The issue in this case involves the standard of review federal habeas courts must apply when reviewing a state court’s determination of harmless error. Davenport was partially shackled during his trial for first-degree murder. On direct appeal, the state appellate courts found that his partial shackling was unconstitutional, but was harmless beyond a reasonable doubt under the standard announced in Chapman v. California (1968).
Davenport subsequently filed a petition for a writ of habeas corpus in the U.S. District Court pursuant to 28 U.S.C. § 2254. Because Davenport’s partial shackling claim had been addressed by the state appellate courts, the federal habeas judge analyzed his petition pursuant to the provisions of the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA). The District Court judge denied habeas relief after finding that the state court’s harmless error determination was neither contrary to nor involved an unreasonable application of clearly established federal law (applying 28 U.S.C. § 2254(d)(1)). A divided panel of the Sixth Circuit Court of Appeals reversed, finding that the District Court applied the incorrect standard for addressing harmless error. The majority found that Brecht v. Abrahamson (1993), not AEDPA, supplies the correct standard, and based on Brecht alone, the shackling error was not harmless. The Michigan Attorney General’s Office filed a petition for a rehearing en banc. The Sixth Circuit denied rehearing by an 8-7 vote. Several judges provided written dissents from the denial rejecting the Brecht-only approach utilized by the panel majority. Instead, the state court’s harmless error determination was entitled to AEDPA deference and before a federal court can grant habeas relief, it must analyze the case pursuant to both AEDPA and Brecht
Earlier today Kent’s blog post (here) summed up the gist of what is at issue in this case. Essentially, if a state prisoner files a petition for a writ of habeas corpus claiming constitutional error in federal court, he or she is entitled to relief only if the error is not harmless. AEDPA requires the federal court to give great deference to the state court’s resolution of federal law, which includes a finding of harmless error. On the other hand, Brecht (a case that predates AEDPA by three years) held that federal courts need not give any deference to a state court’s harmlessness determination and could decide from scratch whether the claimed error affected the verdict.
It is CJLF’s position that because AEDPA was enacted to reduce delay and is a prerequisite to habeas relief, the habeas court must review the state court’s determination of harmlessness under AEDPA first. If the habeas petitioner satisfies the demands of AEDPA, he or she is not automatically entitled to relief, but rather Brecht must then be applied. In other words, if a federal habeas court is going to grant relief, both AEDPA and Brecht must be addressed and the Sixth Circuit erred when it skipped over AEDPA and went straight to Brecht to grant relief to Davenport.
As Kent’s blog post noted, in Fry v. Plyer (2007), a case in which the state court did not make a harmless error determination, Justice Scalia commented that “it certainly makes no sense to require formal application of both tests (AEDPA/Chapman and Brecht) when the latter obviously subsumes the former.” That “subsumes” language has caused a lot of problems regarding how federal habeas courts are to analyze cases like this one where the state court did make a harmless error determination.
At oral argument this morning, Justice Alito acknowledged that the “subsumed” language muddied the waters and perhaps it is time to dump it. Here is the full passage, on pages 38-39 of the transcript:
Well, maybe our ––maybe our opinions have confused things by introducing this concept of one subsuming the other. Why shouldn’t we just get rid of that?
AEDPA is a statute. It says in unequivocal terms you can’t grant federal habeas relief unless the decision is based on an unreasonable application of federal law defined in a certain way. Period.
There’s no way that federal relief, federal habeas relief, can be granted unless that is satisfied. So forget about what subsumes something subsuming the other. Brecht was an equitable decision. It continues to have force in a situation in which there isn’t a ––an applicable AEDPA provision, which is what Fry addressed when there wasn’t a harmlessness determination by the — by — by the state court. Isn’t that — doesn’t that simplify things? And is there anything wrong with it?
It does simplify things, and no, nothing is wrong with that analytical framework.
The Sixth Circuit took that “subsumes” language to find that if the habeas petitioner satisfies the high hurdle imposed by Brecht, then there is no reason to separately address the AEDPA question because it’s been “subsumed” by Brecht. At oral argument, Justice Kagan asked Davenport’s attorney:
“why not just tell courts that both have to be satisfied? You know, it seems like kind of a waste of pages and a kind — but, you know, just go through the motions, do it twice. . . .
But, I mean, I guess — I guess my question here is — is, if, one, you know, generally subsumes the other, but maybe contra-Ayala and contra-Fry we could imagine a case in which that wasn’t true, just have the courts go through both and we’ll be sure? . . .
And so why not just say, you know, you — you have to do it just so we’re sure that no errors are taking place and that AEDPA is being considered in the right way?”
Good question. Davenport’s attorney responded that federal habeas courts “can do it” that way, but it would seem “unnecessary” to do so. It is not unnecessary in cases where there is a state court decision on the merits because AEDPA is a prerequisite to habeas relief. So, yes – courts must “go through both” to make sure “AEDPA is being considered in the right way.” The Sixth Circuit’s leap frog over AEDPA straight to Brecht ignores the demands of the statute.
The post The Riddles of Harmless Error and Habeas Corpus – Part II appeared first on Crime & Consequences.