions as to its application were settled in 1986.⁽¹⁹⁾ With no significant changes in this rule between 1986 and the present, it cannot explain a sharp drop in grant rates during that period. Similarly, the exclusion of fully litigated Fourth Amendment claims from habeas corpus has been in effect throughout the modern capital punishment era,⁽²⁰⁾ and cannot explain changes within those years.

Another major limitation is the strong bar against second and subsequent petitions by prisoners who have already had one full federal review.⁽²¹⁾ Denial of successive petitions on this ground cannot explain the drop in the grant rate, because successive petitions were not a substantial portion of the earlier, high grant rate. CJLF's earlier study found not a single case of relief being granted on a successive petition.⁽²²⁾ The entire 40% was made up of first petitions.

This leaves what is perhaps the most controversial of the recent limitations: the nonretroactivity doctrine of Teague v. Lane.⁽²³⁾ The Teague case held, in essence, that if the state courts entering and affirming the conviction obeyed the constitutional rules in effect at that time, the conviction would not be rendered void by new rules created afterward. An important corollary to this principle is that new rules cannot be created on habeas corpus.⁽²⁴⁾

The Teague rule was invoked in 10 of the 51 cases in our sample in which relief was denied.⁽²⁵⁾ None of these cases invoked Teague to deny a claim which would have been clearly meritorious without that rule. Six of the ten were Texas cases involving the Penry-Graham-Johnson problem, and in all six the facts of the case resembled Johnson more than Penry.⁽²⁶⁾ The Fifth Circuit could have rejected the claims under either Graham, a Teague case, or Johnson, a direct review case, and chose the simpler of the two.

In the other four cases, the habeas petitioners were similarly trying to "push the envelope," making claims that ranged from dubious to frivolous even as of the date of the Court of Appeals' decision.⁽²⁷⁾ Thus, Teague has only been invoked in this sample to reject claims on the far fringe of the law that would likely have been rejected even without Teague. While one or two of these claims might have been granted by a defense-leaning court, the rejection of these marginal claims cannot explain the dramatic drop in the grant rate.

     2. Guilt reversal rate.

One of the three cases, Burden, was a case accepted for review twice by the U. S. Supreme Court.⁽³⁰⁾ The Supreme Court can review a case after an adequate state habeas proceeding as well as it can after a federal habeas proceeding, so cases requiring Supreme Court intervention do not support an argument for duplicating adequate state proceedings in federal court.

     3. Circuit differences.

There are two possible reasons for this dramatically different grant rate in the Ninth Circuit. Either the state courts in the states of the Far West are doing an exceptionally poor job relative to the other states with large numbers of capital cases, or the Ninth Circuit is granting relief in cases where other circuits would not.

The first possibility is difficult to swallow. Death penalty opponents for years lauded the California Appellate Project as a model for other states to emulate.⁽³²⁾ When the majority of the American Bar Association's Habeas Corpus Task Force issued its heavily defense-oriented recommendations, the "keystone" of the entire proposal was to require all the states to do largely what California had already been doing with regard to appointment and compensation of counsel.⁽³³⁾ The ABA Criminal Justice Section's sharpest criticisms were invariably directed at the South, not the West.⁽³⁴⁾ Thus, it seems extremely unlikely that the higher grant rate is the result of state courts in the Far West being of vastly lower quality than those of other regions of the country.

The other potential source of the problem is the Ninth Circuit itself. Is the high grant rate attributable to wrong or dubious decisions by that court?

In the previous CJLF study, I undertook to examine cases individually to see if either the state court or the federal court was simply making a wrong decision. This resulted in criticism of the study as "subjective" rather than "objective."⁽³⁵⁾ I believe the earlier methodology was valid and the "subjectivity" criticism overblown, but in the interest of keeping the debate focussed on habeas corpus rather than methodology, for this study I will adopt the approach of the opposition and simply count case results. To determine whether the exceptionally high grant rate in the Ninth Circuit is due to poor judging by state courts or by the Ninth Circuit, this study will simply count cases.

The rate at which the Ninth Circuit is reversed by the Supreme Court in both civil and criminal law has been noted for many years.⁽³⁶⁾ In the October 1993 term, for example, the Ninth Circuit was reversed, at least in part, in 12 cases and fully affirmed in only 2, for a reversal rate of 85%, far worse than any other circuit. That rate is also high in death penalty cases.

A computer search of Supreme Court opinions similar to the one described earlier turned up 25 capital cases from the Ninth Circuit or the states which comprise it, in which the high court decided the case on the merits. These cases include decisions on the merits of whether a last-minute stay was proper. The cases are listed in Table 3 at the end of this part.

There were a total of nine cases in which the Supreme Court granted certiorari to resolve a conflict between the Ninth Circuit and state courts. Eight of these were review of Ninth Circuit opinions,⁽³⁷⁾ and one was directly from a state court decision.⁽³⁸⁾ The Walton case involved four issues, so there were a total of thirteen issues in the nine cases. On thirteen issues, the Ninth Circuit was determined to be wrong on twelve, a stunning reversal rate of 92%.

If we broaden the scope to look at all 25 cases, we see that the state courts were reversed on six,⁽³⁹⁾ affirmed on nine,⁽⁴⁰⁾ and largely affirmed on one.⁽⁴¹⁾ Excluding the latter in the interest of being conservative, that is a 40% reversal rate. The Ninth Circuit was affirmed in only one case, Shuman, but was reversed in eight,⁽⁴²⁾ for an 88.9% reversal rate, more than double the rate of the state courts.

The sample is, of course, strongly affected by the Supreme Court's decision to review the case or not. A case is more likely to be reviewed if several justices believe it is wrongly decided, and thus discretionary review results in a higher reversal rate than mandatory review. Nonetheless, the Court's grants of certiorari are not limited to cases it believes are wrongly decided, as the state courts' 60% affirmance rate amply demonstrates. Nor does a denial of certiorari connote any approval whatsoever of the result in a case.⁽⁴³⁾ Thus, if eight out of nine of the Ninth Circuit cases in which certiorari was granted were wrongly decided, it stands to reason that a substantial number of the cases in which certiorari was denied reached wrong and unjust results.

The Ninth Circuit cases are also distinguished from the decisions of the other circuits by disagreement within the court. The four cases from other circuits granting relief are all unanimous panel decisions.⁽⁴⁴⁾ Two of the Ninth Circuit cases, Wade and Hamilton, are 2-1 reversals of a district court's denial of relief. Of the four federal judges to review each of these cases, half of them believed the state court was correct. In a third case, Blazak, the Ninth Circuit panel split three ways, with one judge believing the state court was correct, one believing the appeals court had no jurisdiction, and only one believing the state court was wrong. These cases hardly make compelling arguments that state courts are disregarding federal rights.

In short, the anomalously high grant rate of the Ninth Circuit makes a stronger argument against de novo review than for it.

E. Conclusions.

2. The rate at which any relief is granted remains high only in the Ninth Circuit, where the rate is anomalously high. The lack of any systemic deficiencies in state courts of this region, relative to other regions, the high rate of reversal of the Ninth Circuit by the Supreme Court, and the dissention within the Ninth Circuit all run counter to the notion that this high rate is the result of inadequacies of state remedies.

3. Excluding the Ninth Circuit, relief is granted in only 8% of habeas cases. The dramatic drop in recent years is not the result of changes in the law regarding procedural defaults or abuse of the writ, and it is only minimally related to changes in the law of retroactivity. The hypothesis most consistent with the data is that the high grant rates of prior years were primarily the result of uncertainty and rapid changes in the law, and that the rates have dropped as the law has clarified and stabilized.

4. The assertion that high grant rates indicate a systemic disregard for federal rights by state courts is no longer valid, if it ever was.

Please note: Table 2 is in three parts