OPINION ON REMAND A jury found appellant guilty of murder and assessed punishment at twenty years’ confinement. This court initially affirmed his conviction, rejecting his contention that the admission of his historical cell site location information (CSLI) violated the United States and Texas Constitutions. The Court of Criminal Appeals reviewed this court’s holding under the Texas Constitution and affirmed. See Hankston v. State, 582 S.W.3d 278, 279 (Tex. Crim. App. 2019) (citing Hankston, 517 S.W.3d 112, 121–22 (Tex. Crim. App. 2017)). Appellant petitioned for a writ of certiorari from the United States Supreme Court. Id. While the petition was pending, the Court decided Carpenter v. United States, 138 S. Ct. 2206 (2018), holding that a defendant has an expectation of privacy under the Fourth Amendment in at least seven days of historical CSLI. Hankston, 582 S.W.3d at 279 (citing Carpenter, 138 S. Ct. at 2217). The Supreme Court vacated the Court of Criminal Appeals’ judgment in this case and remanded for further consideration. Id. at 280. The Court of Criminal Appeals then vacated this court’s judgment and remanded for this court “to reexamine its Fourth Amendment holding in light of Carpenter.” Id. We agree with the parties that appellant’s CSLI was erroneously admitted because appellant’s Fourth Amendment right to privacy was violated by a warrantless search. Because the admission was erroneous under the Texas exclusionary statute, the analysis for harm is based on non-constitutional error. Considering the entire record, we hold that the erroneous admission of appellant’s CSLI affected his substantial rights. Thus, we reverse the trial court’s judgment of conviction and remand for a new trial. I. Fourth Amendment Violation The parties agree that the State conducted a warrantless search of appellant’s historical CSLI, which exceeded seven days of CSLI and violated appellant’s Fourth Amendment right against unreasonable searches. We agree with the parties. See id. (citing Carpenter, 138 S. Ct. at 2217). Because the source of the alleged error in this case is the admission of evidence that was obtained in violation of a federal constitutional right, we must next determine whether the federal or Texas exclusionary rule applies to require the exclusion of the evidence. II. Federal vs. Texas Exclusionary Rule The federal exclusionary rule is a judicially created remedy that serves as a deterrent sanction to bar the State from introducing evidence obtained in violation of the Fourth Amendment. See Perez v. State, 495 S.W.3d 374, 390 (Tex. App.— Houston [14th Dist.] 2016, no pet.). Exclusion under the federal rule has always been the courts’ last resort, and various principles constrain the application of the exclusionary rule. See Herring v. United States, 555 U.S. 135, 140 (2009). For example, the federal exclusionary rule does not apply to a search performed in good-faith reliance on a statute that is later declared unconstitutional. Id. at 142 (citing Illinois v. Krull, 480 U.S. 340, 349–50 (1987)). Although the federal exclusionary rule applies in Texas, our Legislature has provided for an exclusionary rule that does not incorporate all the federal exceptions. See Garcia v. State, 829 S.W.2d 796, 798 (Tex. Crim. App. 1992); see Tex. Code. Crim. Proc. art. 38.23(a) (“No evidence obtained by an officer or other person in violation of any provisions of the Constitution or laws of the State of Texas, or of the Constitution or laws of the United States of America, shall be admitted in evidence against the accused on the trial of any criminal case.”). The State’s good-faith reliance on a statute in conducting a warrantless search is not a recognized exception to the Texas exclusionary statute. See Perez, 495 S.W.3d at 391. At the suppression hearing in this case, the trial court found that police officers obtained appellant’s CSLI by “legally proper means” because a court had issued an order that complied with state and federal statutes for obtaining CSLI.[1] As the State acknowledged below, those statutes did not require a finding of probable cause but instead a reasonable belief that the records would be relevant to a legitimate law enforcement inquiry or ongoing criminal investigation. Appellant does not challenge the trial court’s finding that the search was “legally proper” under the statutes. And it does not appear from the record or appellant’s briefing that he contends that the evidence in this case should have been excluded under the federal exclusionary rule. He has relied solely on the Texas statute, Article 38.23. We agree with appellant that Article 38.23 required exclusion of the evidence. The trial court erred by failing to suppress evidence related to appellant’s CSLI. III. Harm Analysis for Non-Constitutional Error The parties agree that the erroneous admission of appellant’s CSLI should be reviewed for harm under the standards for constitutional error. See Tex. R. App. P. 44.2. But, it is this court’s duty to assess harm, not the litigants’ burden. See Thomas v. State, 505 S.W.3d 916, 926 (Tex. Crim. App. 2016). And this court is not bound by the State’s concession regarding an issue of law. See Oliva v. State, 548 S.W.3d 518, 520 (Tex. Crim. App. 2018); see also Hill v. State, 426 S.W.3d 868, 876–77 (Tex. App.—Eastland 2014, pet. ref’d) (assessing whether error was constitutional or non-constitutional despite State’s apparent concession that it was constitutional); Kelly v. State, No. 06-13-00009-CR, 2013 WL 6254956, at *2–3 (Tex. App.—Texarkana Dec. 4, 2013, no pet.) (mem. op., not designated for publication) (assessing non-constitutional error for harm despite State’s concession that error was reversible). After the parties filed their briefs on remand, the Court of Criminal Appeals issued its opinion in Holder v. State, 639 S.W.3d 704 (Tex. Crim. App. 2022), which overruled part of its prior decision in Love v. State, 543 S.W.3d 835 (Tex. Crim. App. 2016). In Love, the court held that the defendant had a reasonable expectation of privacy to the contents of his text messages under the Fourth Amendment to the United States Constitution. See id. at 844–45. The court held that the State’s use of the same statutes involved here—authorizing disclosure to the State by court order without first obtaining a warrant supported by probable cause—violated the defendant’s rights under the Fourth Amendment. See id. The court held that the text messages should have been excluded from evidence under Article 38.23, rather than the federal exclusionary rule. See id. at 845. The court, nonetheless, analyzed the error for harm as if it were constitutional error rather than statutory error. See id. at 846. This final part of the Love opinion, the court reasoned in Holder, was incorrect. See Holder, 639 S.W.3d at 707. When the trial court’s error is the admission of evidence in violation of Article 38.23, rather than the federal exclusionary rule, an appellate court should review the harm based on the standards for non-constitutional error. See id. Accordingly, we reject the parties’ contentions that the error in this case was constitutional for purposes of Rule 44.2 of the Texas Rules of Appellate Procedure. We will review for harm under the standards applicable to non-constitutional error. A. Standards A non-constitutional error must be disregarded unless it affects a defendant’s substantial rights. Tex. R. App. P. 44.2(b); Barshaw v. State, 342 S.W.3d 91, 93 (Tex. Crim. App. 2011). A substantial right is affected when the error had a substantial and injurious effect or influence in determining the jury’s verdict. Thomas v. State, 505 S.W.3d 916, 926 (Tex. Crim. App. 2016). On the other hand, if the error did not influence the jury, or had but very slight effect, the verdict and the judgment should stand. Id. The focus is not on whether the outcome of the trial was proper despite the error, but whether the error had a substantial or injurious effect or influence on the jury’s verdict. Barshaw, 342 S.W.3d at 93–94. A conviction must be reversed if this court has “grave doubt” that the result of the trial was free from the substantial effect of the error. Id. at 94. “‘Grave doubt’ means that ‘in the judge’s mind, the matter is so evenly balanced that he feels himself in virtual equipoise as to the harmlessness of the error.’” Id. (quoting Burnett v. State, 88 S.W.3d 633, 637 (Tex. Crim. App. 2002)). In assessing the likelihood that the jury’s decision was improperly influenced, we must consider the entire record, including the nature of the evidence supporting the verdict, the character of the alleged error and how it might be considered in connection with other evidence, the parties’ theories of the case, arguments, voir dire, and whether the State emphasized the error. See Thomas, 505 S.W.3d at 927, Barshaw, 342 S.W.3d at 94. B. The Record 1. Voir Dire During voir dire, the State discussed the topic of a child testifying as a witness and how to assess the credibility of a child. The State asked: What about—what about corroborating evidence? Does anyone know what corroborating evidence is? Kind of like circumstantial clues. Would that be important to you in determining whether or not a child would—or to what they’re saying is credible, if it’s corroborated by other facts in the case? Several of the venire members agreed that corroborating evidence would be important to them. One of them was selected as a juror. 2. Opening Statements The State. The State began its opening statement by discussing the decedent’s child, who was present during the murder and testified as an eyewitness in the case. The State noted that the child was “the only witness that saw this defendant” at the scene. The State discussed how the child identified appellant in a photo spread, continuing: You’re going to hear from [the child], who was the only witness that can identify this defendant in this case. But this case is not just an eyewitness case. This is a circumstantial evidence case. The State emphasized, “The motive in this case is huge.” And the State argued that there were “other pieces to the puzzle,” and the jurors would “have to be the detectives, and get all the different puzzle pieces and put them together.” The State discussed these other puzzle pieces: You’re going to hear about a vehicle that was observed, that matches the vehicle driven by the defendant. You’re going to see cell phone records, and cell site information that you’ll learn about, all about and pinpoint where the phone is at what time. You’re going to hear that he had the opportunity, because he was there, he was in the area. And two minutes after this murder is committed, he’s leaving the area, based on his cell phone. At the same time, the car that matches his car was seen heading down the street right after the gunshots. The State told the jury, “You’re going to have to put it all together. And with the witness testimony and the circumstantial evidence combined, it’s going to lead you to a guilty verdict.” Appellant. Appellant’s counsel argued that the child’s identification was unreliable based on several factors discussed in greater detail below. Counsel also detailed evidence regarding other suspects. 3. Non-CSLI Evidence Some of the evidence is discussed in this court’s prior opinion. See Hankston v. State, No. 14-13-00923-CR, 2015 WL 3751551, at *1–4 (Tex. App.— Houston [14th Dist.] June 16, 2015), vacated, 582 S.W.3d at 280. But we will again discuss the non-CSLI evidence in the context of the entire record. Identification. The decedent was shot while answering the front door. His nine-year-old son was in the home at the time and identified appellant at trial as the shooter. The child testified that he saw appellant through a window for about three or four seconds. It was dark outside, but the porch light was on. When asked if the child got a good look at the shooter, the child testified, “Not very good, but enough to—for me to see his face.” He testified that the first thing he saw was the gun, but then he immediately looked at the person’s face. He was afraid at the time. His mother (the decedent’s wife) testified that the child had told her that evening, “[H]e saw the guy, but he didn’t know what his face looks like.” The child later told the police that the shooter was bald, wearing a tank-top, and did not have tattoos. Other evidence showed that near the date of the murder, appellant’s hair was shaved on the sides but not on the top, and he had tattoos on his arms and shoulders. The child told police that there were two other people with the shooter, but he didn’t know for sure at trial if there were other people with the shooter. A sketch artist worked with the child to create a composite sketch of the suspect; the sketch depicted someone who was completely bald. About six months after the shooting, and after the child did not identify different suspects in two previous photo spreads, the child identified appellant in a third photo spread. He testified that it took him twenty or thirty minutes to identify appellant, although a police officer testified that it took only three or four minutes. At the time of the photo-spread identification, the child told the officer that he was “pretty sure” the man he identified was the shooter. He clarified at trial that what he meant by “pretty sure” was, “I’m sure this is the guy that I saw at the window.” The officer testified that he created the photo spread and showed it to the child while the child’s mother was present. The officer acknowledged that, since the child made this identification, the procedures for administering photo spreads had changed to have the person who prepares the photo spread be different from the person who administers it, so it is done in a “blind” fashion. The officer testified that the procedures have changed to reduce the likelihood of misidentification. The jury also heard from appellant’s expert witness, a psychology professor, about various factors that can cause misidentifications. The expert noted that in cases in which DNA has been used to exonerate wrongfully convicted persons, more than three-fourths of those convicted were because of inaccurate eyewitness identifications. The expert identified factors in this case that could undermine the reliability of the child’s identification: Children don’t remember faces as well as adults and are much more likely to make false identifications compared to adults. There was an extended length of time between the murder and the photo spread identification. The child saw the shooter for only a brief period. There could have been “weapon focus,” where a witness tends to remember the face of the perpetrator less than the weapon due to the fear or anxiety of seeing a weapon. In a non-blind photo spread identification, where the person administering the photo spread knows which person is the suspect, there can be unconscious cues to the witness that causes them to select a suspect. The mother’s presence during the photo spread identification, an emotionally invested witness, may have signaled, consciously or unconsciously, for the child to select someone from the photo spread. There was a potential for “misinformation effect” because, before the child identified appellant, the child may have been impacted by (1) the composite sketch drawing and (2) the fact that the first photo spread showed all bald men, the second photo spread showed all men with hair, and the third photo spread showed all men with some hair, thus indicating the child should select someone with hair rather than trust his original memory. The trial court allowed one of the State’s police-officer witnesses to testify as an expert about eyewitness identifications The officer testified that corroborating evidence is a “very important factor with eyewitness identifications” because in the “vast majority of DNA exoneration cases, where eyewitness testimony has been involved, most of those cases involved an uncorroborated eyewitness ID.” He identified some of the evidence in this case that corroborated the child’s identification: Appellant’s vehicle matched the description of a vehicle seen near the scene of the murder. Appellant had a motive to murder the decedent. A frequency analysis of appellant’s phone records indicated that a significant event occurred with appellant around the time of the murder. Vehicle. A neighbor told the police that they saw a 2000 or 2006 burgundy Honda Civic being driven near the scene of the murder. The neighbor did not witness the murder, nor did he testify at trial. Appellant was known to drive a 2002 maroon or burgundy Honda Accord. Motive. A police officer testified that he believed appellant had a “very compelling” motive for the murder: That the victim in the case had been harassing and stalking [appellant's] girlfriend over an extended period of time, ultimately culminating in [the decedent] showing up at her apartment on the night that he was killed, approximately 45 minutes before he was shot. The decedent had shown up at the girlfriend’s apartment on the evening of the murder and knocked on her door. The girlfriend called the police. Her subsequent interaction with police officers upset her because it did not seem to her like they were taking the situation seriously. The girlfriend testified that appellant did not seem upset or bothered by the knocking incident. But, appellant wasn’t happy with the police officers because they told him to be quiet and not talk. He was upset but not irate; nothing led her to believe appellant was going to kill the decedent. She never knew him to own a gun. Appellant’s friend, who was also appellant’s barber, testified that appellant was real laid back and mild mannered. The friend had never seen appellant get riled up. Location. The girlfriend testified that she went to her parents’ house on the same street as the decedent’s house before the murder. Appellant arrived within minutes of her. When she heard gunshots, however, she was not sure where appellant was at the time. She testified that she and appellant left her mother’s house together, about twenty minutes later, and went to appellant’s mother’s house. She testified that she was not sure if they rode in the same car, but she was sure they went together because they walked in the door at the same time. She testified that it was difficult to remember what took place, exactly what time, and who was there. Phone Records. The State adduced evidence concerning the call and message activity of appellant’s cell phone, irrespective of the CSLI. The 911 call for the murder occurred at 9:32 p.m. A police officer testified that there was a six-minute period immediately preceding the murder in which appellant’s phone had no activity. Then, there was a “huge flurry of activity, a burst of activity,” with thirty-eight communications in a twenty-three-minute period. For the seven months’ worth of phone records that the officer reviewed, it was the busiest twenty-three-minute period. An officer testified that between 9:16 p.m. and 10:43 p.m., there were many calls exchanged between appellant and his girlfriend, which indicated to the officer that they were not together. Other Suspects. A police officer testified about eliminating at least five other potential suspects from consideration. About a week before the murder, the decedent had reported to a police officer that the first suspect had “jumped” the decedent and robbed him regarding a bet that the decedent had made on a basketball game. The first suspect drove a maroon Pontiac Bonneville. The first suspect was arrested on outstanding warrants and brought in for questioning. The officer eliminated the man as a suspect after reviewing his cell phone records, which indicated that the suspect was not in the area at the time of the murder. The second suspect was developed because the decedent’s wife heard someone outside of the door on the night of the shooting say their name was “Chad.” The decedent’s wife knew a Chad who made music with the decedent in the 1990′s. The wife was not aware of any beef or dispute going on between the man and the decedent. The man was included in a photo spread shown to the decedent’s son; the child did not identify anyone from the photo spread. The officer did not develop any connection between the man and a maroon vehicle. The third suspect was disclosed by the decedent’s wife because the man had sent hostile text messages to the decedent regarding an ongoing dispute between the two about a music record that the suspect had been recording for the decedent. The third suspect was not associated with a maroon car, and the decedent’s son knew the suspect and did not indicate that the man was the shooter. The fourth suspect was an associate of the third suspect. He fit the description of the shooter and drove a maroon Chrysler Sebring. He was included in a photo spread for the decedent’s son, but the child did not identify anyone from the photo spread. The fifth suspect was another associate of the third suspect. He drove by the scene when police officers were canvasing the area. An officer initially thought that the man looked like the artist’s sketch, but not as much when the officer got closer to him. The officer took some additional steps “to make sure he wasn’t connected to the victim,” but did not do much else other than that. 4. CSLI-Related Evidence Data Analysis. A Houston Police Department officer, Robert Brown, testified as the state’s expert regarding CSLI. His testimony consists of about 150 pages in the reporter’s record. He testified that from 8:17 p.m. to 8:49 p.m. on the night of the murder, appellant’s phone was in an area consistent with his parents’ house. From 8:52 p.m. to 9:32 p.m., appellant’s phone was hitting a cell phone tower and sector that was consistent with the murder scene. Then, at the same time the 911 call was placed regarding the murder, the CSLI showed movement of appellant’s phone away from the murder scene in an eastbound direction and then a rapid southbound movement towards appellant’s parents’ house. From 9:32 p.m. to 10:25 a.m. the next day, appellant’s phone stayed in an area consistent with his parents’ house. The court admitted five exhibits depicting maps showing the relevant locations, cell towers, and headings describing the locations of appellant’s phone at various times. Brown testified that appellant’s movement was consistent with someone fleeing the area. Corroboration of Identification. As mentioned above, a police officer testified as an expert about the corroboration of the child’s eyewitness identification. The officer testified that, in addition to the evidence referenced above, the CSLI was relevant corroborating evidence. This officer testified that Brown was an expert in cell phones, and it was “very” significant to the officer that he met with Brown and went over the CSLI. 5. Closing Arguments Appellant’s Closing. During appellant’s closing argument, counsel attempted to undermine the reliability of the child’s eyewitness identification and suggest that the police officers did not fully investigate other leads or collect potential evidence. Counsel addressed the CSLI as follows: And then the cell site records. You heard that today. They’re estimations. They’re assumptions. It’s a guess. This is not like GPS technology that says this is exactly where the person is or this is exactly where the person is. But you know what folks, it doesn’t matter because at the end of the day, there’s nothing in dispute. They’re all in that area. We’re not disputing that Gareic Hankston went to [the girlfriend's] apartment. After [her] apartment, went to her parents’ house. After her parents’ house, went home. The records show that. But they don’t show that he was standing in front of [the decedent's house] at 9:31 or 9:32. They just don’t show that. State’s Closing. The State argued that while the decedent was crawling on his hands and knees, grasping for his last breath and bleeding at 9:32 p.m., appellant was “fleeing the scene” because he was guilty of murder. The State recited the girlfriend’s testimony that she and appellant were together at her parents’ house and then left together about twenty minutes after the murder. The State argued, “That’s not true, because Gareic Hankston was nowhere to be found. At 9:32 p.m., he’s leaving that area and he doesn’t move again until 10:00 o’clock.” The State noted, “This is a circumstantial evidence case. It is not just about [the child].” The State encouraged the jury to “[t]ake the cell phone records back there and go through them” because the records undermined the girlfriend’s testimony that the couple was together the whole night. The State reiterated its theme from opening statements that the jurors would have to review the evidence and “put it altogether like a puzzle, piece by piece by piece by piece. The motive, the car, the description, the cell site, the call records, all of it, one stacked on top of the other and the other and the other. And then you get the identification.” 6. Deliberation and Verdict On the fourth and final day of deliberations, the jury indicated it was stuck at an 11–1 vote. After the court instructed the jury to keep deliberating, they reached a guilty verdict. C. Analysis The State used appellant’s CSLI for two primary purposes at trial: (1) to show appellant had the opportunity to commit the murder because he was in the area at the time of the murder; and (2) to show appellant’s flight from the scene, thus establishing a consciousness of guilt. In both regards, the State adduced evidence and made arguments for the jury to rely on the CSLI as evidence that corroborated the child’s eyewitness identification. Appellant’s counsel conceded during closing arguments that appellant was in the area of the murder at the relevant time because the girlfriend testified that they were at her parents’ house on the same street as the decedent’s house. Although there was evidence that appellant drove a vehicle very similar to one observed near the scene of the murder, the CSLI evidence was the strongest evidence showing appellant’s flight from the scene and, thus, his consciousness of guilt. See, e.g., Clay v. State, 240 S.W.3d 895, 905 n.11 (Tex. Crim. App. 2007) (“Evidence of flight evinces a consciousness of guilt.”). The CSLI evidence directly contradicted the girlfriend’s testimony that she and appellant left the area together twenty minutes after the murder, and the State pointed to the CSLI to contradict the girlfriend’s testimony. The CSLI evidence showing appellant’s flight from the scene was emphasized by the State throughout trial, was not cumulative, and was elicited from an expert. See Bagheri v. State, 119 S.W.3d 755, 763–64 (Tex. Crim. App. 2003) (erroneous admission of retrograde extrapolation testimony was harmful when evidence was emphasized throughout trial and was elicited from expert; although other evidence showed guilt, it was not overwhelming). As developed throughout the entire case, from voir dire to closing arguments, the parties disputed the reliability of the child’s identification. Although the State presented direct evidence of appellant’s guilt—the child’s identification—the State emphasized in opening and closing arguments that this case was a “circumstantial evidence” case. The State emphasized the importance of other corroborating evidence, including the CSLI specifically, and adduced substantial CLSI evidence. The case was not presented as one in which there was an “unshakeable” eyewitness. Cf. Clay, 240 S.W.3d at 905 (erroneous admission of hearsay was harmless under constitutional standard when the complainant and other eyewitnesses identified the defendant as a perpetrator of aggravated robbery and he fled from police). Although a rational jury might have convicted appellant based on the child’s identification, appellant’s presence near the scene, his motive, and his call log activity, we cannot conclude from this record that the admission of the CSLI evidence—showing his flight from the scene, contradicting his girlfriend’s testimony, and bolstering the child’s identification—had no effect or only a slight effect on the jury’s verdict. After considering the entire record, we have grave doubt that the result of the trial was free from the substantial effect of the error. See Barshaw, 342 S.W.3d at 94. The matter is so evenly balanced regarding the effect of the error that we are in virtual equipoise as to the harmlessness of the error. See id. Under these circumstances, appellant’s conviction must be reversed. See id. IV. Conclusion Appellant’s sole issue is sustained. and the case is remanded for a new trial. The trial court’s judgment is reversed, /s/ Ken Wise Justice Panel consists of Chief Justice Christopher and Justices Wise and Poissant. Publish —Tex. R. App. P.47.2(b)
