¶ 71. (dissenting). The majority correctly concludes that negligent treatment and failure to obtain informed consent in the context of chiropractic malpractice are two different issues that require separate verdict questions. Majority op., ¶¶ 2, 68. I therefore join that portion of the decision relating to the scope of the chiropractor's duty to obtain informed consent. I also agree with and join the majority's conclusion that the circuit court's failure to submit a special verdict on informed consent after separately instructing the jury on negligent treatment and informed consent constituted error. Majority op., part IVB. I write separately because I disagree with the majority's conclusion that the trial court's error in *705failing to submit a special verdict on informed consent was prejudicial. Majority op., part IVC. I therefore respectfully dissent.
¶ 72. The majority correctly notes that an error does not require reversal unless it affects the substantial rights of the party seeking to set aside the judgment. Majority op., ¶ 57. In determining whether the error is prejudicial or harmless, we apply the same test for harmless error in civil cases as in criminal cases. Town of Geneva v. Tills, 129 Wis. 2d 167, 184-85, 384 N.W.2d 701 (1986). Nevertheless, the harmless error test is not easy to articulate or apply. See State v. Hale, 2005 WI 7, ¶¶ 60-61 n.9, 277 Wis. 2d 593, 691 N.W.2d 637.
¶ 73. As I indicated in my concurring opinion in Hale, most constitutional errors1 are analyzed using the basic harmless error test set forth in Chapman v. California, 386 U.S. 18, 24 (1967), reh'g denied, 386 U.S. 987 (1967). Hale, 277 Wis. 2d 593, ¶ 111 (Butler, J., concurring). For these types of errors, the analysis begins with an evaluation of the type of error and the harm it is alleged to have caused in order to determine whether the error did not contribute to the verdict obtained beyond a reasonable doubt. Id. The appropriate standard for such errors is not whether there is sufficient evidence, absent the error, to support the *706verdict. Id. (citing State v. Weed, 2003 WI 85, ¶¶ 28-32, 263 Wis. 2d 434, 666 N.W.2d 485).
¶ 74. Some errors are considered "structural" in nature, and are considered so fundamental and pervasive that they require reversal without regard to the facts or circumstances of the particular case. Id., ¶ 110 (citing Chapman, 386 U.S. at 23 n.8; Delaware v. Van Arsdall, 475 U.S. 673, 681 (1986); Neder v. United States, 527 U.S. 1, 8 (1999)). These errors include a complete denial of counsel,2 a biased trial judge,3 racial discrimination in the selection of a grand jury,4 denial of self-representation at trial,5 denial of a public trial,6 and a defective reasonable-doubt instruction.7 Id. (citing Neder, 527 U.S. at 8.)
¶ 75. A yet third classification of harmless error involves the type of error that by its very nature lends itself to a form of an "outcome determinative" approach. See Id., ¶¶ 112-13. This third class of error includes ineffective assistance of counsel8 and errors in jury instructions.9 Id. For such errors, the court looks at whether there is a reasonable probability that the jury verdict would have been the same absent the error.10 Id., ¶ 112. A reasonable probability is one sufficient to *707undermine confidence in the outcome. Id. See also United States v. Bagley, 473 U.S. 667, 682 (1985).
¶ 76. I conclude that the failure to submit a special verdict on the question of informed consent after separately instructing the jury on both negligent treatment and informed consent falls within the third class of harmless error. In looking at the nature of the error and the harm it is alleged to have caused, the failure to give a special verdict after instructing a jury on a specific issue more closely approximates the type of error in failing to instruct a jury in the first instance than other types of trial errors. In both situations, the jury has been deprived of certain information upon which to base its decision. Since we cannot look subjectively into the minds of the jurors11 when an error occurs in the verdict form, I would apply the harmless error analysis adopted for jury instruction errors by the United States Supreme Court in Neder and by our court in State v. Harvey, 2002 WI 93, ¶ 47, 254 Wis. 2d 442, 647 N.W.2d 189. I would approach the inquiry by looking at whether the force of evidence, presumably considered by the jury in accordance with the instructions, was so overwhelming that we must conclude that the verdict would have been the same had the jury received the proper verdict form. Hale, 277 Wis. 2d 593, ¶ 113 (citing Yates v. Evatt, 500 U.S. 391, 404-05 (1991), (Butler, J., concurring).
¶ 77. This is not the first time our court has had to determine whether a jury verdict could be supported when the jury was faced with alternative methods of proof. In State v. Crowley, 143 Wis. 2d 324, 422 N.W.2d *708847 (1988), a jury found the defendant guilty of aggravated battery of a disabled person. Id. at 327. However, the State attempted to prove its case on alternate grounds, and it was not clear on which ground the jury convicted. First, the State tried to establish that the defendant's conduct directly created a high probability of great bodily harm. Id. at 328. Alternatively, the State contended that because the victim had a physical disability, if the evidence established the defendant's intent was to inflict bodily harm on a disabled person, then the defendant's conduct presumptively created a high probability of great bodily harm. Id.
¶ 78. This court concluded that when alternative methods of proof resting upon different evidentiary facts are presented to a jury, it is necessary for an appellate court to conclude that the evidence was sufficient under both of the alternative modes of proof in order to uphold the verdict. Id. at 329. While the court recognized that it was unclear which mode of proof the jury relied upon, the court nevertheless reasoned that it was obliged to search the record in an effort to determine whether the evidence was sufficient under each mode of proof. Id. at 331, 334. The court summarized the rule to be applied as follows: "[T]he rule .. . 'requires a verdict to be set aside in cases where the verdict is supportable on one ground, but not on another, and it is impossible to tell which ground the jury selected.'" Id. at 334-35 (quoting from United States v. Sales, 725 F.2d 458, 459 (1984)). See also Yates, 354 U.S. 298; Zant v. Stephens, 462 U.S. 862 (1983); Stromberg v. California, 283 U.S. 359 (1931). The court ultimately concluded that there was sufficient evidence under either method or mode of proof to sustain the jury's verdict. Crowley, 143 Wis. 2d at 345.
*709¶ 79. Similarly, in State v. Caldwell, 154 Wis. 2d 683, 454 N.W.2d 13 (Ct. App. 1990), petition for review denied, 457 N.W.2d 324 (unpub. table decision) (1990), the court of appeals was confronted with a conviction for obstructing an officer. The trial court instructed the jury that the crime of obstructing could be committed in alternative ways: by making more difficult the performance of the officer's duties or by knowingly giving false information to the police with intent to mislead. Id. at 690. Citing Crowley, the court of appeals reasoned that for the resulting conviction to stand, the evidence would have to be sufficient under both modes of proof. Id. at 691. The court ultimately determined that the evidence was sufficient for the jury to convict under either theory of obstruction and affirmed the conviction. Id. at 692, 695.
¶ 80. Neither Crowley nor Caldwell was decided using a harmless error analysis. Nevertheless, both decisions provide an analytical framework that fits well under the harmless error test. Like this case, both cases involve a general verdict even though the jury was presented with alternative grounds for reaching its decision. We are not presented with a situation where the verdict is supportable on one ground, but not another, and it is impossible to tell which ground the jury relied upon. Instead, by determining whether there was sufficient evidence under either method of proof, this court necessarily reaches the question of whether the evidence is so overwhelming that we must conclude that the verdict would have been the same had the jury received the proper verdict form. See Hale, 277 Wis. 2d 593, ¶ 113 (citing Yates, 500 U.S. at 404-05), (Butler, J., concurring).
¶ 81. Ample evidence was offered at trial that Hannemann received negligent treatment from Boyson *710following a chiropractic adjustment to his lower back and neck. Subsequent to the treatment, Hannemann developed numbness and tingling in his leg and under his foot. After complaining of the numbness, Boyson performed a second adjustment. Ultimately, Hanne-mann was unable to move one side of his body and experienced urinary tract problems. He was admitted to the emergency room and diagnosed as having suffered a stroke. Hannemann's expert witness testified that the treatment performed by Boyson fell below the chiropractic standard of care because Boyson did not perform a series of diagnostic tests that would indicate whether a patient was susceptible to neurovascular injury. While that evidence was disputed, we must resolve sufficiency of the evidence questions in the light most favorable to the verdict. State v. Burkman, 96 Wis. 2d 630, 643, 292 N.W.2d 641 (1980). I conclude that there was overwhelming evidence to support the jury's verdict with respect to the negligent treatment alternative.
¶ 82. Regarding informed consent, the evidence was undisputed that Boyson did not warn Hannemann that chiropractic treatment carried a risk of stroke or other neurovascular injuries. Hannemann's expert witness testified that there were no informed consent forms in the records. The expert testified that informed consent should include a warning to the patient that one of the risks of injury of chiropractic treatment includes stroke. Once again, while Boyson disputed the evidence, overwhelming evidence is present in the record that supports the jury's verdict with respect to the informed consent alternative.
¶ 83. Because the evidence is sufficient under either method of proof, I conclude that the error in failing to submit a special verdict to the jury on the *711question of informed consent is harmless. I would therefore reverse the court of appeals, and affirm the trial court. Accordingly, I respectfully dissent.
Examples of such errors include the following: Confrontation Clause violation (Delaware v. Van Arsdall, 475 U.S. 673, 684 (1986)); illegally seized evidence (Fahy v. Connecticut, 375 U.S. 85 (1963), and Gilbert v. California, 388 U.S. 263 (1967)); the right to consult with counsel (Satterwhite v. Texas, 486 U.S. 249 (1988)); involuntary confessions (Arizona v. Fulminante, 499 U.S. 279 (1991)); and comments on a defendant's silence (Chapman v. California, 386 U.S. 18 (1967), reh'g denied, 386 U.S. 987 (1967)).
Johnson v. United States, 520 U.S. 461, 468 (1997).
Tumey v. Ohio, 273 U.S. 510 (1927).
Vasquez v. Hillery, 474 U.S. 254 (1986).
McKaskle v. Wiggins, 465 U.S. 168 (1984).
Waller v. Georgia, 467 U.S. 39 (1984).
Sullivan v. Louisiana, 508 U.S. 275 (1993).
Strickland v. Washington, 466 U.S. 668 (1984).
Neder v. United States, 527 U.S. 1 (1999); Yates v. Evatt, 500 U.S. 391 (1991); and Pope v. Illinois, 481 U.S. 497 (1987).
While the burden of proof is normally on the beneficiary of the error in harmless error cases, it shifts to the defendant in *707ineffective assistance of counsel cases. Compare Chapman, 386 U.S. at 24, with Strickland, 466 U.S. at 694.
See Yates, 500 U.S. at 404-05.