Friday, February 16, 2007

Harmless Error and Confrontation

It is clear that Confrontation Clause errors are subject to harmless error analysis. Delaware v. Van Arsdall, 475 U.S. 673 (1986). But it seems to me that there is a significant limitation on how the harmless-error inquiry may be conducted: If a testimonial statement is admitted against an accused without his having had an opportunity for confrontation, the court should not speculate that cross-examination would have been ineffective. On the contrary, the court should address the harmless-error issue by assuming that cross, if it had not been denied, would have been maximally effective.

Where prosecution testimony has been admitted in violation of the confrontation right because the accused never had an opportunity to be confronted with the witness, the reviewing court should consider the case as if that testimony had never been admitted, and declare the error harmless if it is confident beyond a reasonable doubt that even without that evidence the accused would have been found guilty. Van Arsdall, supra, 475 U.S. at 682 n.5 (describing Harrington v. California, 395 U.S. 250 (1969): “Noting that the wrongfully admitted evidence was cumulative and that the untainted proof of the defendant's guilt was overwhelming, the Court concluded that the error was harmless beyond a reasonable doubt.”). One might complain about the exercise that it involves some guesswork about what the jury would have done, but basically it is a practical and necessary judgment: “The tainted evidence is of trivial importance, and a jury that found the accused guilty with that evidence would not fail to do so without that evidence.”

But suppose that after considering the untainted evidence in this way the court cannot say with the requisite confidence that the outcome would have been the same without the tainted testimony. Then it is not appropriate for the court to say, in effect, “The error is still harmless, because if there had been an opportunity for cross-examination of that witness it would have done little or any good.” To do that would be not merely to assess what the jury likely would have done with a slightly altered body of evidence but to speculate what an absent witness would have said if he had faced the adverse questioning that the Constitution guarantees the accused. The Confrontation Clause is meant to prevent the need for speculation, by inducing the prosecution to produce the witness for confrontation.

Van Arsdall supports this interpretation. There, the accused had not been denied the confrontation right altogether; he had cross-examined the witness in question, but the trial court had erroneously prevented him from pursuing a line of questioning that might have revealed his bias; thus, the immediate error was not admitting prosecution evidence but excluding defense evidence. Extending Harrington to this situation, the Court held that the error was subject to harmless-error analysis, and so remanded to the state supreme court, perhaps with a hint that because the testimony of the witness in question was cumulative and peripheral it would be entirely proper to deem the error harmless. The Court said: “The correct inquiry is whether, assuming that the damaging potential of the cross-examination were fully realized, a reviewing court might nonetheless say that the error was harmless beyond a reasonable doubt.” (emphasis added)

Thus, the reviewing court must act under the assumption that the accused gets the full “damaging potential of the cross-examination.” It is appropriate for the reviewing court to say, “Even under that assumption, the error doesn’t matter, because the most that cross could have done would have been to nullify the testimony of this witness, and the case was still overwhelming.” Possibly, but much more dubiously, it is appropriate for the court to say, “Even if the cross-examination that the accused says he would have conducted had gone entirely his way, achieving its full damaging potential by showing the witness was willing to lie in favor of the prosecution, the testimony would have had some value because even a dishonest witness could not likely have told this story if it were not true, and given that the error is harmless beyond a reasonable doubt.” But what the court should not do is say, “The error is harmless because if the accused had had an opportunity for cross-examination the answers of the witness would in all probability have been unhelpful to the accused.”


Arti Rolfingsmeier said...

This reminds me of another aspect of a case on which I'm working. That is the Strickland v. Washington standard of ineffective assistance of counsel. In my case, the issue was whether the outcome would have been different had the attorney at trial effectively cross-examined a law enforcement witness. The witness was available at the preliminary hearing, but not at trial. It seems the harmless error rule is becoming the rule in cases everywhere when deep analysis is not preferred by the courts.

Heather said...

It can't really have success, I feel so.