joined by Justice O’NEILL, dissenting.
The Court concludes that Cox’s testimony amounts to “no more than a mere scintilla” of evidence on causation. 159 S.W.3d at 911. To the contrary, Cox testified that the Passat experienced a “catastrophic failure of the wheel bearing assembly” while it was traveling in the eastbound lane of U.S. Highway 83, before the Passat entered the median. He both tested and rejected Volkswagen’s alternative theory— that damage to the wheel bearing assembly occurred after the Passat’s collision with the Mustang. Reasonable jurors could have accepted Volkswagen’s theory and rejected Cox’s (as they did in the first trial), or accepted Cox’s and rejected Volkswagen’s (as they did here), but unlike the jury, this Court lacks constitutional *914authority to weigh conflicting evidence. Accordingly, I respectfully dissent from the Court’s rendition of judgment for Volkswagen.
I agree with the Court, however, that the trial court erred in admitting the testimony of the unidentified witness, whose deliberate steps to ensure his anonymity and calm demeanor throughout the interview negate the trial court’s conclusion that this hearsay was admissible under the excited utterance exception. Furthermore, that error probably caused the rendition of an improper judgment. I would accordingly reverse the court of appeals’ judgment and remand the case to the trial court for a new trial.
I
Cox’s Testimony on Causation1
Causation was a hotly contested issue at trial. The Ramirez plaintiffs argued that a defective left-rear wheel bearing assembly caused Sperling to lose control of her Pas-sat, which crossed the median and collided with Guerra’s Mustang. In contrast, Volkswagen contended that driver error, specifically Sperling’s overreaction to the contact between her Passat and the Cáma-ro, triggered a chain of events that caused Sperling to lose control and cross the median. To that end, Volkswagen , asserted' that the left-rear wheel separation was not the cause of the accident, but the result of the impact at the end of the accident sequence. Each party offered expert witness testimony to support their respective causation theories.
Cox testified that defects in the Passat’s left-rear wheel bearing assembly caused an abrupt, catastrophic bearing failure that occurred from the inside out when the car was traveling in the eastbound lane of U.S. Highway 83. He asserted that cracks in the bearing assembly produced metal fragments that eventually became lodged within the assembly, causing it to jam and disintegrate abruptly. He contended that as the bearing assembly came apart, pieces of it were ejected with force sufficient to knock off the thrust washer and adjusting nut that held the wheel assembly intact. Without the adjusting nut in place, the hub and - rotor moved outward until the rotor contacted the brake pads and induced a retarding force, which led to erratic vehicle behavior and caused Sperling’s vehicle to enter the median.
Cox specifically testified regarding three vehicle defects. First, based on microscopic tearing at the base of the threads in the adjustment nut, which holds the bearing assembly intact, he contended that the nut was looser than it should have been. Second, based on false brinell marks on the rollers within the bearing assembly, he concluded that the bearing assembly was damaged in transit because the car was cinched down too tightly. *915Third, based on the presence of cracking and metal fragments in the bearing assembly, Cox concluded that the metal in the assembly became embrittled by temperature inconsistencies in the manufacturing process. He testified that the embrit-tled material structure led to cracks, which created steel bearing fragments that were found embedded in the roller cage of the Passat, trapped between the inner and outer races and embedded in the surface of the rollers.
The Court concludes that “Cox’s testimony that the results of this defect would be consistent with erratic vehicle behavior causing the driver to make a corrective steer and veer into the median is an unsupported conclusion.” 159 S.W.3d at 912. I submit, however, that the jury was entitled to make the reasonable inference that the wheel bearing assembly’s disintegration (which all agree would cause erratic vehicle behavior) was the catalyst for the accident. Cox testified:
Q: And in that failure that occurs back here 500 feet or so, that’s when this explosion takes place when the roller fragment ... comes shooting out and does what it has to do, correct?
A: I can’t tell you the number of feet. I haven’t been involved in the accident reconstruction. But I can tell you, you do have a catastrophic failure of the bearing which does, at that point, take the nut off which holds the wheel intact and which then causes the hub to start moving outward. And as the hub and rotor move outward, the rotor contacts the brake pads, which would then induce a retarding force because of contact between the rotor and the brake pads, which would certainly be consistent with why there is an unusual behavior of the vehicle at that point in time. Now, where it occurs in the highway, I can’t tell you. That’s Mr. Walker’s analysis. But we do know that those facts of the failed fastener assembly and the friction between the rotor and the brake pads would certainly lead to an erratic vehicle behavior which would then cause the driver to make a corrective steer.
The Court states that Cox cited “no testimony, tests, skid marks, or other physical evidence to support []his opinion.” 159 S.W.3d at 912. In fact, as the Court itself notes, Cox testified on direct examination that the presence of grass mixed with the grease in the hub could be explained only if the bearing failure occurred before the vehicle went into the median. Id. at 912. Volkswagen itself elicited similar testimony:
Q: I don’t know if you were present in the courtroom when I was talking with Mr. Walker, but we were looking at this diagram that Mr. Walker brought in where we see the beginning of the marks here that the police had indicated. We talked about the car leaving the eastbound lanes and going in the median, and I had a dialogue with Mr. Walker about the fact that the failure had to occur back 528 feet or 564 feet from the point at which the Passat enters the median. Were you present for that discussion?
A: Yes, I was.
[[Image here]]
Q: Okay. In other words, back here in your opinion and his opinion, there was this sudden catastrophic failure of the wheel bearing assembly, correct?
A: Sure. It has to happen before the vehicle goes into the median because we know from the grass in the hub that’s your only opportunity to get the grass there.
Furthermore, Cox testified regarding additional physical evidence of a bearing failure that occurred from the inside out prior *916to impact (contrary to Volkswagen’s assertions):
Q: Now, go ahead and tell me about the time elements with the different parts now that we know what those parts are, if you would.
A: Okay. We know from the physical evidence that there are small fragments in the bearing that have been run over by the rollers. Those fragments were being generated by the bearing before it faded....
We have in this instance literally friction welded a piece of debris onto rollers and pieces of the inner race. So before everything comes apart, we have a bearing that’s starting to disintegrate. It’s malfunctioning, which also is consistent with the grease becoming black, because grease normally does not discolor in such a short time period....
But yet the grease inside the cavity of the hub was coal black, the bearings, and contained lots of these little fragments.
That was part of what led up to the failure. But we also had some damage in the bearings that preexisted much of the operation life. What happened—
[[Image here]]
Q: What made those [brake pad] markings?
A: There’s only one thing that could make those marks, and that’s from this rotor, the edge of this rotor. The brake is clamped. And, again, it’s undisputed that the parking brake was activated when the Volkswagen crashed frontally into the Mustang. So the brakes were activated at that time, and this shows us the position of this rotor at that point in time.
[[Image here]]
Q: So this is the brake pad only hitting the edge of the rotor because the rotor is off center?
A: Right, when the brake is activated. And there’s only one point in time when the brake is activated, and that’s at the point of the frontal collision. So that tells us definitively that we have a broken rotor at the time of impact.
Q: Does it tell you whether or not the failure occurred before or after the time of impact?
A: Yes.
Q: Conclusively?
A: Yes.
Q: And what does it tell you?
A: It tells you that at the time that the two vehicles collided, the wheel was already broken away.
[[Image here]]
Q: Is this a failure inside-out or outside forces in?
A: This is an internal failure where the bearing itself is starting to digest itself. Q: Is the wheel moving when this is happening?
A: Sure. Because if the wheel is not moving, the rollers aren’t going around the race, and so the rollers can’t roll over anything. So this has to be a rotational event.
Q: Does it happen before or after the vehicles collide together?
A: It has to occur before.
While Cox’s causation testimony is neither ironclad nor exhaustive, it is surely some evidence that the Passat’s bearing failure occurred in the eastbound lane, contributed to Sperling’s loss of control, and ultimately caused a catastrophic accident in the westbound lane.2 Whether the *917wheel separated completely in the eastbound or westbound lanes is important but not dispositive; indeed, the jury could have accepted Volkswagen’s theory that a total separation occurred toward the end of the accident sequence, yet also accepted Cox’s testimony that the bearing failure began a-process in which the hub began moving outward and eventually resulted in a total wheel separation in the median (which explains grass in the wheel hub), immediately before, the collision in. the westbound lane. We are supposed to indulge inferences in favor of the verdict, not against it.
Coastal Transport Co. v. Crown Central Petroleum, Corp., 136 S.W.3d 227, 231 (Tex.2004), does not support the result the Court reaches today. In that case, we concluded that an expert’s testimony on gross negligence was conclusory on its face. Id. at 233. The sum total of testimony on gross negligence in Coastal was as follows:
Q: When viewed objectively from Coastal’s point of view at the time of the September ’93 incident, in your opinion, did Coastal’s failure to stop using probes that could have [sensor failure] problems, did that involve a high degree of risk, considering the probability and magnitude of the potential harm to others?
A: Yes, it did, very high.
Q: In your opinion, did Coastal have an actual subjective awareness of the risk involved in failing to stop using probes that can have [sensor failure] problems?
A: Yes, again and again.
Q: And in your opinion, did Coastal nevertheless proceed with conscious indifference to the rights, safety, or welfare of others?
A: That’s the only conclusion I can draw.
Id. at 231. By equating Cox’s testimony here with the paltry testimony at issue in Coastal, the Court sets a dangerous precedent that threatens to fundamentally alter the nature of no-evidence review.3
Rather than indulging every reasonable inference in favor of the jury’s finding, the Court adopts a contrary approach, tipping the scale in the opposite direction to dismiss as “conclusory” expert testimony that supports the verdict. This Court is constitutionally bound to conduct only a legal— not factual — sufficiency review.4 See Tex. *918Const. art. V, § 6; Tex. Gov’t Code § 22.225(a); Herbert v. Herbert, 754 S.W.2d 141, 143 (Tex.1988); Hall v. Villarreal Dev. Corp., 522 S.W.2d 195 (Tex.1975). While a jury or court of appeals may find Cox’s testimony factually insufficient on causation, it is at least some evidence. I therefore respectfully dissent from the Court’s rendition of judgment for Volkswagen.
II
Unidentified Witness’s Testimony
The trial court admitted into evidence an unidentified eyewitness’s assertions that the Passat had a blown-out tire before it crossed the median. I agree with section IIB of the Court’s opinion, which concludes that the witness’s statements were not an excited utterance and thus, were inadmissible hearsay. The videotaped testimony reveals that the witness’s first reaction was one of self-interest; he agreed to the interview only if his identity was concealed. He was composed and reflective when giving his statement. I therefore agree that the statement does not qualify as an excited utterance.
I write separately to address whether the trial court’s erroneous admission of the witness’s testimony amounted to harmful error requiring reversal.
A
Standard of Review
Whether to include or exclude evidence is a matter committed to the trial court’s sound discretion. Interstate Northborough P’ship v. State, 66 S.W.3d 213, 220 (Tex.2001); Tex. Dep’t of Transp. v. Able, 35 S.W.3d 608, 617 (Tex.2000). “Erroneous admission of evidence requires reversal only if the error probably (though not necessarily) resulted in [the rendition of] an improper judgment.” Nissan Motor Co. v. Armstrong, 145 S.W.3d 131, 144 (Tex.2004); see also Tex.R.App. P. 61.1(a); Interstate, 66 S.W.3d at 220. To make this determination, we review the entire record. Nissan, 145 S.W.3d at 144; Able, 35 S.W.3d at 617; City of Brownsville v. Alvarado, 897 S.W.2d 750, 753-54 (Tex.1995). We will not reverse a judgment for erroneous rulings if, as the plaintiffs argue here, the evidence in question is cumulative. Interstate, 66 S.W.3d at 220; Able, 35 S.W.3d at 617-18.
While it is clear that a judgment should not be reversed for erroneous admission of evidence that is merely cumulative, beyond that, determining whether the erroneous admission was harmful “is more a matter of judgment than precise measurement.” Nissan, 145 S.W.3d at 144; see also Interstate, 66 S.W.3d at 220; Able, 35 S.W.3d at 617. To make this judgment, we have looked to counsel’s efforts to emphasize the evidence in question. See Nissan, 145 S.W.3d at 144; Spohn Hosp. v. Mayer, 104 S.W.Sd 878, 883-84 (Tex.2003); Alvarado v. Farah Mfg. Co., Inc., 830 S.W.2d 911, 917 (Tex.1992). In addition, we have examined whether there was contrary evidence that the improperly admitted evidence was calculated to overcome. See Nissan, 145 S.W.3d at 144; Spohn Hosp., 104 S.W.3d at 884.
B
Harmful Error Analysis
As noted above, the parties vigorously disputed causation in this case. Whereas the plaintiffs contended that a defective *919bearing assembly led to a catastrophic bearing failure that caused the accident, Volkswagen argued that driver error caused the accident and that the bearing failure was the result, not the cause, of the accident. The main dispute at trial centered around whether the bearing failure occurred in the eastbound lanes, thereby causing the Passat to enter the median and collide with the Mustang, or in the westbound lanes, as a result of the impact at the end of the accident sequence.
The trial court admitted into evidence a local news crew’s videotaped interview of a witness at the accident scene. In the interview, the witness stated: “The station wagon was headed from here toward over there and apparently had a blown-out tire.... The tire blew up and it crossed the median and it hit the car that was headed from over there toward over here.” The witness refused to identify himself and was never located or deposed. The court of appeals held that, if the trial court committed error in admitting the witness’s testimony, the error was harmless because the appellant did not show that the error probably resulted in an improper judgment. 79 S.W.3d at 122. I disagree.
The unidentified witness’s testimony was not merely cumulative of other evidence in this case. The witness’s statements constituted the only disinterested, eyewitness account that the plaintiffs offered to support their causation theory — that the bearing failure occurred at the beginning of the accident sequence, thereby leading Sper-ling to lose control of the Passat, which crossed the median and collided with the Mustang. Although Cox also provided some evidence of causation, his testimony was derived not from direct observation of the accident, but from post-accident analysis and testing. Moreover, unlike the unidentified witness, Cox conducted his research and formed his opinion solely for the purpose of litigating this case. See E.I. du Pont de Nemours & Co. v. Robinson, 923 S.W.2d 549, 559 (Tex.1995) (“[0]pinions formed solely for the purpose of testifying are more likely to be biased toward a particular result.”); Nissan, 145 S.W.3d at 144 (noting that “we have consistently required competent expert testimony and objective proof’ in product defect cases). Finally, whereas Volkswagen was able to challenge the credibility of Cox’s testimony through cross-examination, it did not have the same opportunity with the unidentified witness. See Davidson v. Great Nat’l Life Ins. Co., 737 S.W.2d 312, 314 (Tex.1987) (“Cross-examination is a safeguard essential to a fair trial and a cornerstone in the quest for truth.”); State v. Hilton, 412 S.W.2d 41, 42-43 (Tex.1967) (holding that error in denying cross-examination of witness was harmful because witness was “the real supporting witness”).
At trial, Plaintiffs’ counsel repeatedly emphasized the importance of this evidence. Counsel presented the evidence in three forms: (i) the videotaped interview of the witness at the scene of the accident; (ii) a transcript of the interview read to the jury; and (iii) live testimony of the reporter who interviewed the unidentified eyewitness. In arguing for the admission of the evidence, counsel asserted that the videotape containing the interview was “a very important piece of evidence for the plaintiffs” and that they were “only going to show the video clips of the evidence that [they] needled] to prove their case.” (Emphasis added.) They referred to the evidence in their opening statement, and it was the first evidence presented at trial. Most notably, plaintiffs’ counsel made the following statements during closing argument:
We heard from an eyewitness, an eyewitness who gave some testimony, testi*920mony that hasn’t been controverted, hasn’t been disputed in this case, hasn’t even been addressed. It’s just kind of been ignored by Volkswagen. And what did that man say? That man told us very clearly, very simply and very honestly ... that what he saw from his position looked like a blowout to him; that a tire blew out, a car went out of control, it skidded across a median, it killed two girls and permanently damaged another one. He has no ax to grind. He has no beef with anybody. He’s not being paid to say any of this. He’s just an honest man giving an opinion, an opinion that was captured on film and was given to you just minutes, minutes after he experienced it.
It wasn’t like he went back out to the scene to measure. It wasn’t like he looked at a bunch of photographs. It wasn’t like he ran hundreds of tests and lots of exhibits. He told you that just minutes after he witnessed what he thought was a blowout.
What does that mean to us? It means something very, very, very clearly. It means that something obviously went wrong with this tire. This man didn’t get to see an accident that took place in two seconds stopped frame by frame. What he saw was something come loose, and that’s exactly what Mr. Walker told you.
Additionally, the unidentified witness’s testimony was very likely calculated to overcome contradictory evidence. To support its causation theory and discount the opposing theory, Volkswagen, like the plaintiffs, offered extensive expert witness testimony through an accident reconstruc-tionist and metallurgist. It also presented an eyewitness who testified that the car appeared to be bouncing as it crossed the median and that its back end lifted off of the ground upon impact with the Mustang. Finally, the sergeant who investigated the accident testified that the inspection indicated, albeit not definitively, that the Pas-sat left four tire marks in the median and westbound lanes.
In sum, the evidence in question was not merely cumulative, it was heavily emphasized by the Ramirez’s counsel, and it was very likely calculated to overcome contradictory evidence at trial. Considering these factors, I would hold that its erroneous admission was harmful and therefore necessitates reversal.
Ill
Conclusion
Cox’s testimony is legally sufficient to support the jury’s causation finding. I would not affirm the court of appeals’ judgment on this ground, however, because the erroneous admission of the unidentified witness’s statements probably resulted in the rendition of an improper judgment. Accordingly, I would reverse the court of appeals’ judgment and remand the case to the trial court for a new trial. See Tex.R.App. P. 60.2(d).
. Texas Rule of Appellate Procedure 53.2(f) states: ‘'The petition must state concisely all issues or points presented for review. The statement of an issue or point will be treated as covering every subsidiary question that is fairly included.” See also Tex.R.App. P. 55.2 (brief on merits must be confined to issues stated in petition for review). Volkswagen, in petitioning this court for review, did not complain that Cox’s testimony on causation was legally insufficient. The issue presented in Volkswagen’s petition for review was not briefed and referred to Walker, not Cox: “Was the unsupported testimony of the plaintiffs’ accident reconstruction expert. [Walker] legally sufficient evidence of causation?” (Emphasis added.) Volkswagen first challenged Cox on causation in its reply brief on the merits and then devoted only two sentences suggesting, inaccurately, that Cox did not testify about when the bearing failed or whether it could have contributed to the accident. Given this procedural background, the Court has essentially concluded that Rules 53.2(f) and 55.2, although stated as mandates, are merely aspirational.
. The jury answered "yes” to this question: "Was there a defect in the Volkswagen Passat stationwagon at the time it left the possession of Volkswagen of America, Inc., that was a *917producing cause of the occurrence in question?”
.Justice Hecht’s colorful concurrence misses the point. Volkswagen has not challenged in this Court the methodology employed by Cox or the extent to which his conclusions are supported by testing, which form the bulk of Justice Hecht’s concurrence. Cox does not say "the moon is made of green cheese,” or "the Earth is the center of the solar system,” but rather that objective evidence supports his conclusion that the bearing failure occurred before the Passat left the eastbound lane and caused the accident. See Maritime Overseas Corp. v. Ellis, 971 S.W.2d 402, 420 (Tex.1998) (Hecht, J., concurring). A court may decide as a matter of law that the former examples are "no evidence,” but when more than a scintilla of objective evidence supports an expert’s conclusions in a technical area in which judges have no particular expertise, and when that expert’s methodology is not challenged on appeal, the question becomes one of factual, and not legal, sufficiency.
.Whether the evidence is sufficient, as a factual matter, to support the jury's finding is beyond this Court’s jurisdiction. This is not to deny the trial court’s obligation, as gatekeeper, to assess reliability during the Daubert-Robinson hearing, see Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), and E.I. du Pont de Nemours & Co. v. Robinson, 923 S.W.2d 549 (Tex.1995), nor the appellate court’s obligation to evaluate the trial court’s determination when a party challenges reliability on appeal. But here, as the Court notes, "Volkswagen does not object to the *918reliability of Cox’s testimony.” 159 S.W.3d at 912. Our limited task, then, is to determine whether the evidence is legally sufficient.