OPINION
CHARLES KREGER, Justice.A jury found beyond a reasonable doubt that Lester Winkle suffers from a behavioral abnormality that predisposes him to engage in a predatory act of sexual violence. The trial court committed Winkle to outpatient treatment as a sexually violent predator. See Tex. Health & Safety Code Ann. § 841.081 (West 2010). On appeal, we hold that the trial court erred in excluding the testimony of Winkle’s sole expert witness. Accordingly, we reverse the order of commitment and remand the case to the trial court for a new trial.
Winkle contends in issue two that the trial court erred by excluding his sole expert witness from testifying at trial on the basis of his supposed disbelief in the existence of “behavioral abnormality.” “A trial court abuses its discretion in excluding expert testimony if the testimony is relevant to the issues in the case and is based on a reliable foundation.” State v. Cent. Expressway Sign Assocs., 302 S.W.3d 866, 870 (Tex.2009). In this case, the State filed a pre-trial motion to exclude the testimony of forensic psychiatrist John Tennison pursuant to Rules 402 and 702 of the Texas Rules of Evidence, which the trial court granted after an oral hearing. See Tex.R. Evid. 402 (“All relevant evidence is admissible, except as otherwise provided by Constitution, by statute, by these rules, or by other rules prescribed pursuant to statutory authority. Evidence which is not relevant is inadmissible.”); see also Tex R. Evid. 702 (“If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education may testify thereto in the form of an opinion or otherwise.”).
The State did not challenge Tennison’s qualifications and limited its challenge to the relevance and reliability of his testimony. In the motion, the State alleged that “[bjecause Dr. Tennison does not believe that ‘behavioral abnormalities’ exist at all, his opinion is not relevant in this case and should, therefore, be excluded.” Citing deposition excerpts from two other SVP commitment cases,1 the State argued that “[a]n expert in this case should look at the particular facts of Defendant’s history to educate the jury regarding whether Defendant has a behavioral abnormality.” The State claimed that
Dr. Tennison does not offer such testimony. Rather, his evaluation ends at his interpretation of the statute. Therefore, Dr. Tennison’s opinion is rendered irrelevant when he refuses to accept as *244valid the statutory starting block of the definition or concept of a ‘behavioral abnormality,’ as it is defined by our Texas Legislature — he will never have the opinion that any sex offender has a behavioral abnormality.
Tennison’s pre-trial hearing testimony demonstrates the inaccuracy of the factual assertion presented by the State in its motion. Tennison did look to the particular facts of Winkle’s history in forming his professional opinion. For example, Tenni-son explained:
[TENNISON]: Well, I look at the record, Your Honor. And, obviously, by the time I receive the file, there’s generally several hundred pages at the very least. And I look at evidence that someone has had serious difficulty controlling their behavior. That certainly has to be present. But I apply the five-prong test that Dr. Rogers elaborated on, which I referred to, which is, first of all, to see if they have serious difficulty controlling their behavior.
THE COURT: Wait a minute. I’m going to write this down.
[TENNISON]: Okay. Actually, Your Honor, I prepared a packet of information for you as promised, a reading list, as well as that article from Dr. Richard Rogers, and I can give you a copy right now, if you would like.
THE COURT: No.
[TENNISON]: Okay.
THE COURT: Five-prong test. One is to see if they have serious difficulty controlling their behavior.
[TENNISON]: Okay. Actually, I was stating the steps out of order. But, yes, I first want to see if there’s evidence of serious difficulty controlling behavior as related to some sort of sexual offense.
THE COURT: So, like, if they have two, three, four rape convictions, that would be some evidence of that, or no, not necessarily?
[TENNISON]: Well, the — not necessarily.
THE COURT: Okay. That’s fine.
[TENNISON]: Let me — let me clarify that. The mere repetition of the behavior is not enough to conclude that someone has volitional incapacity. For example, I get out of bed every morning, which is a repetition of behavior, but that doesn’t mean I have volitional capacity just because I repeat that behavior.
THE COURT: If you get out of— never mind. You’re getting me off task here.
[TENNISON]: I’m sorry.
THE COURT: So go back. What’s the next of the five-prong [test] you look for?
[TENNISON]: Right. So first I would want to see that there’s evidence of having serious difficulty controlling behavior as pertaining to the sexually coercive behaviors. There might be some ambiguity regarding what kind of sexual behaviors, but certainly any sexual behavior that could be considered coercive or threatening to another person could be an example.
I would then want -to see if this serious difficulty controlling behavior was the result of some sort of underlined congenital or acquired condition; as contrast to say if someone just, you know, drank a lot of alcohol or snorted a lot of cocaine and they were behaving sexually inappropriate, that would be a temporary metabolic disturbance and not an example of what physicians would consider an acquired or a congenital condition. So you want to make that distinction as well.
If there is this serious difficulty controlling behavior and if that behavior is *245resulting from a congenital or abnormal condition, you then also want to ask the question, “Is this serious difficulty controlling behavior likely to continue?” The law requires that it be likely. And for me even if-obviously, I’m not the trier of fact in these situations, but I have to define what the word “likely” means to me.
So, based on the caselaw and based on what other experts have done, including Dr. Gaines, for example, I use the term “more likely than not.” If I think someone is more likely than not to continue to have serious difficulty controlling behavior as a result of a congenital or acquired condition, then those are the fundamental requirements that would need to be present.
As far as volitional incapacity, I could elaborate my definition of that.
THE COURT: On your five-prong test here, you told me one, see if they have serious difficulty controlling behavior regarding sex; and then if that’s from an acquired condition; and number three was whether it’s likely to continue. What’s number four?
[TENNISON]: Okay. I actually collapsed the five prongs into, I think, just three prongs—
THE COURT: Those three I just said?
[TENNISON]: — in my description. Yes. What I just said — I mean, if you review what I just said, I don’t think I actually enumerated it out to be five. But it is — it is representative of the five-prong or five-step test that Dr. Richard Rogers outlined in his 2005 article in the Journal of Psychiatry and the Law.
THE COURT: Okay. So you look at the records, you see if that happened. Then what do you do?
[TENNISON]: Well, of course, I want to evaluate the person. If it’s apparent that someone has a history of this propensity from the record alone, then the evaluation of the person might not be as pertinent. But it’s always routine to evaluate the person because certainly new information can be had from that. So one could potentially have a direct observation of someone’s comments or behaviors, which was not apparent from the record, which would conclude — allow one to conclude that one had volitional incapacity as caused by a congenital or acquired condition.
THE COURT: So you evaluate the person. Then what do you do?
[TENNISON]: Right. Well, I mean, there are — there are lots of things. If I evaluate the person and I look at the record, I then try to formulate an initial opinion based on what I — the information I have at that time. However, sometimes new information comes into play. You know, what I do depends on the individual case.
In the case of Mr. Hinkle, his case was vastly complex as compared to Mr. White and Mr. Winkle, whose cases were much simpler. So there’s a lot of collateral information. Because of the lack of clarity and the accusations of Mr. Hinkle being a liar that I had to look into, none of that was necessary for Mr. White or Mr. Winkle. They did not deny what they had done. They did not — you know, there was no lack of clarity. The only lack of clarity was the degree to which volitional incapacity was likely in the case of Mr. White. And, based on his deposition, it appears to be much more likely than was apparent at the time of the evaluation.
THE COURT: Okay. [Counsel for Winkle], you can continue. I’m sorry. Is that all you do when you — that’s your evaluation process right there?
*246[TENNISON]: You asked me to summarize, Your Honor. I can talk all day in detail if-I mean, I didn’t — if you want me — I can elaborate on how I define the term “volitional incapacity.”
THE COURT: No. I want to know what you do when you get the case.
[TENNISON]: Well, I think I—
THE COURT: How you determine whether or not somebody has a behavioral abnormality.
[TENNISON]: Part of what I do is the thoughts I have. The thoughts I have and the way I define things is inclusive of what I do. So, if you’re asking what I do, I mean, behaviorally I’m having — thinking, I’m analyzing this, I’m thinking about the data and saying do they have it or do they not. But what I do is my thought process. That’s what I get paid for.
And to answer the question of what I do requires me to elaborate my thought process, which requires that I tell you the definition that I’m using for the terms if you want that degree of elaboration in my answer.
Tennison also explained how he applied his methodology in Winkle’s case, as follows:
[TENNISON:] Mr. Winkle gave no evidence in his — during my examination of him or from his record of not having the volitional capacity, unlike Mr. White who appears to lack that — likely lack that capacity in many instances. Mr. Winkle describes what he did as having been voluntary, although he did describe it as being under the influence of alcohol. It’s possible, if that is true, that he may not remember his state of volitional capacity. But certainly there’s no evidence of any sort of acquired or congenital condition even if there was a temporary intoxicational condition that might have affected his volitional capacity at the time.
[WINKLE’S COUNSEL:] Now, you are aware that Mr. Winkle, he has, you know, three sexual assault cases against — against elderly women?
[TENNISON:] Yes.
[WINKLE’S COUNSEL:] You are aware of that?
[TENNISON:] Yes, I am!
[WINKLE’S COUNSEL:] And what fact [sic] does that have on your diagnosis?
[TENNISON:] Well, the fact that a sexual assault case is certainly a red flag. I mean, you want to look into those assault cases and the details surrounding them. But the mere reputation — repetition of a sexual assault is not in and of itself proof of volitional incapacity. So one needs to look at the state of mind leading up to and, if possible, during those events.
[WINKLE’S COUNSEL:] Okay. Now, you read the file in the Winkle case?
[TENNISON:] Yes.
[WINKLE’S COUNSEL:] And Mr. Winkle has not taken SOTP?
[TENNISON:] That is correct.
[WINKLE’S COUNSEL:] And what effect does that have on your diagnosis of Mr. Winkle?
[TENNISON:] Well, the — that is a consideration that I have made. The lack of SOTP as the data has shown over and over again, lack of sex offender treatment or lack of motivation for it is not a risk factor for repeated sexual recidivism. It’s been studied over and over and I can cite the specific studies.
But my opinion is that whether he had received it or not, it would not change my opinion about his propensity for repeated acts of sexual violence, whether *247they involve volitional capacity or not. However, I should say that the MnSOST does reduce its risk score as a result of such treatment. But based on the data of Karl Hanson, who is one of the creators of the Static-99, it doesn’t appear that such treatment is actually associated with reduced recidivism and, therefore, it’s probably not appropriate to have an actu — the evidence shows it’s not appropriate to have an actuarial instrument.
[WINKLE’S COUNSEL:] Okay. Now, Mr. Winkle, he was intoxicated during his offenses; correct?
[TENNISON:] So he says.
[WINKLE’S COUNSEL:] Okay. And what effect does that have on your determination if Mr. Winkle has a behavioral abnormality?
[TENNISON:] Well, one can imagine a lack of volitional capacity that can be associated with a metabolic disturbance, such as alcohol intoxication. Certainly no physician would regard that as a congenital or acquired condition.
However, it’s possible that, at the very least, his judgment was poor. I think we can conclude that conclusively. Whether he had an intact sense of self of — whether he had an intact sense of planning and doing what he wanted to do as a result of this choice in contrast to not being able to help himself or acting reflexively in the kind of the way Mr. White did when he fell off his bicycle, it’s not clear. But, obviously, these sorts of evaluations occur when someone is in a non-intoxicated state. You’re not looking at a temporary mental state when they’re in a state of intoxication, but rather their normal baseline state that can be considered acquired or congenital.
So the fact that he might or might not have been inebriated when he committed these sex crimes is not really relevant to the here now evaluation, but it certainly could be relevant in terms of a risk factor for him. It probably is a risk factor, indeed, that if he were to start drinking again, he might very well be more likely to commit crimes, perhaps sex crimes, but that would not be an example of behavioral abnormality. That would be someone behaving inappropriately as a result of a metabolic disturbance, not unlike delirium in which people were out of their minds behaving in ways that are entirely inconsistent with their physical condition.
[WINKLE’S COUNSEL:] Okay. Now, what effect does it have on your evaluation that Mr. Winkle has taken, you know, numerous classes, programs in TDC in order to help himself when he gets out of prison? How does that affect your evaluation?
[TENNISON:] Well, I think that shows he has planning for the future and tends to be one of his strengths. It also is consistent with the fact that he has a non-psychopathic PCL-R score. You know, that’s a — I think is — even though there was some point differences between Dr. Thorne’s and my PCL-R score. But I think we can probably both agree that Mr. Winkle has a capacity to plan for the future, which would not— which otherwise would have elevated his PCL-R score had he not.
The State also argued that Ten-nisoris opinion is not reliable. The nonexclusive considerations for determining the reliability of expert testimony: (1) the extent to which the theory has been or can be tested; (2) the extent to which the technique relies upon the subjective interpretation of the expert; (3) whether the theory has been subjected to peer review or publication; (4) the technique’s potential rate of error; (5) whether the underly*248ing theory or technique has been generally accepted as valid by the relevant scientific community; and (6) the non-judicial uses which have been made of the theory or technique. E.I. du Pont de Nemours & Co. v. Robinson, 928 S.W.2d 549, 557 (Tex.1995). An expert’s opinion may be unreliable if the tests conducted by the expert do not support the testifying expert’s opinion, or if the expert fails to explain how the facts support his conclusion. Whirlpool Corp. v. Camacho, 298 S.W.3d 631, 637-38 (Tex.2009). In the motion to exclude Ten-nison as an expert witness, the State argued that Tennison “uses flawed reasoning because his opinion is based on the incorrect assertion that there is no such thing as a behavioral abnormality. Thus, Dr. Tennison brings only his credentials, flawed reasoning, and subjective opinion to this case.” The record of the pre-trial hearing refutes the State’s basis for concluding that Tennison’s opinion is not reliable due to his supposed personal disbelief in the existence of “behavioral abnormality.”
During the hearing Winkle’s counsel asked Tennison for a citation to professional literature “concerning whether or not one can measure if a person is having serious difficulty controlling their behavior.” Tennison replied, as follows:
Yes. The literature that I would like to cite in response to your question is a 2005 book by John Q. LaFond, who is an expert on mental health law. In this book entitled Preventing Sexual Violence, from the American Psychological Association Press, Mr. LaFond has a chapter entitled Sexually Violent Predator Laws. On page 140 in that chapter, he — there is a section entitled Impossible to Determine Volitional Impairment.
It says, quote, in Kansas v. Crane, comma, the Supreme Court required the government to prove in SVP trial that the defendant’s mental condition significantly impaired the ability to control his sexual behavior. Period. Yet there is virtually unanimous consensus that mental health professionals cannot determine when an individual has significant difficulty controlling his behavior. Period. In 1982 slash or dash 1983, comma, both the American Bar Association and the American Psychiatric Association recommended eliminating the volitional prong of American Law Institute insanity defense because experts simply could not make this determination. Period.
The trial court asked, “Is that what you believe, too, Dr. Tennison?” Tennison explained, as follows:
[TENNISON]: Scientifically it cannot be determined, inferentially it can. And I use inference; therefore, I can determine that someone has a behavioral abnormality, and I have.
THE COURT: Okay. But so that whole thing you just read you don’t believe?
[TENNISON]: No, that’s not what I’m saying, Your Honor. What I just read refers to what can be determined scientifically.
THE COURT: Okay. So you don’t think it can happen scientifically, but it can be done other ways?
[TENNISON]: Yes. Absolutely.
If an expert’s opinion does not fit the facts of the case, significant analytical gaps in an expert’s opinion may undermine its reliability. See TXI Transp. Co. v. Hughes, 306 S.W.3d 230, 235 (Tex.2010). On appeal, the State argues the manner in which Tennison applies his methodology is suspect because it is “colored by and driven by his political leanings.” As the testimony described above shows, however, Tennison’s opinion is based upon his methodology, which he explained and applied to *249the particular facts of the case. The State argues that the trial court could exercise discretion to exclude the evidence because Tennison “dismissed any negative information about Winkle.” But Tennison did consider the commission of repeated sexual assaults against elderly nursing home patients to be a “red flag,” and stated that one should consider “the state of mind leading up to and, if possible, during those events” in making a behavioral abnormality determination. From his interview with Winkle, Tennison concluded that the sexual assaults were voluntary conduct committed in a temporary state of intoxication. The State’s psychiatry expert, Michael Ar-ambula, acknowledged that the diagnostic manual refers to intoxication as mitigating the diagnosis of paraphilia, but he doubted that Winkle experienced that level of intoxication because he could recall significant parts of one of the sexual assaults. This is not a case in which the proffered expert ignored the facts of the case in forming his opinion.
The State also argues that Tennison would have confused the jury because Ten-nison both acknowledged that the law is valid and discussed various authors who have called such laws a misuse of psychiatry. Tennison cited to a 1999 Task Force Report of the American Psychiatric Association to explain his earlier statement that the American Psychiatric Association recognized behavioral abnormality statutes, “but not as having a scientific basis for existence.” He was explaining the distinction between the rule of law, which is honored by the association, and the association’s disapproval of the use of what appears to be clinical language that gives “the impression of being something more than just a legal definition.” Thus, Tenni-son would not have confused the jury by acknowledging the validity of a law that his professional association argued should not be honored; rather, he clearly explained that the Task Force’s concern was with the use of “psychiatric commitment to effect nonmedical societal ends that cannot be openly evaled.”
Winkle presented an expert who could provide testimony that was both relevant and that could assist the jury in determining a fact in issue. We hold that the trial court erred when it excluded admissible testimony. The exclusion of evidence is likely harmless if the evidence was cumulative, “or the rest of the evidence was so one-sided that the error likely made no difference in the judgment.” Cent. Expressway, 302 S.W.3d at 870. But if erroneously excluded evidence was crucial to a key issue, the error is likely harmful. Id. The jury might not have found Tennison’s opinion, that a nursing home employee who repeatedly became intoxicated and sexually assaulted nursing home patients while at work did not lack volitional capacity, credible. However, when admissible evidence is excluded, the issue is not whether the jury would have failed to find that Winkle suffers from a behavioral abnormality, but rather the issue is whether the judgment turns on the excluded evidence and, consequently the error probably resulted in an improper judgment. City of Brownsville v. Alvarado, 897 S.W.2d 750, 753-54 (Tex.1995). Here, the improperly excluded testimony related to a critical issue submitted to the jury; that is, whether Winkle suffers from a behavioral abnormality that predisposes him to engage in a predatory act of sexual violence. Winkle provided no other expert testimony to support his contention regarding the volitional nature of the conduct relied upon by the State to prove the existence of a congenital or acquired condition that, predisposes Winkle to commit a sexually violent offense. In addition to providing his own opinion about whether Winkle suffers from a behavioral abnor*250mality, Winkle’s admittedly qualified expert could have critiqued the thoroughness of the State’s expert’s analysis of Winkle’s case and whether that analysis was consistent with the appropriate professional manuals.
The Texas Supreme Court has recognized the impossibility of prescribing a specific test to determine whether a particular error is harmful, and has entrusted that determination to the sound discretion of the reviewing court. State v. Cent. Expressway Sign Assocs., 302 S.W.3d 866, 870 (Tex.2009). However, “it is not necessary for the complaining party to prove that ‘but for’ the exclusion of evidence, a different judgment would necessarily have resulted.” McCraw v. Maris, 828 S.W.2d 756, 758 (Tex.1992). The complaining party must only show “that the exclusion of evidence probably resulted in the rendition of an improper judgment.” Id. The role that excluded evidence plays in the context of the trial is important. Cent. Expressway, 302 S.W.3d at 870. “[I]f erroneously admitted or excluded evidence was crucial to a key issue, the error is likely harmful.” Id. The reason the jury’s verdict and the resulting judgment in this case was “improper” for the purposes of Rule 44.1(a)(1) of the Texas Rules of Appellate Procedure is that Tennison’s testimony, which the trial court refused to allow the jury to consider, was the only evidence the defendant had favoring a finding that he would not, beyond a reasonable doubt, likely reoffend, and that the defendant’s risk of reoffending was one of the central disputed issues to be guided by expert testimony. See Tex.R.App. P. 44.1(a)(1). In that light, we cannot view the excluded testimony as harmless.
We hold the trial court committed harmful error, and accordingly, we sustain Winkle’s second issue. Tex.R.App. P. 44.1(a). Because Winkle’s other issues would not result in greater relief, it is not necessary that we address them. Tex.R.App. P. 47.1.
REVERSED AND REMANDED.
. In the course of the hearing on the State's motion, the participants frequently referred to two other SVP commitment cases in which the trial court sustained the State’s objection to Tennison’s testimony as an expert. See In re Hinkle, No. 09-09-00548-CV, 2011 WL 2420884, at *6-12 (Tex.App.-Beaumont June 16, 2011, pet. filed) (mem. op.); In re White, No. 09-10-00018-CV, 2011 WL 1420760, at *1 (Tex.App.-Beaumont Apr. 14, 2011, no pet.) (mem. op.). We reversed the judgment in Hinkle due to what we held was the improper exclusion of Tennison’s proffered expert testimony. Hinkle, 2011 WL 2420884, at *12. As Tennison noted in the pre-trial hearing in this case, upon re-examining White, Tennison concluded that White had a behavioral abnormality because he “showed a propensity for volitional incapacity to a very high degree of likelihood[.]’’ In White, appellate counsel certified that no arguable issues would support an appeal, and we affirmed the commitment order in that case. See White, 2011 WL 1420760, at *1.