State v. Sayles (Dissent)

• Court: Supreme Court of Connecticut
• Decided: March 26, 2024
• Docket: SC20575
• Status: Published

Opinion

*********************************************** The “officially released” date that appears near the be- ginning of each opinion is the date the opinion will be pub- lished in the Connecticut Law Journal or the date it was released as a slip opinion. The operative date for the be- ginning of all time periods for filing postopinion motions and petitions for certification is the “officially released” date appearing in the opinion.

All opinions are subject to modification and technical correction prior to official publication in the Connecticut Reports and Connecticut Appellate Reports. In the event of discrepancies between the advance release version of an opinion and the latest version appearing in the Connecticut Law Journal and subsequently in the Connecticut Reports or Connecticut Appellate Reports, the latest version is to be considered authoritative.

The syllabus and procedural history accompanying the opinion as it appears in the Connecticut Law Journal and bound volumes of official reports are copyrighted by the Secretary of the State, State of Connecticut, and may not be reproduced and distributed without the express written permission of the Commission on Official Legal Publica- tions, Judicial Branch, State of Connecticut. *********************************************** STATE v. SAYLES—DISSENT

ECKER, J., dissenting. In conducting harmless error review of a constitutional violation, it is tempting for a reviewing court to take on the role of a thirteenth juror by reconstructing a hypothetical trial at which the tainted evidence was not admitted and then asking whether the properly admitted evidence is so strong that the court can be confident that it establishes the defendant’s guilt beyond a reasonable doubt. Our case law teaches that we must avoid this temptation because the inquiry asks and answers the wrong question. The correct question is whether there is a reasonable possi- bility that the improperly admitted evidence had a ten- dency to influence the judgment of the particular jury in the case before us. ‘‘The inquiry, in other words, is not whether, in a trial that occurred without the error, a guilty verdict would surely have been [returned], but whether the guilty verdict actually [returned] in this trial was surely unattributable to the error.’’ (Emphasis in original.) Sullivan v. Louisiana, 508 U.S. 275, 279, 113 S. Ct. 2078, 124 L. Ed. 2d 182 (1993). In my view we can only answer that question ‘‘no’’ in the present case. The state has not come close to providing the level of assurance required to find the alleged constitutional error harmless beyond a reasonable doubt. I base my conclusion principally on the lack of physical evidence connecting the defendant, Dwayne Sayles, to the charged crimes, the kind and quality of the state’s cir- cumstantial evidence, the highly inculpatory nature of some of the tainted evidence procured from the defen- dant’s cell phone in presumptive violation of the prophy- lactic rules created by Miranda v. Arizona, 384 U.S. 436, 478–79, 86 S. Ct. 1602, 16 L. Ed. 2d 694 (1966), and Edwards v. Arizona, 451 U.S. 477, 483–85, 101 S. Ct. 1880, 68 L. Ed. 2d 378 (1981), and the prosecutor’s heavy reliance at trial on that tainted evidence to persuade the jury of the defendant’s guilt. As a result, I believe that we must reach the constitutional issues certified by this court and briefed and argued by the parties on appeal.1 I LEGAL PRINCIPLES GOVERNING HARMLESS ERROR REVIEW The error that we are presuming is of constitutional magnitude. This means that the process by which the defendant was convicted and sentenced to eighty years of incarceration violated our most fundamental norms of justice. Under these circumstances, the burden prop- erly falls on the state to demonstrate that the error, despite its grave nature, nonetheless did not possibly affect the jury’s verdict and, therefore, was harmless beyond a reasonable doubt. This standard is ‘‘demanding . . . .’’ State v. Mangual, 311 Conn. 182, 212, 85 A.3d 627 (2014). ‘‘[W]e must examine the impact of the evi- dence on the trier of fact and the result of the trial. . . . If the evidence may have had a tendency to influence the judgment of the jury, it cannot be considered harm- less. . . . That determination must be made in light of the entire record [including the strength of the state’s case without the evidence admitted in error].’’ (Internal quotation marks omitted.) Id., 214–15. Harmless error review analyzes the impact of the constitutional error on the result of the trial, rather than on whether the jury arrived at a correct finding of guilt, because the United States constitution guarantees every defendant the right to a trial by the actual jury convened to hear the evidence. A criminal conviction cannot be based on the verdict of a hypothetical jury. The doctrine governing constitutional harmless error review is designed with this precise principle in mind. The leading case remains Chapman v. California, 386 U.S. 18, 87 S. Ct. 824, 17 L. Ed. 2d 705 (1967), which, as construed by the United States Supreme Court, prescribes the required analysis: ‘‘Consistent with the [jury trial] guar- antee, the question [that Chapman] instructs the reviewing court to consider is not what effect the constitutional error might generally be expected to have [on] a reason- able jury, but rather what effect it had [on] the guilty verdict in the case at hand. . . . [Harmless error] review looks, [the United States Supreme Court has] said, to the basis on which the jury actually rested its verdict. . . . The inquiry, in other words, is not whether, in a trial that occurred without the error, a guilty verdict would surely have been [returned], but whether the guilty verdict actually [returned] in this trial was surely unattributable to the error.’’ (Citations omitted; emphasis in original; internal quotation marks omitted.) Sullivan v. Louisiana, supra, 508 U.S. 279. Thus, ‘‘[t]he [harmless error] 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 v. United States, 328 U.S. 750, 765, 66 S. Ct. 1239, 90 L. Ed. 1557 (1946); see also State v. Mangual, supra, 311 Conn. 214 (‘‘[W]e must examine the impact of the evidence on the trier of fact and the result of the trial. . . . If the evidence may have had a tendency to influence the judgment of the jury, it cannot be considered harmless.’’ (Internal quotation marks omitted.)). It follows that harmless error review is not the same as sufficiency of the evidence review. See, e.g., Fahy v. Connecticut, 375 U.S. 85, 86–87, 84 S. Ct. 229, 11 L. Ed. 2d 171 (1963) (‘‘We are not concerned . . . with whether there was sufficient evidence on which the petitioner could have been convicted without the evi- dence complained of. The question is whether there is a reasonable possibility that the evidence complained of might have contributed to the conviction.’’); State v. Bruno, 197 Conn. 326, 336, 497 A.2d 758 (1985) (Shea, J., concurring) (‘‘Legal sufficiency of the evidence is not the test for harmless error even if only a nonconsti- tutional error is involved. The harmlessness of an error depends [on] its impact on the trier and the result . . . .’’), cert. denied, 475 U.S. 1119, 106 S. Ct. 1635, 90 L. Ed. 2d 181 (1986); see also State v. Torres, 343 Conn. 208, 245, 273 A.3d 163 (2022) (Ecker, J., dissenting) (‘‘the legal sufficiency of the evidence is not the issue, and the . . . marshaling of evidence sufficient to sup- port the conviction misapprehends the point of harm- less error analysis’’). Many courts and commentators have emphasized this important distinction. The following words of the Flor- ida Supreme Court summarize the point: ‘‘The test is not a [sufficiency of the evidence], a correct result, a not clearly wrong, a substantial evidence, a more proba- ble than not, a clear and convincing, or even an over- whelming evidence test. Harmless error is not a device for the appellate court to substitute itself for the [trier of fact] by simply weighing the evidence. The focus is on the effect of the error on the [trier of fact]. The question is whether there is a reasonable possibility that the error affected the verdict. The burden to show the error was harmless must remain on the state. If the appellate court cannot say beyond a reasonable doubt that the error did not affect the verdict, then the error is by definition harmful.’’ (Emphasis omitted; internal quotation marks omitted.) Ventura v. State, 29 So. 3d 1086, 1089–90 (Fla. 2010); see also State v. Gibson, 391 So. 2d 421, 427 (La. 1980) (‘‘[a]lthough the [harmless error] standard requires a reviewing court to consider the evidence in order to determine if there is a reason- able possibility that the error had prejudicial effect, it does not permit a court to substitute for the verdict its judgment of what the jury would or should have decided in the absence of error’’); State v. Alvarez-Lopez, 136 N.M. 309, 319–320, 98 P.3d 699 (2004) (‘‘[C]onstitutional error cannot be deemed harmless simply because there is overwhelming evidence of the defendant’s guilt. Our focus must remain squarely on assessing the likely impact of the error on the jury’s verdict.’’), cert. denied, 543 U.S. 1177, 125 S. Ct. 1334, 161 L. Ed. 2d 162 (2005). In sum, the proper standard to determine whether a constitutional error is harmless does not, as the major- ity states, assess whether the result of the trial would have been the same without the admission of the pre- sumptively improper evidence. The correct inquiry, rather, is whether we are assured beyond a reasonable doubt that the result of the trial would have been the same despite the admission of the presumptively improper evidence. The distinction between these two formulations is subtle but important. The former inquiry incorrectly focuses on the properly admitted evidence and the correctness of the jury’s verdict, whereas the latter inquiry correctly focuses on the improperly admit- ted evidence and its likely impact on the jury’s verdict. II THE TAINTED EVIDENCE AND THE STATE’S RELIANCE ON IT The majority accurately recounts the facts that the jury reasonably could have found, and I will not repeat those facts at any length here. I consider the majority’s legal analysis flawed because it alternatively under- states or altogether overlooks the full extent of the tainted evidence and the prosecutor’s heavy reliance on that evidence at trial. The proper analysis must begin by examining more closely the evidence extracted from the defendant’s cell phone, the fruits of that evidence,2 and the manner in which the prosecutor emphasized all of the presumptively inadmissible evidence during closing and rebuttal arguments to support an inference of guilt. The most damaging tainted evidence is the text mes- sages extracted from the defendant’s cell phone. Eman- uel Hatzikostas, a digital forensic examiner with the computer analysis response team at the Federal Bureau of Investigation (FBI), testified about these text mes- sages. According to Hatzikostas, at 3:55 a.m. on April 15, 2015, the date on which the search warrant for the residence of the defendant’s mother was executed during the early morning hours, someone texted the defendant, ‘‘[d]o not come here.’’ Approximately three hours later, at 7:48 a.m., someone texted the defendant, ‘‘M sai call his [s]hit now.’’ On that same date, the defen- dant drafted the following text message, which was never sent, to an unknown recipient: ‘‘If I get locked up tell sheema put them shits in the river some where worda loc.’’ It should come as no surprise that the content and meaning of these text messages became a focal point of the state’s case. The text messages are central to the harmlessness analysis, and I will return to them shortly, but it is important to understand at the outset that those mes- sages were by no means the only presumptively inad- missible evidence extracted from the defendant’s cell phone and presented to the jury. Multiple screenshots of a news article regarding the murder and robbery also were admitted into evidence. Specifically, the jury heard that, following the search of his mother’s residence on April 15, 2015, the defendant searched the news feed of WTNH, a local New Haven news station, for information regarding the crimes for which he was on trial. On that same date, the defendant also accessed a news article on Instagram pertaining to the crimes. This tainted evi- dence, largely unexamined by the majority, was relied on extensively by the prosecutor during closing and rebuttal arguments to buttress the state’s theory of guilt. The tainted evidence also included the fact that the name associated with the defendant’s cell phone was ‘‘Blackhead’’ and that the name associated with the Instagram account on the defendant’s cell phone was ‘‘Black Hoodie,’’ consistent with the testimony of the state’s witnesses, Leighton Vanderberg, Derrick Hoo- ver, and Jeremiah Samuels, regarding the defendant’s street names. A photograph of the defendant also was among the tainted evidence, depicting him wearing white sneakers, black jeans, and a shirt emblazoned with ‘‘Born Fly As F*ck.’’ The photograph not only por- trays the defendant as proudly transgressive, using pro- fanity to describe his nature, but, more important, shows him wearing the same color shoes and jeans worn by the perpetrator of the charged crimes.3 This is the kind of detail that can matter to a jury. The inadmissible contents of the defendant’s cell phone yielded additional probative evidence that the prosecutor used at trial to establish the defendant’s guilt.4 The unsent text message referencing ‘‘sheema’’ led the state to the defendant’s then girlfriend, Tys- heema Barker. Douglas Jowett, an inspector in the state’s attorney’s office, testified that he visited Barker to ‘‘[confront] her with the text message’’ and to serve her with a subpoena to testify at the defendant’s trial. Jowett explained that Barker was ‘‘irate’’ and that, not long afterward, the Department of Correction notified Jowett that she had gone to visit the defendant at the New Haven Correctional Center. Barker appeared at trial pursuant to the state’s subpoena and testified regarding her relationship with the defendant, the unsent text message, and her visit to the defendant at the cor- rectional center. Barker explained that she lived with the defendant in April, 2015, and that she never saw the defendant with any guns. Barker also testified that she never received the unsent text message drafted on the defendant’s cell phone and that the term ‘‘loc’’ refers to ‘‘somebody who passed away.’’ According to Barker, she visited the defendant for the first time since ‘‘[a]lmost [one] year ago’’ because she was very upset that she had been subpoenaed to testify at his criminal trial. During that visit, she informed the defendant that she had been asked about the unsent text message on his cell phone, and the defendant replied: ‘‘You’re straight. Just don’t worry. They’re going to try to get you mad.’’ The defendant also said about the unsent text message something to the effect of, ‘‘[y]eah, fuck that shit. That don’t mean anything . . . . You never got it. That don’t mean anything.’’ Barker acknowledged that, afterward, she and the defendant communicated nonverbally through the glass separating them during their noncon- tact visit so that a portion of their conversation would not be recorded by the Department of Correction. The testimony of Jowett and Barker was fruit of the pre- sumed constitutional violation, and, as set forth in detail in this opinion, the prosecutor relied on it extensively in closing argument to urge the jury to find that the ‘‘sheema’’ referred to in the unsent text message was Barker and that the ‘‘shits’’ was the gun used in the commission of the crimes with which the defendant was charged. The state relied heavily on the tainted evidence to prove its case in a manner significantly more extensive and rhetorically persuasive than acknowledged by the majority. The prosecutor used the contents of the defen- dant’s cell phone and the fruits of those contents during closing and rebuttal arguments not only to demonstrate the defendant’s participation in the commission of the crimes but also to lend credibility to the disparate bits and pieces of the state’s admissible evidence that either were unsupported, uncorroborated, or inherently suspect. The prosecutor made strong and effective use of the most powerful piece of evidence that presumptively was admitted in violation of the defendant’s constitu- tional rights, namely, the unsent text message on the defendant’s cell phone instructing his girlfriend, Barker, to throw ‘‘them shits in the river’’ if he were arrested. If the jury credited the prosecutor’s claim that ‘‘shits’’ meant ‘‘gun,’’ an extremely plausible interpretation, then the unsent text message was the functional equivalent of a confession by the defendant. The persuasive force of this evidence was obvious, as the prosecutor point- edly highlighted for the jury during initial closing argu- ment: ‘‘You heard from FBI analyst . . . Hatzikostas. . . . Hatzikostas testified that he was able to analyze a [cell] phone belonging to the defendant. If you recall, it had the [defendant’s] name . . . on it, Blackhead. He was able to retrieve a series of text messages from April 15, 2015. Now, if you recall, there is testimony that, in the early hours of April 15, [at] approximately 3 a.m., [the] New Haven Police Department executed a search warrant . . . [at] the defendant’s residence. In the series of text messages that [was] presented, there’s a text message that was received by the defen- dant’s phone at 3:55 a.m. on April 15, 2015, that says, ‘[d]o not come here.’ There was also testimony that, later that day, the same day, April 15, 2015, the defen- dant and his mother went to the New Haven Police Department to talk to detectives about why they were at . . . his house. In the same string of text messages, there’s a text message that . . . was not sent, although . . . Hatzikostas said there were numerous reasons why a text message may not be sent. Maybe [the cell phone] lost contact with the server, but there’s a text message that was typed on the defendant’s [cell] phone, ‘[i]f I get locked up tell sheema put them shits in the river some where [worda loc].’’5 The prosecutor also emphasized the significance of the unsent text message by tying the gun (the ‘‘shits’’) allegedly referenced in that message to the testimony of other witnesses, with the effect of providing a mutually reinforcing narrative. The prosecutor explained to the jury that Barker testified that ‘‘she also goes by the name Sheema. . . . She testified that, after [one] year of not going to visit . . . the defendant in jail, after she was served with a subpoena by . . . Jowett . . . she went to go visit him on January 27, 2018, a little [more than one] week ago. She testified that she knows those telephone calls while you’re in a face-to-face visit are recorded. She also testified that, when they want to talk about things that they don’t want recorded, they put the phone down and talk through the glass. She told you that she told the defendant that . . . Jowett had come to see her and that . . . Jowett ha[d] asked her about the text message, ‘[i]f I get locked up tell sheema . . . put them shits in the river’ or to lie. The defendant at that point told her not to worry.’’ The inference was both obvious and compelling: the suspi- cious and secretive communications between Barker and the defendant at the correctional center were fur- ther evidence that the unsent text message meant exactly what the prosecutor suggested it meant. Similarly, the prosecutor reminded the jury that there was evidence from both Samuels and Hoover that the defendant had stashed the gun used in the commission of the crimes at Barker’s residence. The prosecutor referenced Samuels’ prior inconsistent statement to the police, which was admitted into evidence for substan- tive purposes under State v. Whelan, 200 Conn. 743, 753, 513 A.2d 86, cert. denied, 479 U.S. 994, 107 S. Ct. 597, 93 L. Ed. 2d 598 (1986), ‘‘that the defendant had a girlfriend by the name of Tysheema, that he had a baby’s mother by the name of Imani, and that the guns were probably at either one of those locations.’’ The prosecu- tor later referenced Hoover’s testimony that ‘‘[t]he defendant told . . . Hoover that the guns were at [the defendant’s] girlfriend’s house [and] that, after the police raided his house, he moved them.’’ On the basis of this testimony, the prosecutor simultaneously tied together disparate threads of evidence and bolstered the credibility of the jailhouse informants by implicitly but unmistakably suggesting to the jury that the ‘‘shits’’ mentioned in the defendant’s unsent text message referred to the gun used in the commission of the crimes, consis- tent with the statement of Samuels and the testimony of Hoover. The prosecutor continued to hammer home the signif- icance of the tainted evidence during rebuttal argument to construct a compelling narrative of guilt. In one instance, the prosecutor argued that ‘‘April 15 is kind of an interesting time frame because . . . Vanderberg goes down April 14, he gives the information to [the] police, and then everything starts rolling. . . . The next day, the defendant [was] on the WTNH app . . . look- ing up the story. Pretty coincidental. And he’s looking up this story when he’s got nothing to do with it? He’s so concerned that he’s reading details in the afternoon time on April 15 about this Forbes [Avenue] gas station and homicide. . . . He got a series of texts, right? . . . And what were those texts? . . . [H]e got the first one right around the time the police [were] actually tossing his house, from someone to [the defendant], ‘[d]o not come here.’ Later that day, a text [message] that we know was never sent, but it was drafted on that phone, ‘[i]f I get locked up tell sheema put them shits in the river some where [worda loc].’ Shits can mean anything, right? People are constantly throwing stuff in the river. Does that make sense, or does it make more sense that he keeps his gun at his girlfriend’s house? He knows he’s going to get locked up for this offense, and he’s desperately trying to get the word out to get rid of those guns.’’ The prosecutor continued: ‘‘How about the conversa- tion with Sheema? They’re talking about the text mes- sage. She had met with . . . Jowett a couple [of] days earlier. Of course, she hasn’t seen [the defendant] in over [one] year, but she decides to go and pay him a visit days before she gets on the stand. And he says, ‘[t]hat don’t mean nothing. You never got it. That don’t mean nothing.’ He’s not worried at all. What’s he refer- ring to? Is he referring to the guns? Is he referring to the text message? I don’t know. It certainly shows knowledge of something though, doesn’t it?’’ The record, in short, establishes without question that the prosecutor’s use of the tainted evidence was extensive, integral to the state’s theory of guilt, and rhetorically effective. III HARMLESS ERROR ANALYSIS Applying the proper standard of review, I cannot con- clude on this record that the presumed constitutional error is harmless beyond a reasonable doubt. The incul- patory nature of the tainted evidence strikes me as self-evident. The timing of the text messages and the defendant’s searches for news articles was highly suspi- cious because they occurred on the same day and around the same time that his mother’s residence was being searched for evidence of the murder and robbery with which the defendant eventually was charged. The jury heard that, at the time the search warrant was executed at his mother’s residence, someone texted the defendant ‘‘[d]o not come here.’’ Additionally, soon thereafter, the defendant searched the WTNH news app and Instagram for information regarding the very crimes at issue. Even more damning is the unsent text message: ‘‘If I get locked up tell sheema put them shits in the river some where worda loc.’’ It is irrelevant to the analysis that this text message never was sent—the important point is that the message reflects the defen- dant’s own acknowledgment that he might get ‘‘locked up,’’ i.e., arrested, presumably for the crimes for which his mother’s residence was searched and for which he was soon to be charged. The jury readily could have inferred, as the prosecutor plainly argued, that the unsent text message captures the defendant himself acknowledging that there is inculpatory evidence of his participation in those crimes, namely, the ‘‘shits’’ that the defendant tells his girlfriend to throw into the river. This is strong evidence of guilt by any standard. The prosecutor knew that it was powerful and deployed it accordingly. The prosecutor invited the jury to infer that the word ‘‘shits’’ in the unsent text message referred to the gun used in the commission of the murder and robbery, which never was recovered. She implored the jurors to use their common sense and to find that there was no innocent explanation for the text: ‘‘Shits can mean anything, right? People are constantly throwing stuff in the river. Does that make sense, or does it make more sense that he keeps his gun at his girlfriend’s house? He knows he’s going to get locked up for this offense, and he’s desperately trying to get the word out to get rid of those guns.’’ The prosecutor expressed even greater incredulity that there could be an innocent explanation for the defendant’s Internet searches for news of the crimes: ‘‘The next day, the defendant [was] on the WTNH app . . . looking up the story. Pretty coincidental. And he’s looking up this story when he’s got nothing to do with it? He’s so concerned that he’s reading details in the afternoon time on April 15, about this Forbes [Avenue] gas station and homicide.’’ The prosecutor also suggested to the jury that an innocent man, one unconnected to the crimes with which the defendant was charged, would not be instructed ‘‘[d]o not come here’’ during the execution of a search war- rant. These are good arguments, and I cannot imagine being confident that they had no tendency to persuade the jury to reach a guilty verdict. The other evidence extracted from the defendant’s cell phone led directly to additional inculpatory evi- dence, which the prosecutor used to strengthen the case against the defendant. The jury heard that the unsent text message led the state to Barker, who, after being confronted with the text message, visited the defendant for the first time in approximately one year. According to Barker, the defendant was unconcerned about the unsent text message, purportedly saying to Barker: ‘‘Yeah, fuck that shit. That don’t mean anything . . . . You never got it. That don’t mean anything.’’ Through questioning and argument, the prosecutor used the inculpatory nature of the text message to sug- gest to the jury that Barker and the defendant continued their cover-up when the defendant and Barker inten- tionally communicated nonverbally so that their conver- sation would not be overheard by the Department of Correction: ‘‘How about the conversation with Sheema? They’re talking about the text message. She had met with . . . Jowett a couple [of] days earlier. Of course, she hasn’t seen [the defendant] in over [one] year, but she decides to go and pay him a visit days before she gets on the stand. And he says, ‘[t]hat don’t mean noth- ing. You never got it. That don’t mean nothing.’ He’s not worried at all. What’s he referring to? Is he referring to the guns? Is he referring to the text message? I don’t know. It certainly shows knowledge of something though, doesn’t it?’’ The prosecutor appealed to the jurors to use their common sense and experience to find that the defen- dant’s text messages, searches for news articles, and conversation with Barker reflected the defendant’s guilt and an attempt to cover up his commission of the crimes. This would have been entirely appropriate advo- cacy if the evidence had been admissible. See, e.g., State v. Courtney G., 339 Conn. 328, 347, 260 A.3d 1152 (2021) (prosecutor is permitted to ‘‘appeal to [the jurors’] com- mon sense in closing remarks’’ (internal quotation marks omitted)). But we are presuming in the present appeal that the evidence was improperly admitted in violation of the defendant’s constitutional rights, as guaranteed by Miranda v. Arizona, supra, 384 U.S. 478–79, and Edwards v. Arizona, supra, 451 U.S. 483– 85. The prosecutor’s pointed and repeated reliance on the tainted evidence during closing and rebuttal argu- ments undermines the majority’s conclusion that the admission of this evidence was harmless beyond a rea- sonable doubt. We previously have observed that a prosecutor’s ‘‘fre- quent and repeated emphasis on [inadmissible evidence] during [his or her] closing and rebuttal arguments indi- cates that [its] admission was not harmless.’’ State v. Culbreath, 340 Conn. 167, 195, 263 A.3d 350 (2021); see also State v. Ayala, 333 Conn. 225, 235, 215 A.3d 116 (2019) (‘‘in evaluating harm [we] look to see how the state used [the inadmissible] evidence in its closing argument’’); State v. Sawyer, 279 Conn. 331, 360–61, 904 A.2d 101 (2006) (finding harm because, among other reasons, state repeatedly emphasized improperly admit- ted evidence during its closing argument), overruled in part on other grounds by State v. DeJesus, 288 Conn. 418, 953 A.2d 45 (2008). Although we have not explicitly articulated our rationale for this principle, the logic is unmistakable. The prosecutor’s own selection of evi- dence to use during closing and rebuttal arguments at trial, chosen specifically for the purpose of persuading the jury of the defendant’s guilt, is among the very best indicators of what evidence would have had a tendency to impact the jury’s assessment of guilt. Why, after all, would the prosecutor repeatedly emphasize the con- tents of the defendant’s cell phone to persuade the jury of his guilt if that evidence was trivial, inconsequential, equivocal, or immaterial? If we are attempting to recon- struct what evidence reasonably was relied on by this jury to arrive at its verdict, and if our inquiry must strive (as much as possible) to avoid the risk of retrospective appellate hypothesizing, I would think that the state would have great difficulty meeting its burden of dem- onstrating harmlessness on appeal when, as here, the particular evidence that it now seeks to characterize as having had no tendency to persuade the jury was, in fact, evidence handpicked by the prosecutor pre- cisely for its persuasive force at trial and explicitly and repeatedly relied on by the prosecutor to convince the jury of the defendant’s guilt. The majority’s failure to adequately consider the effect that the prosecutor’s heavy reliance on the improp- erly admitted evidence likely had on the jury marks a departure from our usual harmless error analysis. Rather than examine what the prosecutor said and the likely impact it had on the jury, the majority focuses on the number of impermissible arguments in comparison to the permissible arguments, concluding that, ‘‘[v]iewed in context . . . the prosecutor’s focus on the unsent text message was minimal relative to the other evidence admitted at trial.’’ This analysis is flawed for two reasons. First, even if I were to accept the suggestion that the harmlessness analysis focuses on the relative quantity of references to the tainted evidence vis-à-vis the untainted evidence, and even if numerous references to the tainted evidence throughout the prosecutor’s arguments could be consid- ered insubstantial—both flatly erroneous propositions in my view—the tainted evidence in this case was far more extensive than just the unsent text message. It also included all of the other evidence found on the defendant’s cell phone and the fruits of the constitu- tional violation, such as the defendant’s searches for news articles regarding the crimes of conviction, the text message sent to him when his mother’s residence was being searched (‘‘[d]o not come here’’), and Bark- er’s testimony about her off-the-record conversation with the defendant regarding the unsent text message at the correctional center. The prosecutor relied on all of this evidence, at length, during her closing and rebut- tal arguments. Indeed, it is fair to say that the tainted evidence pervaded the prosecutor’s closing and rebut- tal arguments. Second, regardless of the number of references, when examining the likely effect that inadmissible evidence had on the jury, our review must be qualitative, not quantitative. See, e.g., State v. Van Kirk, 306 Mont. 215, 224, 32 P.3d 735 (2001) (when conducting harmless error review, appellate courts cannot ‘‘simply tally the quantity of the admissible evidence of guilt, [but] instead [must] evaluat[e] the qualitative impact the inadmissi- ble evidence might have had on the finder of fact’’ (emphasis in original)). That is to say, we must examine the quality and nature of the improperly admitted evi- dence in relation to the other evidence adduced at trial and carefully assess the manner in which the prosecutor highlighted that evidence to the jury and encouraged the jury to rely on it to find the defendant guilty. In my view, the inculpatory nature of the inadmissible evidence combined with the prosecutor’s pointed and repeated references to that evidence during closing and rebuttal arguments demonstrate why the state cannot meet its burden of proving harmlessness beyond a rea- sonable doubt.6 Even in the strongest of cases, it would be difficult to conclude that such damaging evidence had no possible tendency to affect the jury’s verdict. And this is not the strongest of cases. There was no physical evidence connecting the defendant to the crimes. For example, there were no impartial eyewitnesses who identified the defendant as a participant, there was no forensic or DNA evidence placing the defendant at the scene of the crimes, and there was no ballistic evidence connect- ing the defendant to the gun used in the commission of the crimes. The admissible evidence of the defendant’s guilt was derived predominantly from the testimony of convicted felons (Vanderberg, Hoover, and Samuels), jailhouse informants (Hoover and Samuels), and the defendant’s accomplice (Vanderberg). Although each of these wit- nesses implicated the defendant in the charged crimes, they suffered from serious credibility problems and had a powerful incentive, fueled by self-interest, to falsely implicate the defendant in the crimes charged. See, e.g., State v. Jones, 337 Conn. 486, 496, 254 A.3d 239 (2020) (Jailhouse informants have ‘‘a powerful incentive, fueled by self-interest, to implicate falsely the accused. Conse- quently, the testimony of such an informant . . . is inevitably suspect.’’ (Internal quotation marks omit- ted.)); State v. Patterson, 276 Conn. 452, 469, 886 A.2d 777 (2005) (‘‘[a]s the United States Supreme Court observed . . . ‘[t]he use of informers, accessories, accomplices, false friends, or any of the other betrayals which are ‘‘dirty business’’ may raise serious questions of credibility’ ’’), quoting On Lee v. United States, 343 U.S. 747, 757, 72 S. Ct. 967, 96 L. Ed. 1270 (1952). The testimony of jailhouse informants and accomplices regarding a defendant’s allegedly inculpatory admis- sions is inherently suspect, particularly damaging, and has a significant influence on conviction rates. See State v. Jones, supra, 502 (‘‘false confession evidence from informants is the leading factor associated with wrong- ful convictions in capital cases and a major factor con- tributing to wrongful convictions in noncapital cases’’ (emphasis in original; internal quotation marks omit- ted); J. Neuschatz et al., ‘‘The Effects of Accomplice Witnesses and Jailhouse Informants on Jury Decision Making,’’ 32 Law & Hum. Behav. 137, 146 (2008) (‘‘the presence of a secondary confession provided by a coop- erating witness ha[s] a strong influence on conviction rates when compared with the absence of such testi- mony’’). For this reason, the jury in the present case was instructed to ‘‘look with particular care at the testi- mony of’’ these witnesses and to ‘‘scrutinize it very carefully before . . . accept[ing] it.’’ The credibility of these witnesses plainly was important to the state’s case, and any evidence that tended to bolster their credibility likely would have affected the jury’s verdict. See State v. Fernando V., 331 Conn. 201, 223–24, 202 A.3d 350 (2019) (‘‘[when] credibility is an issue and, thus, the jury’s assessment of who is telling the truth is critical, an error affecting the jury’s ability to assess a [witness’] credibility is not harmless error’’ (internal quotation marks omitted)). The tainted evi- dence did just that—it corroborated the testimony of Vanderberg and Hoover that the defendant was involved in the commission of the crimes charged and tended to support their testimony that he hid the murder weapon at the home of his girlfriend. Given ‘‘[t]he com- monsense inference that corroborated statements tend to be true,’’ the inadmissible evidence necessarily enhanced the credibility of these crucial witnesses. State v. Fauci, 282 Conn. 23, 40, 917 A.2d 978 (2007). Further compounding the harm to the defendant is the fact that the accuracy and believability of the evi- dence from the defendant’s cell phone—particularly the unsent text message, ‘‘[i]f I get locked up tell sheema put them shits in the river some where worda loc’’— could not seriously be doubted once heard by the jury. The jury could not easily discount this evidence as self- serving, mistaken, or unreliable because, after all, it came from the defendant himself. See Zappulla v. New York, 391 F.3d 462, 473 (2d Cir. 2004) (inculpatory admissions from defendant are ‘‘the most probative and damaging evidence that can be admitted against him’’ because they ‘‘come from the actor himself, the most knowledgeable and unimpeachable source of informa- tion about his past conduct’’ (internal quotation marks omitted)), cert. denied, 546 U.S. 957, 126 S. Ct. 472, 163 L. Ed. 2d 358 (2005). Together, the tainted evidence and the testimony of Vanderberg, Hoover, and Samuels had a synergistic effect, which reinforced the state’s theory of guilt and likely affected the outcome of the jury’s verdict. The majority acknowledges ‘‘that the testimony of Vanderberg and Hoover is properly viewed with some skepticism, given their obvious self-interest in testifying for the state as an accomplice and a jailhouse informant, respectively,’’ but finds this evidence to be compelling in large part because the testimony of each was corrobo- rated by the testimony of the other and the prior incon- sistent statement of Samuels, another jailhouse informant. I cannot agree with the majority that the suspect testi- mony of a jailhouse informant ‘‘grow[s] in strength con- siderably’’ simply because it is repeated by another jailhouse informant or accomplice. In determining whether independent corroboration exists in the context of harm- less error review, we must exercise care that we do not abandon our well justified skepticism about the untrustworthy nature of testimony provided by accom- plices and jailhouse informants. The untrustworthy nature of this testimony exists because these witnesses have a powerful incentive, animated by self-interest, to falsely implicate the defendant in the crimes charged. I fail to understand how this powerful incentive is elimi- nated or even reduced by the fact that multiple wit- nesses all share the same motivation. Indeed, some states will not even permit the testimony of accomplices or jailhouse informants to come into evidence without corroboration by a source other than another accom- plice or jailhouse informant. See, e.g., State v. Harris, 405 N.W.2d 224, 227 (Minn. 1987) (‘‘[a]ccomplice testi- mony, it is clear, may not be corroborated solely by the testimony of another accomplice’’); People v. Ohlstein, 54 App. Div. 2d 109, 112, 387 N.Y.S. 2d 860 (1976) (‘‘[t]es- timony of each of several accomplices is not corrobora- tive of the other’’), aff’d, 44 N.Y.2d 896, 379 N.E.2d 222, 407 N.Y.S.2d 696 (1978); Chapman v. State, 470 S.W.2d 656, 660 (Tex. Crim. App. 1971) (‘‘it is a fundamental principle that the testimony of one accomplice witness cannot corroborate another accomplice witness’ testi- mony’’); see also Schnidt v. State, 357 S.W.3d 845, 851 (Tex. App. 2012, pet. ref’d) (standard for corroboration of accomplice and jailhouse informant testimony is same, and one cannot corroborate the testimony of other).7 We need not go so far in the present case, which does not involve the admissibility or sufficiency of evi- dence but, rather, the different question of whether the evidence demonstrates the harmlessness of the pre- sumed constitutional error beyond a reasonable doubt. The inherently dubious testimony of accomplices or jailhouse informants and accomplices simply cannot supply what the majority characterizes as ‘‘overwhelm- ing’’ evidence of guilt; nor is it sufficiently strong to remove the taint of the improper admission of the defen- dant’s cell phone and its contents. To the extent that the jury found the testimony of Vanderberg, Hoover, and Samuels credible, I consider it most likely that they were persuaded to believe this testimony only because it was corroborated by the highly inculpatory contents of the defendant’s own cell phone, which the defendant could not explain away. This is the very definition of harmful error. The cell site location data, which the majority charac- terizes as ‘‘the most powerful evidence corroborating the testimony of Hoover and Vanderberg,’’ appears to be fruit of the poisonous tree because the warrant used to procure that data was based on information obtained as a direct and proximate result of the presumed consti- tutional violation. See footnote 4 of this opinion. Regard- less, even if we were we to include the cell site location data in the harmless error analysis, it did not establish any of the essential elements of the crimes charged or that the defendant was in the actual location of the murder and robbery when those crimes occurred. At most, this evidence established that the defendant was in the general vicinity of the Forbes Avenue gas station approximately seventeen minutes beforehand. The defen- dant’s location at 7:33 p.m., the approximate time of the murder and robbery, is unknown. Indeed, the cell site location data indicate that the defendant was mov- ing in and around the city during the one and one-half hours between 7:00 and 8:30 p.m. At 7:15 p.m., he was in the Long Wharf area of New Haven, but, one minute later, at 7:16 p.m., he was across the Quinnipiac River in the area where the gas station on Forbes Avenue was located. More than one hour later, at 8:29 p.m., he was again across the Quinnipiac River in the area of New Haven where his mother’s residence was located. His movements between 7:16 and 8:29 p.m. are unclear, but what is clear is that the cell site location data do not disclose the defendant’s precise location at the time of the crimes charged. Moreover, given that the defen- dant’s cell phone pinged on two different cell towers in two separate areas of New Haven within less than two minutes, it is reasonable to infer that the defendant was moving around the city at the time and could have traveled a considerable distance in the seventeen minute time period between 7:16 and 7:33 p.m. It may be equally reasonable to infer that the defendant’s movements remained static during this time period and that he was in the vicinity of Forbes Avenue when the crimes occurred, but, under any scenario, the defen- dant’s presence in the general area of the crime is hardly powerful evidence of guilt because the imprecision of the cell site location data and the dense population of the city would at best make him one of countless potential suspects. The testimony of the impartial eyewitness, Jonathan Gavilanes, who observed two men fleeing the scene of the crime, does nothing to remove the doubt sur- rounding the state’s case. Gavilanes did not identify the defendant as one of the men and was unable to provide a detailed description of the perpetrators, aside from their clothing and height. As to their height, Gavilanes’ testimony was equivocal. Gavilanes, who is six feet tall, described the perpetrators’ height as ‘‘[a]bout five feet, seven inches [tall], almost my height.’’8 On cross-exami- nation, Gavilanes acknowledged that there is ‘‘a big difference between five foot, seven inches, and six feet,’’ and reiterated that the perpetrators were about his own height—six feet. On redirect examination, Gavilanes testified that he was ‘‘guessing’’ that the perpetrators were about five feet, seven inches tall. Given the incon- sistent nature of Gavilanes’ testimony and the fact that the average height of all men is between five feet, seven inches, and six feet tall, his description adds no strength to the state’s case. I likewise see no significant force added to the state’s case by the black ski mask and gloves found in the residence of the defendant’s mother. There was DNA evidence indicating that the defendant was a possible contributor—as were one out of every five other African Americans in the population. The majority candidly acknowledges the tenuous connection between the ski mask and gloves and the defendant, stating that ‘‘the DNA evidence standing alone is far from definitive . . . . ’’ I agree with that assessment. The connection between the black ski mask and gloves and the crimes charged is even weaker. There was nothing distinctive about the black ski mask and gloves used in the murder and robbery—no unique markings, patterns, or colors—to distinguish those items from the type or color of ski masks and gloves commonly found in a typical Connect- icut household. There is nothing unusual or inherently inculpatory in possessing a black ski mask and gloves. Given Connecticut’s cold winters, countless residents of this state have these items in their closets. Absent some evidence connecting the black ski mask and gloves found in the residence of the defendant’s mother with the black ski mask and gloves used in the commission of the crimes, I cannot conclude that this evidence constitutes physical evidence connecting the defendant to the murder and robbery. For the foregoing reasons, the properly admitted evi- dence was of a circumstantial nature and dubious or imprecise quality that does not come close to the kind of evidentiary showing that would remove or diminish the harmful effect that the improperly admitted evidence likely had on the jury’s verdict in the present case. Accord- ingly, I cannot conclude, beyond a reasonable doubt, that the alleged constitutional violation was harmless. I am not any more eager than the majority to reach the constitu- tional issues analyzed by the Appellate Court, certified by this court, and briefed and argued by the parties on appeal, but I do not believe that we can dispose of this case on the ground of harmless error. The defendant has a constitutional right to have a duly constituted jury deter- mine his guilt on the basis of lawfully obtained evidence. Absent waiver of this right, only a jury, not a panel of judges, can find the defendant guilty of the crimes charged beyond a reasonable doubt. As a result of this fundamental mandate, when an error of constitutional magnitude occurs at trial, the defendant is entitled to a new jury trial unless a reviewing court can say beyond a reasonable doubt that the error did not have a tendency to influence the jury’s verdict. I cannot reach that conclusion on this record, and, accordingly, I respectfully dissent. 1 We granted the defendant’s petition for certification to appeal, limited to the following issues: (1) ‘‘Did the Appellate Court properly uphold the trial court’s denial of the defendant’s motion to suppress the contents of his [cell phone] in reliance on United States v. Patane, 542 U.S. 630, 124 S. Ct. 2620, 159 L. Ed. 2d 667 (2004), and State v. Mangual, 311 Conn. 182, 85 A.3d 627 (2014), when the seizure of those contents was the result of questioning after he had invoked his Miranda rights, on the basis that a cell phone and its stored data constitute ‘physical’ (i.e., nontestimonial) evidence that need not be suppressed if seized as the result of a Miranda violation?’’ And (2) ‘‘[d]id the Appellate Court properly reject the defendant’s claim that the holding in Patane does not comport with the broader protections against compelled self- incrimination afforded under article first, § 8, of the Connecticut constitution?’’ State v. Sayles, 336 Conn. 929, 247 A.3d 578 (2021). Additionally, we granted the defendant permission to raise on appeal a fourth amendment claim challeng- ing the sufficiency and particularity of the search warrant used to obtain the contents of his cell phone pursuant to our recent decision in State v. Smith, 344 Conn. 229, 246–52, 278 A.3d 481 (2022). 2 ‘‘As a general principle, the exclusionary rule bars the government from introducing at trial evidence obtained in violation of the . . . United States constitution. . . . The rule applies to evidence that is derived from unlawful government conduct, which is commonly referred to as the fruit of the poisonous tree . . . .’’ (Citation omitted; internal quotation marks omitted.) State v. Bro- cuglio, 264 Conn. 778, 786, 826 A.2d 145 (2003). 3 The perpetrator also was wearing white sneakers and dark colored jeans. 4 It appears that the cell site location data, which were utilized by the prosecu- tion to track the defendant’s movements on the night at issue, also are fruits of the presumed constitutional violation. The record reflects that the search warrant used to obtain the call detail records and cell site location information, like the search warrant used to obtain the contents of the defendant’s cell phone, was based on information acquired as a direct and proximate result of the alleged violation of the defendant’s Miranda rights. Specifically, the warrant recites that, ‘‘[o]n April 15, 2015, [the defendant] came to [p]olice [h]eadquarters with his mother. Prior to any questioning, [the defendant] was given his Miranda [r]ights from a New Haven Police Department Miranda [w]aiver form. [The defendant] requested an attorney and no questioning took place. Prior to [the defendant’s] leaving, his mother handed to [the] detectives a [cell phone that] she said belonged to [the defendant] and provided [a certain ten digit number] as the phone number. The phone was seized and placed in an electronic protective bag to prevent remote erasure of data. A [s]earch and [s]eizure [w]arrant was obtained to retrieve data from within the [cell phone] . . . . Detectives were unable to gain access to the [cell phone] without the required passcode.’’ Just as we are presuming in the present appeal that the defendant’s cell phone and its contents should have been suppressed as fruits of the poison- ous tree because the phone was obtained from the defendant’s mother as a result of the alleged violation of the defendant’s Miranda rights, so, too, must we presume that the defendant’s cell phone number and the information obtained as a result of the procurement of the defendant’s cell phone number must be suppressed because they were obtained from the defendant’s mother in the same encounter pursuant to the same alleged Miranda violation. Because the information provided by the defendant’s mother is the only information in the warrant that links the defendant to the cell phone number used to generate the cell site location data, and the record plainly reflects that it was obtained ‘‘by exploitation of [the alleged] illegality,’’ it is fruit of the poisonous tree. (Internal quotation marks omitted.) Wong Sun v. United States, 371 U.S. 471, 488, 83 S. Ct. 407, 9 L. Ed. 2d 441 (1963). The majority does not address the substance of my assertion that the cell site location data are fruits of the presumed constitutional violation and affords the evidence ‘‘full consideration’’ in its harmless error analysis on the ground that the defendant abandoned any challenge to this evidence through inadequate briefing. Footnote 8 of the majority opinion. I cannot agree for three reasons. First, the defendant’s challenge to the cell site location data is subsumed within and intertwined with his claim that that the evidence obtained as a result of the violation of his Miranda rights improperly was admitted into evidence at trial. The police obtained his cell phone and cell phone number from his mother as the direct and proximate result of an interrogation that we presume was illegal, and precisely the same analysis applies to both. See, e.g., Meribear Productions, Inc. v. Frank, 340 Conn. 711, 732, 265 A.3d 870 (2021) (‘‘[w]e may . . . review legal arguments that differ from those raised by the parties if they are subsumed within or intertwined with arguments related to the legal claim before the court’’ (internal quotation marks omitted)). Second, the burden is not on the defendant to establish that the presumed constitutional violation was harmless; the burden instead rests on the state, and the state has failed to explain why the cell site location data can be treated any differently from the cell phone itself in this respect. Cf. State v. Jacques, 332 Conn. 271, 294, 210 A.3d 533 (2019) (reversing defendant’s conviction because state did not claim in its appellate brief that constitutional violation was harmless beyond reasonable doubt). Third, the overriding question that we must answer when conducting harmless error review, under any standard, is whether we have confidence in the fairness and integrity of the verdict despite the error, and answering this question requires a careful and searching review of the entire record. See, e.g., State v. Mangual, supra, 311 Conn. 214–15. Our appellate review must independently assess whether the state has met its heavy burden of establishing that a constitutional error is harmless beyond a reasonable doubt. See, e.g., State v. Alexander, 343 Conn. 495, 510, 275 A.3d 199 (2022) (conducting ‘‘our harmless error analysis . . . on the basis of our independent review of the record’’). This mandate does not authorize the court to ferret out new issues unrelated to the constitutional claims raised by the defendant, but, by the same token, I do not see how our independent assessment of the entire record can treat evidence as untainted when it comes from the very same poisonous tree that rendered the defendant’s cell phone and its contents inad- missible. 5 Defense counsel also referred to the unsent text message in his closing argument, stating, ‘‘[t]his text message [about] the shits . . . could mean any- thing. They want you to think, oh, it was guns, right? That’s what they want you to think. Who knows, right? It could be drugs. It could be stolen property. It could be anything. It doesn’t have to be guns. You don’t have to believe this because they say it. And [the unsent text message] never even gets sent.’’ 6 I agree with the majority that a prosecutor’s reliance on illegally obtained evidence in closing argument is not dispositive, but I find the cases cited by the majority to be distinguishable because, in those cases, the tainted evidence paled in comparison to the strength and quality of the properly admitted evi- dence, such that we could say beyond a reasonable doubt that the result of the trial would have been the same despite the admission of the tainted evidence and the prosecutor’s reliance on it. See, e.g., State v. Tony M., 332 Conn. 810, 823–25, 213 A.3d 1128 (2019) (improper admission of defendant’s statement that he tossed infant son off bridge in manner similar to ‘‘ ‘free throw’ ’’ was harmless beyond reasonable doubt as to defendant’s intent in light of defendant’s numerous text messages detailing his intent and testimony from multiple wit- nesses); State v. Jordan, 314 Conn. 89, 104, 101 A.3d 179 (2014) (improper admission of evidence of drugs found in closet was harmless beyond reasonable doubt as to issue of intent to sell because number of pills found on defendant’s person, combined with fact that these pills ‘‘contained identical markings to pills [the defendant] previously had sold, [was] consistent with a drug dealer soliciting repeat business’’). Importantly, in the cases relied on by the majority, the remaining evidence of guilt did not derive predominantly from ‘‘[t]he use of informers, accessories, accomplices, false friends, or any of the other betrayals [that] are ‘dirty business’ [and] may raise serious questions of credibility.’’ On Lee v. United States, 343 U.S. 747, 757, 72 S. Ct. 967, 96 L. Ed. 1270 (1952). 7 The majority contends that my reliance on State v. Harris, supra, 405 N.W.2d 227, People v. Ohlstein, supra, 54 App. Div. 2d 112, Chapman v. State, supra, 470 S.W.2d 660, and Schnidt v. State, supra, 357 S.W.3d 851, is misplaced because Connecticut has not adopted a corroboration rule for the testimony of accomplices and jailhouse informants. See footnote 10 of the majority opinion. The majority misapprehends my point. I do not intend to suggest that indepen- dent corroboration of accomplice and jailhouse testimony is necessary to adduce sufficient evidence of a criminal defendant’s guilt. As I previously explained, sufficiency of the evidence review is distinct from harmless error review, and the two types of appellate review should not be conflated. My argument, which the majority does not refute, is that the testimony of accomplices and jailhouse informants is inherently suspect and, absent objectively verifiable substantiation, is inadequate in the present case to satisfy the state’s burden of establishing that the presumed constitutional violation is harmless beyond a reasonable doubt. 8 The defendant is five feet, seven inches tall.