146
Per Curiam
to a conviction at the conclusion of direct review. 507 U. S., at 633. It protects the State's sovereign interest in punishing offenders and its "good-faith attempts to honor constitutional rights," id., at 635, while ensuring that the extraordinary remedy of habeas corpus is available to those " 'whom society has grievously wronged,' " id., at 634 (quoting Fay v. Noia, 372 U. S. 391, 440-441 (1963)).
A federal court upsets this careful balance when it sets aside a state-court conviction or sentence without first determining that the error had a substantial and injurious effect on the jury's verdict. The social costs of retrial or resentencing are significant, and the attendant difficulties are acute in cases such as this one, where the original sentencing hearing took place in November 1981, some 17 years ago. No. C89-1906, App. to Pet. for Cert. A-101, n. 45. The State is not to be put to this arduous task based on mere speculation that the defendant was prejudiced by trial error; the court must find that the defendant was actually prejudiced by the error. Brecht, supra, at 637. As a consequence, once the Court of Appeals determined that the giving of the Briggs instruction was constitutional error, it was bound to apply the harmless-error analysis mandated by Brecht.
The Boyde test that the Court of Appeals applied instead is not a harmless-error test at all. It is, rather, the test for determining, in the first instance, whether constitutional error occurred when the jury was given an ambiguous instruction that it might have interpreted to prevent consideration of constitutionally relevant evidence. Boyde v. California, 494 U. S. 370, 377, 380 (1990). In such cases, constitutional error exists only if "there is a reasonable likelihood" that the jury so interpreted the instruction.
Although the Boyde test for constitutional error, like the Brecht harmless-error test, furthers the "strong policy against retrials years after the first trial where the claimed error amounts to no more than speculation," 494 U. S., at 380, it is not a substitute for the Brecht harmless-error test.
Page: Index Previous 1 2 3 4 5 6 7 8 9 10 11 12 NextLast modified: October 4, 2007