HEFFERNAN, CHIEF JUSTICE (concurring).

The writer joins in the majority opinion but concurs for the purpose of expressing disagreement with the separate concurring opinion of Justice Day, which states dissatisfaction with the harmless error rule recently adopted in State v. Dyess, 124 Wis.2d 525, 370 N.W.2d 222 (1985).

Quite aside from the broader policy reasons for not now switching to a new harmless error rule, I am troubled by the analysis which would discard the Dyess rule because of the dictionary's definitional difference between "possible" and "probable." Dyess was formulated for the express purpose of escaping the enslaving yoke of semantics. It would indeed be absurd to say that the words, "possible" and "probable," are always synonymous. But what we said in Dyess was that henceforth, for the purpose of the application of the test, meanings of those words should make little difference, that we should look to the substance of the test. We pointed out therein that "there are few meaningful distinctions among the various formulations of the test for prejudice." P. 545. Dyess attempted to stress a common-sense approach to whether or not an error seriously undermined confidence in a verdict. We drew on the experience of this court which taught us "that the determination of whether the trial error was harmless is unaffected by the particular formulation of the harmless error test used." P. 543.

I fear that what the concurring opinion urges is that we resort to a "counting-the-angels-on-the-head-of-a-pin" logic or to an undue reliance on Webster's Dictionary. I also point out that anyone who thinks that "possible" and "probable" are discrete words with no congruent meanings will be promptly disabused of this idea after even a cursory examination of some of the meanings ascribed to these words in Burton's Legal Thesaurus (MacMillan 1980).

While I sympathize with, and agree with, Justice Day's quest for an ever better harmless error rule, that quest should put aside the assumption that the solution to this vexing problem is to be found in dictionary definitions. Any rule adopted, or to be adopted, by this court or others will only confirm the appropriateness of the nomenclature used by Chief Justice Traynor: "The Riddle of Harmless Error." No test is likely to be completely satisfactory. What is harmless or prejudicial eventually comes down to a subjective judgment, no matter how carefully we attempt to hedge that subjective decision with recognizable guidelines. What this court needs is a consistent and predictable policy in handling questions of harmless error.

The effect of the concurring opinion is to place in doubt our present admittedly less than perfect harmless error rule. Perhaps we will now replace it with another formulation that will be equally subject to criticism and whose jurisprudential life will soon be terminated. Nothing in the briefs or oral arguments in the present case has suggested any dissatisfaction or argued for a change in Dyess. On the other hand, when Dyess was adopted, there was widespread and publicly acknowledged dissatisfaction with whatever harmless error rule we theretofore had. To my knowledge, there has been no widespread disagreement with the Dyess procedure.

Recognizing, as any reasonable person should, that there is indeed a "Riddle of Harmless Error," I believe the court should, in the effort to give some predictability and reckonability to this aspect of the law, stick with the present test of harmless error. Until or unless we can reach the conclusion, based on experience, that the Dyess rule, properly read, understood, and applied, fails to serve the purpose of seeing to it that criminal convictions are not set aside for trivial reasons but are set aside only where the right to a fair trial has been impaired, the Dyess rule should be preserved. The duty of this court to provide and assure some continuity and predictability mandates that there be no change in the rule unless we can, with some assurance, come up with a better rule that in likelihood can be applied with greater understanding and consistency by the lawyers and the judges of this state. What the concurrence appears to propose is a rule that compounds whatever confusion may arise out of the present rule. I would suggest that whatever rule we have, or which we may adopt in the future, assuming it substantially meets the dual test of foregoing reversals for trivial errors but yet assuring fair trial, we stick with the rule and work with it if necessary to see to it that it is made workable and understandable.

I am authorized to state that JUSTICE BABLITCH joins in this concurrence.

DAY, J. (concurring).

I concur with the results of the majority opinion. However, I write separately because the application of the "reasonable possibility" test under the facts of the instant case is especially disturbing. It makes little sense to equate the erroneous admission of evidence related to violation of a municipal ordinance with a violation of a constitutional right. I believe that the court's decision in Dyess to adopt a uniform test for constitutional and non-constitutional error was ill-advised and that the court should, at the first available opportunity, overrule Dyess. Although one of the Dyess majority, I would restate the law of harmless error in order to reinstate separate tests for constitutional and non-constitutional error.

While it is clear in the instant case that the error complained of did not affect the evidence of the trial under any test for harmless error, this type of error, i.e., non-constitutional error, should not be subject to the strict, "reasonable possibility" test of Dyess. This test is applied by the United States Supreme Court only to constitutional errors.

As noted, Dyess, abandoned distinctive tests for constitutional and non-constitutional error. Observing that "[t]here seems to be no special virtue in having different harmless error tests for constitutional error and non-constitutional error," a majority of this court stated:

"We conclude that, in view of the gradual merger of this court's collective thinking in respect to harmless versus prejudicial error, whether of omission or commission, whether of constitutional proportions or not, the test should be whether there is a reasonable possibility that the error contributed to the conviction." Dyess, 124 Wis.2d at 543. (Footnote omitted.) (Emphasis added.)

There appears to have been two reasons which prompted the Dyess court to conclude that a single harmless error test is appropriate. First, this court, as a matter of practice, has frequently resorted to a single test in cases involving both constitutional and non-constitutional error. Id. at 542.

The single test that has been used was originally formulated by the United States Supreme Court in Kotteakos v. United States, 328 U.S. 750 (1946), a case involving non-constitutional error. Kotteakos is generally cited for the United States Supreme Court's statement of the harmless error test for non-constitutional errors. Courts have relied on various statements of the test appearing in the Kotteakos opinion. The following is one of the key paragraphs:

"[i]f, when all is said and done, the conviction is sure that the error did not influence the jury, or had but very slight effect, the verdict and the judgment should stand, except perhaps where the departure is from a constitutional norm or a specific command of Congress. But if one cannot say, without fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the error, it is impossible to conclude that substantial rights were not affected. The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error. It is rather, even so, whether the error itself had substantial influence. If so, or if one is left in grave doubt, the conviction cannot stand." Kotteakos, 328 U.S. at 764-765. (Footnote and citation omitted).

The Dyess court noted that the Kotteakos test had been adopted by this court in prior cases involving constitutional error, citing State v. Poh, 116 Wis.2d 510, 529, 343 N.W.2d 108 (1984), State v. Burton, 112 Wis.2d 560, 571, 334 N.W.2d 263 (1983), and State v. Billings, 110 Wis.2d 661, 667, 329 N.W.2d 1982 (1983), as examples. 124 Wis.2d at 542.

The Dyess court also noted that in a number of cases involving non-constitutional error this court has concluded that the proper test was whether there was a reasonable possibility that the error contributed to the conviction, citing State v. Cartagena, 99 Wis.2d 657, 671, 299 N.W.2d 872 (1981), Barrera v. State, 99 Wis.2d 269, 295, 298 N.W.2d 820 (1980), Pohl v. State, 96 Wis.2d 290, 312, 291 N.W.2d 554 (1980), and Novitzke v. State, 92 Wis.2d 303, 308, 284 N.W.2d 904 (1979), as examples. The Dyess court reasoned that these cases, using a "reasonable possibility" test, were applying, for all intents and purposes, the Kotteakos test. Dyess, 124 Wis.2d at 542-543. Thus, the majority in Dyess concluded that this court has applied a single test (the Kotteakos test), in cases involving both constitutional and non-constitutional error.

The second reason given by the Dyess court as to why a single harmless error test is appropriate is that "there is little practical difference between the formulations of harmless error" which this court has used. Dyess, 124 Wis.2d at 543. This means that the conclusion as to whether harmless error exists is often "unaffected by the particular formulation of the harmless error test used." Id.

In arriving at its conclusion as to the correct test, the Dyess court took guidance from Strickland v. Washington, 466 U.S. 668 (1984). In Strickland, the court stated the test of prejudice which must be applied to determine whether reversal is required in cases of defective performance by counsel.

Under Strickland, the appropriate test for prejudice requires that: "the defendant. . . show that there is a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different. A reasonable probability is a probability sufficient to undermine confidence in the outcome." 466 U.S. at 694. The Strickland Court stated: "when a defendant challenges a conviction, the question is whether there is a reasonable probability that, absent the errors, the factfinder would have had a reasonable doubt respecting guilt." Id. at 695.

The Strickland Court set forth the defendant's burden:

"Taking the unaffected findings as a given, and taking due account of the effect of the errors on the remaining findings, a court making the prejudice inquiry must ask if the defendant has met the burden of showing that the decision reached would reasonably likely have been different absent the errors." Id. at 696.

The Dyess majority said that although Strickland phrased the test in terms of "reasonable probability," the United States Supreme Court's test "is substantively the same as ours." Dyess, 124 Wis.2d at 544.

As noted in Dyess, "[t]his court for years has been struggling with methodology to rationalize upholding a conviction despite the acknowledgement that error has been committed." Id. at 540. Prior to Dyess, the leading case on harmless error analysis was Wold v. State, 57 Wis.2d 344, 204 N.W. 482 (1973).

In Wold, a defendant convicted of sexual misconduct argued that erroneously admitted evidence constituted prejudicial error. The Wold court stated:

"The test of harmless error is not whether some harm has resulted, but, rather, whether the appellate court in its independent determination can conclude there is sufficient evidence, other than and uninfluenced by the inadmissible evidence, which would convict the defendant beyond a reasonable doubt." Wold, 57 Wis.2d at 356.

In arriving at this formulation of the test, the Wold court relied on three United States Supreme Court cases: Fahy v. Connecticut, 375 U.S. 85 (1963); Chapman v. California, 386 U.S. 18 (1967); and, Harrington v. California, 395 U.S. 250 (1969).

The Wold court noted that in Fahy, the harmless error test was stated in terms of "whether there is a reasonable possibility that the evidence complained of might have contributed to the conviction." Wold, 57 Wis.2d 344, n. 12, quoting Fahy, 375 U.S. at 86-87. Following the Fahy decision, the Court in Chapman held that "before a federal constitutional error can be held harmless, the court must be able to declare a belief that it was harmless beyond a reasonable doubt." Chapman, 386 U.S. at 24. Chapman is the case generally cited for the United States Supreme Court's statement of the harmless error test for constitutional errors.

The Wold court noted that the formulations of the test in Fahy and Chapman were thought to be similar by the Chapman Court. Wold, 57 Wis.2d at 357, n. 12, quoting Chapman, 386 U.S. at 24. The Wold court concluded: "[t]aking these two cases together, it seems the court is saying an error cannot be deemed harmless beyond a reasonable doubt where there is a reasonable possibility that the evidence complained of might have contributed to the conviction." Wold, 57 Wis.2d at 357, n. 12.

The Wold court reasoned, however, that the subsequent Harrington case "exchanged the `possibility' test of Fahy for one based on `probability.'" Id. Thus, the test was "based on reasonable probabilities." Wold, 57 Wis.2d at 356. In explaining the possibility/probability distinction, the Wold court noted: "[a] possibility test is the next thing to automatic reversal. In determining guilt `beyond a reasonable doubt,' the human mind should not work on possibilities, but on reasonable probabilities." Id. at 356-357. (Footnote omitted.) The Wold test was applied in a number of cases, and the statement of the test experienced some variation.

This court struggled with the Chapman "beyond a reasonable doubt" language as incorporated in the Wold test. In assessing the "degree of harm" caused by trial error, whether constitutional or not, the focus has typically been on the strength of evidence "untainted" by the error. In Sheehan v. State, 65 Wis.2d 757, 223 N.W.2d 600 (1974), the defendant was denied his constitutional right to confrontation when the trial court allowed a deposition to be read at trial. The court held the error harmless, concluding that there was sufficient untainted evidence to sustain the conviction. Sheehan, 65 Wis.2d at 767-768.

This approach was followed in State v. Dean, 67 Wis.2d 513, 227 N.W.2d 712 (1975), cert. denied, 423 U.S. 1074 (1976), which involved the admission of inadmissible hearsay evidence. The Dean court stated: "[w]e conclude that because of other evidence offered in the trial that the receipt of this testimony constituted harmless error." Dean, 67 Wis.2d at 533. A similar approach was used in Ramaker v. State, 73 Wis.2d 563, 243 N.W.2d 534 (1976), where constitutional error had been committed through the ex parte reception of evidence, giving the defendant no notice or opportunity to respond. The Ramaker court concluded: "[i]n short, the error is harmless because this court can independently conclude that there is sufficient evidence, other than and uninfluenced by the unconstitutionally received evidence, which would warrant the revocation of Mr. Ramaker's probation." Ramaker, 73 Wis.2d at 570.

Kelly v. State, 75 Wis.2d 303, 249 N.W.2d 800 (1977), highlighted the uncertainty surrounding the Wold harmless error test and marked the departure of one of the members of this court from the Wold test. The error in Kelly involved the admission into evidence of a gun case and a box of shells which had been found upon a warrantless search of the bedroom of the defendant, an individual convicted of first-degree murder. The court concluded that the admission of this evidence constituted harmless error, relying on evidence "uninfluenced" by the inadmissible evidence. Kelly, 75 Wis.2d at 316-317. In a footnote, the majority in Kelly remarked that: "[w]e decline to upset Wold in favor of Kotteakos v. United States, "but noted that "the improperly admitted evidence had `but only slight effect,' Kotteakos, 328 U.S. at 764 and the result in this case would be the same under that test." Kelly, 75 Wis.2d at 317, n. 3.

In his concurring opinion, Justice Heffernan doubted the correctness of the formulation of the harmless error rule in Wold, arguing that Wold erroneously placed "emphasis upon the admissible evidence rather than upon the alleged error." Id. at 321. Justice Heffernan noted that Wold should be reconsidered in light of United States v. Agurs, 427 U.S. 97 (1976), a case which sets forth a rule different from that adopted in Wold.

Justice Heffernan wrote the opinion in State v. Jennaro, 76 Wis.2d 499, 251 N.W.2d 800 (1977), where the defendant appealed his criminal conviction, in part, on the ground that the trial court erred in failing to grant defendant's motion for severance of defendant's trial from the trial of his co-defendant. The Jennaro court held the error to be harmless, offering the following statement on harmless error analysis:

"Under Chapman v. California, 386 U.S. 18 (1967), the United States Supreme Court stated that, before a federal constitutional error can be held harmless, `the court must be able to declare a belief that it was harmless beyond a reasonable doubt.' (at 24) This standard of Chapman was specifically reaffirmed in Harrington v. California, 395 U.S. 250, 254 (1969). Under the Chapman test, it is apparent beyond a reasonable doubt that the error complained of did not contribute to the guilty verdict in respect to Jennaro. Under the test apparently applicable to nonconstitutional error, set forth in Kotteakos v. United States, 328 U.S. 750 (1946), and recently restated in United States v. Agurs, 427 U.S. 97, 49 L.Ed.2d 342, 96 S.Ct. 2392 (1976), the error is equally harmless, because it is clear that the error did not influence the jury or had only slight effect. See Kelly v. State, 75 Wis.2d 303, n. 2.5 at 317, 249 N.W.2d 800 (1977), and concurring opinion at 321. Under any test, the error was harmless." Jennaro, 76 Wis.2d at 509-510.

Here the Jennaro court explicitly recognized that the Chapman test was to be used in instances involving constitutional error, while the Kotteakos test, focusing on whether the error influenced the jury or only had a slight effect, was to be used in assessing non-constitutional error. Despite the recognition of this distinction, the Jennaro court opted not to choose any single test, and concluded that the error was harmless "under any test." Id. at 510.

Jennaro was relied on in Simpson v. State, 83 Wis.2d 494, 266 N.W.2d 270 (1978), where the court reviewed the erroneous admission into evidence of two photos of a murder victim. The Simpson court stated: "[w]e conclude, applying either the test adopted in Wold, 57 Wis.2d [at] 356, or the test suggested in Jennaro, 76 Wis.2d [at] 509-510, that the error was not prejudicial." 83 Wis.2d at 507. See also, Hammen v. State, 87 Wis.2d 791, 801, 275 N.W.2d 709 (1979).

The "under any test" approach seemed to be modified somewhat in State v. Clark, 87 Wis.2d 804, 818, 275 N.W.2d 715 (1979), where the court held that the admission of testimony by the victim of a sexual assault that she had not had sexual intercourse prior to the incident in question constituted error. The court held that the error was "harmless under any test for nonconstitutional error." Clark, 87 Wis.2d at 818. (Emphasis added.) The Clark court again raised the distinction between constitutional and non-constitutional error. The court noted: "[t]he defendant invokes the standard for harmless constitutional error. See Chapman v. California, 386 U.S. 18, 24 (1967). The error here was admission of testimony in violation of state law; this in no way rises to the level of deprivation of the right to a fair trial guaranteed by the sixth and fourteenth amendments." Clark, 87 Wis.2d at 818, n. 3.

The "under any test" approach is also employed in State v. Sarinske, 91 Wis.2d 55, 280 N.W.2d 725; Novitzke v. State, 92 Wis.2d at 308 and Pohl. In Pohl, the court compared various statements of the harmless error test, and appeared to have given up making any distinction among them.

The distinction between the harmless error test for constitutional and non-constitutional errors arose again in State v. Billings, 110 Wis.2d 661, 329 N.W.2d 192 (1983). In Billings, this court stated that the standard for determining whether constitutional error is harmless was found in Chapman, 386 U.S. at 24. 110 Wis.2d at 666-667.

Following Billings, the court in State v. Gavigan, 111 Wis.2d 150, 163, 330 N.W.2d 571 (1983) identified the Wold test as the standard for determining whether a non-constitutional error is harmless. See also, State v. Ruiz, 118 Wis.2d 177, 198, 347 N.W.2d 352 (1984). The application of the Wold test was challenged by the dissenters in Gavigan. Justice Heffernan, in his dissent, stated: "[t]he Wold test is not whether there was sufficient evidence to convict without the tainted evidence, but whether the jury would have done so." Id. at 174. Claiming that the majority misapplied the Wold test, Justice Heffernan stated: "[t]his court in Wold clearly applied the test of harmless error in respect to its effect upon a jury and did not confine its scrutiny solely to the effect that the untainted evidence would have as regarded solely from the viewpoint of an appellate court." 111 Wis.2d at 174, n. 2.

The foregoing brief sketch of this court's treatment of the harmless error test since Wold shows that: 1) the distinction between tests for constitutional and non-constitutional error has not always been preserved; 2) the court has stated the harmless error test in various ways; 3) the "under any test" approach reveals not only imprecision in language, but frustration with the harmless error methodology; and 4) even when agreeing on the proper formulation, this court often has disagreement on the elements of the test, and how the test is to be applied. It was against this background that the Dyess court addressed the question of harmless error analysis.

The decision in Dyess to adopt a single uniform test for harmless error appears to have resulted, in part, from both dissatisfaction with this court's treatment of trial court errors in earlier cases, as well as a desire to set forth a clear rule that can be simply employed. The uniform test for harmless error had been advocated by Chief Justice Roger Traynor of the Supreme Court of California:

"Constitutional though it may be to impose different tests for errors that are not of constitutional dimension and errors that are, each new test brings new complications to appellate review, which may do less to serve justice than to impede it. For the long run it would be much better to evolve a uniform test that could realistically govern errors of constitutional or nonconstitutional dimension, in both civil and criminal cases. Since trials are often marred by both constitutional and nonconstitutional errors, an appellate court can hardly perceive the record whole if it must constantly adjust its sights to various degrees of probability as it scrutinizes various types of error. Such busy work not only befuddles its vision but causes a court to lose sight of the main objective. The real problem of justice is not whether an error is of constitutional or nonconstitutional dimension or whether it mars a criminal or a civil trial, but whether it affected the judgment." R. Traynor, The Riddle of Harmless Error, 48-49 (1970).

Wigmore also advocates a uniform standard. 1 J. Wigmore, Evidence, Sec. 21 (3d ed. 1940). See also, Comment, Harmless Error: The Need for a Uniform Standard, 53 St. John's L. Rev. 541 (1979). As noted by the author of the foregoing Comment:

"Although, as a general rule, constitutional violations may be more egregious than other types of errors, the circumstances of a particular case may render the nonconstitutional error far more prejudicial . . . . Since `the effect of the evidence in the context of [a] particular case' bears little relation to whether the error is constitutional or nonconstitutional, there seems to be little justification for a rule which makes the determination of harmlessness hinge on the label attached to the error." 53 St. John's L. Rev. at 559 (footnotes omitted).

The constitutional/non-constitutional distinction has been criticized on the ground that "it is erroneous to say that constitutional errors are more likely to have an injurious effect at trial than other errors." Saltzburg, The Harm of Harmless Error, 59 Va. L. Rev. 988, 1025 (1973). Professor Saltzburg argues for two distinct tests for harmless error, but the type of test should depend on whether the case is an appeal from a civil or criminal trial, and should not be related to the type of error involved. He states:

"The basis for requiring a different standard for judging the effect of a procedural error in criminal and in civil trials is that a different standard of proof is utilized in the two kinds of cases, and that the standard of proof utilized at trial should relate in large measure to the standard used by an appellate court in evaluating the impact or potential harm of any trial error." 59 Va. L. Rev. at 991.

In discussing the arguments in favor of a uniform test, Saltzburg notes that Wigmore advocated a uniform test due to a desire to reduce the number of criminal convictions overturned on appeal, and did not consider the different burdens of proof in criminal and civil cases and whether that difference should play a part in assessing error. Id. at 995. Saltzburg observes that Justice Traynor did recognize the "relationship between burden of proof in the trial court and the standards of review on appeal," but only with respect to questions involving sufficiency of evidence and not questions related to review of error. Id. at 995-996. Saltzburg concluded that "aside from abstract notions of symmetry or an unsubstantiated assumption that a single standard is more easily applied, [Justice Traynor] supplied no more justification for uniformity in both kinds of cases. Indeed there appears to be none." Id. at 997. (Footnote omitted.)

Since questions of harmless error often involve constitutional rights under the United States Constitution, the starting point, when considering any formulation of the harmless error test, must be with what is minimally required when dealing with federal constitutional rights. In Chapman, the United States Supreme Court stated:

"The application of a state harmless-error rule is, of course, a state question where it involves only error of state procedure or state law. But the error from which these petitioners suffered was a denial of rights guaranteed against invasion by the fifth and fourteenth amendments . . . . Whether a conviction for crime should stand when a State has failed to accord federal constitutionally guaranteed rights is every bit as much of a federal question as what particular federal constitutional provisions themselves mean, what they guarantee, and whether they have been denied." 386 U.S. at 21.

In adopting a uniform test for harmless error, a state must therefore construct the test so that it meets the United States Supreme Court's test for constitutional error. See, Comment, Confusion in the Court — Wisconsin's Harmless Error Rule in Criminal Appeals, 63 Marq. L. Rev. 643, 659 (1980). The test enunciated in Dyess, i.e., whether there is a reasonable possibility that the error contributed to the conviction, amounts to the Chapman harmless error test for constitutional error. The Chapman Court stated:

"There is little, if any, difference between our statement in Fahy v. Connecticut about `whether there is a reasonable possibility that the evidence complained of might have contributed to the conviction' and requiring the beneficiary of a constitutional error to prove beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained." Chapman, 386 U.S. at 24.

As stated previously, the Wold court concluded that, in Harrington, the United States Supreme Court abandoned the "possibility test" and exchanged it for a test based on "reasonable probabilities." An analysis of the Harrington case as well as subsequent United States Supreme Court opinions reveals that Wold was incorrect and that the Chapman test for constitutional error is still phrased in terms of reasonable possibilities.

In a recent case, Morris v. Mathews, 106 S.Ct. 1032 (1986), the United States Supreme Court dealt with a violation of the Double Jeopardy Clause. A five member majority held that "when a jeopardy-barred conviction is reduced to a conviction for a lesser included offense which is not jeopardy-barred, the burden shifts to the defendant to demonstrate a reasonable probability that he would not have been convicted of the non-jeopardy-barred offense absent the presence of the jeopardy-barred offense." Id. at 1038. In treating a violation of the Double Jeopardy Clause, the majority was not applying any harmless error test, since it did not consider the case to be a "`harmless error' case." Id. at 1037. In defining reasonable probability, the Morris majority looked to Strickland. Id. at 1038.

Four members of the court, however, agree that that case could have been treated as a "`harmless error' case." Id. at 1040 (L. Blackmun, J., concurring). The violation or error which occurred was constitutional in nature and thus the Chapman test should apply. Id. Those four Justices reaffirm that the Chapman "harmless beyond a reasonable doubt" standard is essentially the same as the Fahy "reasonable possibility" standard, and that it is the latter test — one of reasonable possibilities — that is appropriate in cases involving constitutional error.

It is clear that the uniform test adopted in Dyess is the traditional test for constitutional error. Applying this test to cases involving nonconstitutional errors, cases may be reversed in instances where a reviewing court may conclude that there was a reasonable possibility that a nonconstitutional error contributed to the result. This situation may be unacceptable given the original purposes of the harmless error rule and the purposes it continues to serve.

It does well to keep in mind the history behind the evolution of the harmless error rule. Wigmore expressed his frustration at the American courts' adoption of the "Exchequer rule," garnered from the early case of Crease v. Barrett, 1 C.M. R. 919, 149 Eng. Rep. 1353 (Ex. 1835), and the use of that rule in such a way so as to overturn cases on the most trivial technical errors. He notes:

"Whether in civil or criminal cases, it has done more than any other one rule of law to increase the delay and expense of litigation, to encourage defiant criminality and oppression, and to foster the spirit of litigious gambling. . . . That the law has gone to the extremes of absurd and provoking technicality in applying this rule is plain enough even in a casual glance through the reports. . . . Just so long as an erroneous ruling on evidence, however, trifling, is described by the highest judges (and in many courts it habitually is) as `working a reversal,' just so long will the reproach of technicality and futility mark our litigation." 1 J. Wigmore Evidence Sec. 21, p. 891 (Tillers rev. 1983).

See also, Mause, Harmless Constitutional Error: The Implications of Chapman v. California, 53 Minn. L. Rev. 519 (1969); Comment, Harmless Error: The Need for a Uniform Standard, 53 St. John's L. Rev. 541, 543 (1979).

This court should not employ a harmless error test which would result in the overturning of a great number of cases on the grounds that a nonconstitutional error could not pass a harmless error test which the United States Supreme Court has said only constitutional errors need be subjected to. As noted by Justice Traynor: "The practical objective of tests of harmless error is to conserve judicial resources by enabling appellate courts to cleanse the judicial process of prejudicial error without becoming mired in harmless error." Traynor, at 81.

I am persuaded that the purposes underlying the harmless error rule are best served by preserving distinctive tests for constitutional and non-constitutional error. There are two primary reasons for retention of the distinctive tests. First, courts must be concerned with the integrity of the outcome of trials. Determining whether an individual received a fair trial is essentially a matter of judgment. In distinguishing between constitutional and non-constitutional error, the underlying assumption is that constitutional errors should undergo greater scrutiny since a violation of a constitutional right would likely have a more injurious effect than commission of a non-constitutional error.

Second, constitutional rights are of fundamental importance and deserve heightened protection. See, Lane, 106 S. Ct. at 730, n. 9 (Court notes "heightened regard" for constitutional protections). It is harder to countenance the affirmance of a judgment which is based on constitutional error, not only because the constitutional error may have indeed contributed to the result based on the fundamental and pervasive nature of constitutional rights in general, but because upholding the result impliedly approves the violation of the constitutional right.

The foregoing reasons, while they suggest a high standard for constitutional error, also suggest that non-constitutional error should receive a less intense review. The likelihood that non-constitutional error contributed to or determined a verdict may be less, and the rights implicated are not of the same magnitude. Moreover, in applying the higher constitutional standard to non-constitutional errors, it is likely that a host of lesser technical errors may be swept in as having affected the determination of the verdict at trial. This result is at cross-purposes with the harmless error test and its underlying rationale: eliminating prejudicial error but not becoming bogged down in endless formulas for determining harmless error.

I disagree with the conclusion reached by the majority in Dyess that there is no difference between our use of the "reasonable possibility" test and the United States Supreme Court's use of the "reasonable probability" test:

"Although the [United States Supreme] Court uses the words `reasonable probability' of a different outcome, in contrast to our use of `reasonable possibility,' it is clear from the Strickland opinion that the Supreme Court's test is substantively the same as ours." Dyess, 124 Wis.2d at 544.

The meaning conveyed by the "reasonable possibility" and "reasonable probability" tests is not the same. Their meaning depends, in turn, on the meaning of the words "possible" and "probable." There is a distinction in meaning between these words.

Possible is defined in Webster's Ninth New Collegiate Dictionary 918 (1983):

"Possible. 1. to be able — more at POTENT; a: being within the limits of ability, capacity, or realization b: being what may be done or may occur according to nature, custom, or manners. 2. a: being something that may or may not be true or actual (explanation). 3. having an indicated potential (a housing site)."

Probable is defined in Webster's Ninth New Collegiate Dictionary 937 (1983) as:

"Probable. to test, approve, prove more at PROVE (1606). 1. supported by evidence strong enough to establish presumption but not proof (a hypothesis) 2: establishing a probability (evidence) 3: likely to become true or real (events)."

The United States Supreme Court has maintained a distinction between tests for constitutional and non-constitutional error. In Connecticut v. Johnson, 460 U.S. 73 (1982), Justice Stevens expressly stated that: "the harmless-error rule which may be applied when federal constitutional error has been committed, see Chapman v. California . . . is not to be confused with either the federal harmless-error rule that is applied in federal courts when non-constitutional error occurs, see Kotteakos . . . or with a State's own harmless-error rule applicable to errors of state law . . . ." 460 U.S. at 88, n. 2 (Stevens, J. concurring).

One of the key problems in dealing with harmless error tests is that there is a crucial difference between formulating the test into a concise statement and then crafting a practical approach which correctly implements that test. This problem was something that the Dyess court apparently thought best solved by the adoption of a single test. Commentators have noted that the United States Supreme Court has employed various approaches in harmless error analysis, despite the fact that the same test was being used. See, Note, Harmless Constitutional Error: An Analysis of Its Current Application, 33 Baylor L. Rev. 961 (1981); Field, Assessing the Harmlessness of Federal Constitutional Error — A Process in Need of a Rationale, 125 U. Pa. L. Rev. 15 (1976); Comment, Harmless Error: The Need for a Uniform Standard, 53 St. John's L. Rev. 541 (1979).

In the area of constitutional error, the United States Supreme Court appears to have employed three different approaches. As noted by one commentator, the Court has:

"(1) Focused exclusively on the erroneously admitted evidence to decide whether it contributed to the conviction;

"(2) Determined whether overwhelming evidence of guilt exists absent the constitutional error; and

"(3) Examined the record to decide whether the unconstitutionally obtained evidence is `merely cumulative' of some other admissible evidence." 33 Baylor L. Rev. at 962 (Footnote Omitted.)

As demonstrated earlier, this court has similarly employed various approaches in implementing harmless error tests.

In assisting a reconsideration of the harmless error test, it is helpful to consult recent pronouncements by the United States Supreme Court on the subject. The recent case of United States v. Lane, 106 S. Ct. 725 (1986) sheds light on the state of the Court's harmless error jurisprudence.

In Lane, a father and son, James and Dennis Lane, were indicted on multiple counts for mail fraud, conspiracy, and perjury in connection with insurance claims that were made and that insurers paid for fire damage to a restaurant and duplex that James had hired a professional arsonist to burn. Prior to trial, the Lanes made motions for severance, contending that the charged offenses were misjoined in violation of Federal Rules of Criminal Procedure, 8(b), but the motions were denied and the trial proceeded jointly before a jury.

On appeal, the Fifth Circuit Court of Appeals concluded that one of the counts should not have been joined with the others; it was held the misjoinder was prejudicial per se and the court reversed the convictions and remanded for new trials. 106 S. Ct. at 728-729. The Supreme Court granted certiorari to determine whether a misjoinder under Rule 8 is subject to the harmless error rule. Id. at 727.

The Lane Court notes that there are federal rules and statutes which deal with the topic of harmless error. Rule 52(a) of the Federal Rules of Criminal Procedure provides:

"RULE 52. HARMLESS ERROR AND PLAIN ERROR. (a) Harmless Error. Any error, defect, irregularity or variance which does not affect substantial rights shall be disregarded."

In addition, there is a "harmless error" statute, 28 U.S.C. § 2111, which provides:

"Sec. 2111. Harmless error. On the hearing of any appeal or writ of certiorari in any case, the court shall give judgment after an examination of the record without regard to errors or defects which do not affect the substantial rights of the parties."

In Lane, the Court relied on Rule 52(a) and sec. 2111 since "the specific joinder standards of Rule 8 are not themselves of constitutional magnitude." 106 S. Ct. at 730. (Footnote omitted.)

The Lane Court noted that, in Kotteakos, the Court construed a harmless error statute with language similar to that seen in Rule 52(a), and observed: "[t]he inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error. It is rather, even so, whether the error itself had substantial influence. If so, or if one is left in grave doubt, the conviction cannot stand." Lane, 106 S. Ct. at 732, quoting Kotteakos, 328 U.S. at 765. Lane invoked the Kotteakos test, holding that "an error involving misjoinder `affects substantial rights' and requires reversal only if the misjoinder results in actual prejudice because it `had substantial and injurious effect or influence in determining the jury's verdict.'" Id. at 732, quoting Kotteakos, 328 U.S. at 776. In assessing the effect of the error, the Lane Court appeared to adopt an "overwhelming evidence of guilt" standard. The court concluded, "[i]n the face of overwhelming evidence of guilt shown here, we are satisfied that the claimed error was harmless." Lane, 106 S. Ct. at 732.

The Lane majority agreed with the dissent that the harmless error inquiry is different from a sufficiency of the evidence inquiry, stating: "[b]ut that does not in any sense mean that overwhelming evidence of guilt is irrelevant; the threshold of overwhelming evidence is far higher than mere sufficiency to uphold conviction." Id. at 732-733, n. 13.

Although Lane dealt with non-constitutional error, the Court also discussed Chapman and the test for harmless error involving constitutional error. In addressing the dissenters, the majority notes that "the standard for harmless error analysis adopted in Chapman concerning constitutional errors is considerably more onerous than the standard for non-constitutional errors adopted in Kotteakos. . . ." Id. at 730, n. 9. (Emphasis added.)

In harmless error analysis, notes the Lane Court, "it is the duty of a reviewing court to consider the trial record as a whole and to ignore errors that are harmless, including most constitutional violations." Id., quoting, United States v. Hastings, 461 U.S. 499, 509 (1983). This review of the whole record is apparently required for all harmless error analysis, whether the error is constitutional or not. Lane, 106 S. Ct. at 731, n. 11.

Other recent cases in the United States Supreme Court show that the distinction between harmless error tests for constitutional and non-constitutional error has been maintained. In Van Arsdall, the error involved was of constitutional proportions, (the denial of defendant's opportunity to impeach a witness for bias which involved the confrontation clause) and the Court held the error to be subject to the Chapman harmless error analysis. 106 S. Ct. at 1438. The court outlined the nature of the test:

"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. Whether such an error is harmless in a particular case depends upon a host of factors, all readily accessible to reviewing courts. These factors include the importance of the witness' testimony in the prosecution's case, whether the testimony was cumulative, the presence or absence of evidence corroborating or contradicting the testimony of the witness on material points, the extent of cross-examination otherwise permitted, and, of course, the overall strength of the prosecution's case." Id.

The Chapman test was also employed in Hasting. In Hasting, the Court reviewed the reversal of respondent's convictions because of prosecutorial allusion to their failure to rebut the government's evidence. 461 U.S. at 500. The Hasting Court formulated the harmless error inquiry as follows: "[t]he question a reviewing court must ask is this: absent the prosecutor's allusion to the failure of the defense to proffer evidence to rebut the testimony of the victims is it clear beyond a reasonable doubt that the jury would have returned a verdict of guilty?" 461 U.S. at 510-511.

Harmless error analysis in Wisconsin should be based on the distinction between constitutional and non-constitutional error. The United States Supreme Court chooses to maintain this distinction, employing the Chapman test for constitutional error, and Kotteakos for non-constitutional error. I agree with this approach.

There are certain elements that the harmless error tests for constitutional and non-constitutional error have in common. First, in each test the reviewing court must take its own view of the record as a whole. Second, in assessing whether the error was harmless the court must determine the probable impact the error had in the minds of the jury. See, Schneble v. Florida, 405 U.S. 428, 432 (1972). Third, in looking at the impact that the error had on the jury, the court must imagine the impact on the hypothetical average jury, acting rationally.

The foregoing makes clear that the method by which a reviewing court determines whether error is harmless is to assess the effect the error had on the jury's deliberations. Since the jury's chief job entails weighing the evidence, the reviewing court must also look to the evidence presented.

In setting forth the methodology by which reviewing courts look at evidence in a harmless error case, the United States Supreme Court has articulated different tests. I would not distinguish among them but would instead adopt a combined approach. In reviewing the evidence presented, the court should look at the following factors:

1. Focusing on the error by itself, how much, if at all, did it contribute to the conviction? In cases of erroneously admitted evidence, the court would determine the importance of the evidence in the prosecutor's case.

2. Aside from the error, and any evidence that may have been admitted pursuant to it, is there overwhelming evidence supporting guilt?

3. Is there admissible evidence which would be considered cumulative to the evidence which may have been admitted as a result of the error?

Based on the foregoing determinations designed to help the court assess the significance of the error in the context of the evidence presented at trial, the court should then apply the standards for determining whether the error can be deemed harmless. Since the purpose of the harmless error test is to assess the probable impact of the error on the jury, the standard for both types of error is appropriately stated in terms of the degree of influence on the jury.

For constitutional error, the court should ask: Can it be said that the error had no more than minimal influence on the jury? If the error is thought to have more than minimal influence, then it is not harmless.

For non-constitutional error, the court should ask: Did the error substantially influence the jury? If it did substantially influence the jury, then it is not harmless.

As with other court-crafted tests, the tests here are not absolutely precise, nor scientific. However, they do provide a means to differentiate between the tests for constitutional and non-constitutional error.

The United States Supreme Court in Chapman stated the test for constitutional error as whether a reviewing court could hold that the error was harmless beyond a reasonable doubt. The test was restated as whether there is a reasonable possibility that the error contributed to the conviction. The minimal influence test mirrors the "reasonable possibility" and "harmless beyond a reasonable doubt" formulations of the test. It provides a means of measuring the effect of the error in a way that parallels the test for non-constitutional error: Both tests now assess the degree of influence on the jury.

The adoption of the minimal influence test would mean that errors which are trivial, inconsequential or insignificant are not grounds for holding the error not harmless. "Minimal influence" means that degree of influence which calls into question the basis upon which the jury reached its verdict, the fairness of the proceedings, and whether the violation of defendant's constitutional right should be tolerated. The minimal influence test is a lower standard than the substantial influence test since the intent would be to give greater protection to constitutional rights which are violated in the course of trial.

The constitutional error test is based on Chapman. The non-constitutional error test is based on the test enunciated in Kotteakos. The substantial influence test imposes a higher burden upon the party seeking to overturn a conviction on the grounds that there was a non-constitutional error committed. In applying this test, the reviewing court should determine whether the error was essential to the jury reaching its verdict.

The two distinct standards articulated above would preserve the difference between constitutional and non-constitutional errors, while at the same time serve the underlying purposes of the harmless error test.

I am authorized to state that JUSTICE DONALD W. STEINMETZ and JUSTICE LOUIS J. CECI join in this concurring opinion.

SHIRLEY S. ABRAHAMSON, J. (concurring in part).

The issue for which we took this case, namely, the state's complaint that the court of appeals erroneously reversed a conviction by improperly applying the harmless error test, does not, as the state candidly admitted in its petition to review (p. 3), meet any of the specific criteria set out in Rule 809.62(1), Stats. 1985-86, for granting a petition for review. As it turns out, however, the case has raised for the court a significant issue of law upon which the court is divided. The issue of law is the validity of one of the court's holdings in State v. Dyess, 124 Wis.2d 525, 370 N.W.2d 222 (1985), namely, that the test for harmless error does not turn on whether the error is of constitutional or non-constitutional magnitude (here after referred to as the Dyess issue). The state, the party involved in all criminal cases, has questioned the validity of the Dyess approach.

Unfortunately, the court decides the issue which does not meet the criteria for review and fails to decide the issue that does meet the criteria. Because the issue of harmless error arises in almost every criminal case in the circuit court, the court of appeals, and this court, I believe the court should decide the Dyess issue now and not leave it open, as it does in this case.

I recognize that the court raised the Dyess issue sua sponte in this case. The parties did not raise, brief, or argue the issue. I further recognize that this court should not decide cases on issues not raised, briefed or argued by the parties without giving the parties an opportunity to be heard of course in this case the court is deciding the case on the issue raised by the parties. It is therefore arguable that the court may decide the Dyess issue — which is not the determinative issue in the case — without affording the parties an opportunity to be heard. Nevertheless, I believe that the court would benefit from adversarial briefs on this important legal issue and that the court's resolution of the issue would, in all probability, be sounder with the benefit of counsels' assistance.

I conclude that the court should, as it has done in previous cases, order the parties to brief the issue the court has raised sua sponte and request amicus briefs so that the court can decide the Dyess issue promptly for the guidance of litigants, the circuit courts and the court of appeals. Accordingly, regardless of my position on the reversal of the decision of the court of appeals, I cannot join the majority opinion because it does not address what has turned out to be the major legal issue presented in the case.

WILLIAM G. CALLOW, J. (concurring).

Although I agree with the result reached by the majority, I write separately to express my concern with this court's prior adoption, in State v. Dyess, 124 Wis.2d 525, 370 N.W.2d 222 (1985), of a single test for harmless error. In Dyess this court adopted a rule recognizing there was but a single test for harmless error without that issue having been raised, briefed, or argued by the parties. Because of the circumstances under which the single test for harmless error was adopted, I believe the test for harmless error is ripe for review.

I do not join Justice Day's concurrence because I reserve judgment on the proper harmless error test until this court is presented with a case in which the validity of the Dyess harmless error test is properly raised.