Appellant was charged in separate indictments for possession of a motor vehicle with an altered vehicle identification number and possession of cocaine. At a pre-trial hearing, the trial court granted the State's motion to consolidate the two indictments for trial over appellant's objection. The jury found appellant not guilty of possession of a controlled substance and guilty of possession of a motor vehicle with obliterated identification numbers. After making findings of true as to the allegations in the two enhancement paragraphs, the trial court assessed appellant's punishment at confinement for twenty-five years in prison.

The Seventh Court of Appeals concluded that the trial court erred in overruling appellant's mandatory right to a severance. It held that under Cain v. State, 947 S.W.2d 262 (Tex. Crim. App. 1997), the severance error in this case was not proven harmless. Llamas v. State, 991 S.W.2d 64, 70 (Tex. App.-- Amarillo 1998). In so holding, the Court stated that error in denying appellant's mandatory right to sever pursuant to a legislative command is precisely the type of error contemplated by us in Cain, wherein we explained that some errors defy analysis by harmless error standards. Id. at 69. Accordingly, the judgment of the trial court was reversed and the case remanded for a new trial. Id. We granted the State's petition for discretionary review to consider whether the Court of Appeals erred in failing to conduct the appropriate harm analysis regarding the trial court's error in denying appellant's motion to sever.⁽¹⁾ We will affirm.

Initially, the State argues that the Court of Appeals ignored the mandates of Cain and Rule 44.2 of the Texas Rules of Appellate Procedure by relying on Warmowski v. State, 853 S.W.2d 575 (Tex. Crim. App. 1993), for the proposition that a trial judge's error in failing to grant a severance is not subject to a harm analysis. However, our reading of the Court of Appeals opinion does not lead us to this conclusion.

While a cursory examination of the Court of Appeals opinion might lead a reader to conclude that a harm analysis was foreclosed by the Court of Appeals in the present case, such a reading is incorrect. From the outset, the Court of Appeals recognized that a Rule 44.2(b) harm analysis must be conducted pursuant to Cain. Llamas, 991 S.W.2d at 68. Indeed, the Court first acknowledged that the proper standard of review to be used when determining if a substantial right is affected pursuant to a Rule 44.2(b) analysis is whether the error had a substantial or injurious effect or influence in determining the jury's verdict. Llamas, 991 S.W.2d at 68. (citing King v. State, 953 S.W.2d 266 (Tex. Crim. App. 1997) (citing Kotteakos v. U.S., 328 U.S. 750, 776, 66 S.Ct. 1239, 1253, 90 L.Ed. 1557 (1946)). In answering this inquiry, it gleaned guidance from our language in Warmowski regarding the importance of a defendant's right to severance:

[t]he legislature cannot have intended for an appellate court to second guess [a defendant's] discretion on a case-by-case basis by declaring the failure to grant his requested severance to be harmless beyond a reasonable doubt. . . . It would be anomalous to take away the benefit of that absolute discretion, by invoking Rule 81(b)(2) on appeal, while the defendant still risks the adverse consequences of consecutive sentences.

Warmowski v. State, 853 S.W.2d 578.

In Warmowski, we held that a trial court's failure to sever cases despite a defendant's objection to consolidating cases for trial was reversible error not subject to a harm analysis. Id. at 581. Four years later, in Cain, we held that except for certain federal constitutional errors labeled by the United States Supreme Court as structural, no error, whether it relates to jurisdiction, voluntariness of a plea, or any other mandatory requirement, is categorically immune to a harmless error analysis. Cain, 947 S.W.2d at 264. Consequently, the language in Warmowski indicating that severance error is never subject to a harm analysis is disavowed. However, Warmowski is still instructive regarding the importance of a defendant's right to severance and the concerns reviewing courts should examine when conducting a harm analysis.

Given our strong language in Warmowski, the Court of Appeals reasoned, "to hold that a mandatory right established by the state legislature can be ignored by trial judges simply because it is not a constitutional right, is to allow the judicial branch to ignore mandatory directives of the legislature." Llamas, 991 S.W.2d at 68. Based on this reasoning, the Court concluded that the trial court's error in denying appellant's request for severance affected a substantial right and was not one that must be disregarded under Rule 44.2(b). Id. So, while there is language in Warmowski that is contrary to Cain, the Court of Appeals did not rely on or follow that particular language.⁽²⁾

The Court of Appeals did not ignore the directives of Cain in this case. On the contrary, it determined that this is precisely the type of case that we were talking about in Cain when we said: "where error defies harm analysis or the data is insufficient to conduct harm analysis, the error will not be proven harmless." Llamas 991 at 68, (citing Cain v. State, 947 S.W.2d at 264). In Cain we explained that if concrete data necessary to conduct a harm analysis is absent, a harmless error test must nevertheless be conducted, and the absence of data is simply taken into account in determining whether or not the harmless error test is passed or failed. Cain, 947 S.W.2d at 264. When we stated in Cain that some errors may "defy" harm analysis we did not mean that a harm analysis need not be conducted. We meant simply that some errors will not be proven harmless because harm can never be determined due to the lack of data needed for analysis. This is precisely the conclusion that the Seventh Court of Appeals reached in the instant case.

While we agree with the Court of Appeal's interpretation of our language in Cain, we disagree with its determination that the record contained insufficient data to measure harm. This is not a case where the error defies harm analysis; everything in the court reporter's record is data from which a court can analyze harm. To judge the likelihood that harm occurred, appellate courts must consider everything in the record including all the evidence admitted at trial, the closing arguments, and, in this case, the jurors' comments during voir dire.

According to the State, the harm that severance was meant to prevent is the risk that a defendant would suffer consecutive sentences. It argues that consecutive sentences were ultimately impossible in the instant case because appellant was found not guilty of the possession of a controlled substance that was tried with the instant offense. However, various potentialities for harm other than consecutive sentencing exist when a defendant is denied the right to severance of offenses. The rule allowing severance rests upon two legitimate concerns: (1) that the jury may convict a 'bad man' who deserves to be punished -- not because he is guilty of the crime charged but because of his prior or subsequent misdeeds; and (2) that the jury will infer that because the accused committed other crimes, he probably committed the crime charged. 2 Wayne R. LaFave, Israel & King, Criminal Procedure Criminal Practice Series, 17.1(d)(2nd ed. Supp. 1999). The Court of Appeals was cognizant of these concerns, as well as our cautionary language in Warmowski, and conducted its harm analysis by following the directives of Cain coupled with the guidelines of Rule 44.2(b). It noted that if not for the error, the jury would not have heard the evidence of appellant's cocaine charge which was admitted at trial. It also noted the following excerpt of the record from jury selection:

Prosecutor: Do you think that might affect you to vote - - cause you to vote guilty, even though you weren't convinced beyond a reasonable doubt?

Venireman: If I felt the drug charge was going to stick then yeah, I might

think he knew what he was doing.

***

Venireman: I would be more likely - - if the drug charge was proven and it was almost to the very top of saying - - you know, you've just got a little bit - - couldn't prove he knew that the car was stolen, then I would be thinking, well, yeah, he did know it was stolen too.

The court concluded that although ultimately most prospective jurors indicated that they would require the State to prove each charge beyond a reasonable doubt, the record reflected that the very threat section 3.04 allows a criminal defendant to avoid was present in this case.

Ironically, despite its holding that the data was insufficient to perform a harm analysis, it essentially performed one when it made the following observations:

It is conceivable that appellant would not have been convicted for either offense had he been granted his request for separate trials. Clearly, evidence as to the cocaine charge was admitted in this trial that would otherwise not be relevant to the possession of altered identification numbers and vice versa. The record in this case shows that no less than five times during voir dire, venire persons commented in front of the entire jury panel that they were concerned that appellant had two charges against him. This is exactly the type of harm that section 3.04 allows a criminal defendant, in his sole and absolute discretion, to choose to avoid.

Llamas, 991 S.W.2d at 69.

The conclusion we reach from our examination of the record is consistent with the aforementioned observations of the Court of Appeals. As such, we can not be sure that the error did not have a substantial or injurious effect on the jury's verdict. Kotteakos v. United States, 328 U.S.750, 66 S.Ct. 1239, 90 L.Ed.1557 (1946); King v. State, 953 S.W.2d 266 (Tex. Crim. App. 1997). Therefore, we affirm the judgment of the Court of Appeals.

Price, J.

Delivered: February 16, 1999

Publish

1. The actual ground for review reads as follows: The Amarillo Court of Appeals erred in failing to conduct the appropriate harm analysis to the trial court's alleged error in allegedly failing to grant a motion to sever.

2. However, in one respect the Court of Appeals does appear to misunderstand a proper Rule 44.2(b) harm analysis by approaching it as if it were a two-step inquiry. The court first asked if the failure to sever affected a substantial right of the appellant. Then, after answering this question in the affirmative, the court implied that some further harm analysis would be necessary but for the fact that the severance error meets the requisites of the first inquiry. Here, the Court of Appeals is mistaken. In order to properly conduct a harm analysis under Rule 44.2(b), an appellate court need only determine whether or not the error affected a substantial right of the defendant. To make this determination, appellate courts must decide whether the error had a substantial or injurious affect on the jury verdict. The very process of reaching this decision, is the performance of a Rule 44.2(b) harm analysis.