¶1
Becker, J.*— Michael Lee Sipin and his friend David Taylor were riding in Sipin’s new BMW Z-3 convertible when it crashed and Taylor was killed. The State charged Sipin with vehicular homicide and, at trial, sought to admit computer-generated simulation evidence using Version 6.2 *406of a program called PC-CRASH to prove that Sipin had been the driver of the Z-3. Following a Frye1 hearing, the trial court admitted the evidence, and the jury convicted Sipin. In light of postverdict evidence, we cannot be confident that a scientific consensus has been achieved among accident reconstructionists that PC-CRASH is capable of accurately performing the predictions to which the State’s expert witness testified. Accordingly, reversal and remand for a new trial with a new Frye hearing is required.
FACTS
¶2 On March 6, 2000, Michael Sipin and David Taylor spent the afternoon together drinking beer, eating, and talking at Sipin’s home in Maple Valley. Sipin owned a new, manual transmission BMW Z-3 convertible with a removable hardtop, and he bragged about the car to Taylor. Sipin subsequently testified that he occasionally had attacks of gout that prevented him from driving a car with a manual transmission. Sipin stated that he had gout on the day in question, was walking with a cane, and could not drive the Z-3, so his daughter had been using the car that day. Sipin’s wife also testified that Sipin had gout on the day in question, could barely walk, and could not drive the Z-3. Paula Luedke, a physician’s assistant who had treated Sipin for many years, confirmed that Sipin suffered from chronic gout, which made it painful for him to walk.
¶3 When Sipin’s daughter arrived home with the Z-3, Taylor wanted to drive it, and Sipin testified that he agreed. Sipin also testified that Taylor had to adjust the driver’s seat as far back as it would go, and had trouble locating reverse, but that eventually Taylor was able to drive the vehicle out of the driveway. Sipin’s daughter testified that Taylor and Sipin left in the vehicle, with Taylor driving, shortly after 5 p.m. Sipin’s wife testified that she saw Taylor in the driver’s seat, and observed that he had some diffi*407culty putting the car into reverse, but that she did not see who was driving when the car left the driveway.
f 4 Sipin testified that the roads around his home were “back country” roads with lots of hills and a 35 mph speed limit. Sipin stated that the roads were not safe to drive at high speeds. Sipin stated that he had planned to have Taylor drive along a scenic, hilly 15-minute loop leading back to the house. Sipin remembered Taylor turning onto the proper road but testified that he did not remember the accident itself.
f5 Soon after Taylor and Sipin left Sipin’s home in the Z-3, three neighbors saw the car traveling at a high rate of speed on the rural road near Sipin’s home. They did not see who was driving. Soon after the vehicle passed from their sight, they heard a “screech” and a loud crash. The neighbors called 911 — at 5:11 p.m. — and then ran to the scene. They observed that the Z-3 had hit a mailbox on a metal pole that had been cemented into the ground with a large amount of concrete, ripping the mailbox and lifting the cement — which one witness said looked like it weighed several hundred pounds — out of the ground. After hitting the mailbox, the Z-3 hit a large tree. The State’s accident reconstructionists testified that the vehicle hit the tree rear-first, forcing the passenger-side door forward into the right front quarter panel, and leaving the passenger seat open to the tree.
|6 The witnesses found the Z-3’s hardtop lying away from the vehicle. Both Taylor and Sipin had been ejected from the car. Taylor was found on the ground between the passenger-side door and the tree, with one foot in the car. Sipin was found about 10 to 15 feet behind the vehicle. The passenger-side airbag had been deployed. The gearshift was found on the right side of the car near the tree. None of the neighbors moved the men, the car, or the mailbox, but waited until emergency personnel arrived.
f 7 Taylor sustained serious injuries and died soon after the accident. Sipin suffered permanent brain damage. He *408later stipulated that his blood alcohol level had been 0.11 two hours after the crash.
¶8 Sipin was charged with one count of vehicular homicide, as the driver, under RCW 46.61.520(l)(a).
Frye Hearing
¶9 Sipin filed a motion to exclude the State’s accident-reconstruction evidence that had been generated by use of a computer program called PC-CRASH, as well as the accompanying testimony by Ronald D. Heusser, the State’s PC-CRASH expert witness. This evidence was offered by the State to prove that Sipin had been the driver of the Z-3 at the time of the crash. Sipin asserted that the computer-generated evidence was not admissible under the Frye test for admissibility of novel scientific evidence. A Frye hearing was held to determine the admissibility of the computer-generated evidence. Heusser, an engineer and accident reconstructionist, testified at the Frye hearing regarding his use of the computer program PC-CRASH to simulate the movement of the occupants in the accident.
¶10 Heusser testified that PC-CRASH is a simulation program, rather than an animation program. Heusser stated that “inputting” variables from the scene and the vehicle, such as steering, braking, and speed, would create a predictive image of the vehicle movement, based on the laws of physics. Heusser stated that he had previously testified at trials regarding the use of PC-CRASH for simulation of accidents involving single vehicles over rough terrain, multiple-vehicle accidents, and accidents involving vehicles and pedestrians. He testified that none of these cases had been appealed.
¶11 Heusser provided two studies, one from 1996, and another from 1999, for purposes of the Frye hearing.2 The 1996 study, “Validation of PC-CRASH — A Momentum-*409Based Accident Reconstruction Program,” showed a comparison between staged collisions of vehicles that measured tire marks, speed, and direction, and PC-CRASH simulations, and found that the computer simulations predicted speeds in car crashes that were in agreement with “real world” results. The 1999 study, “The Pedestrian Model in PC-CRASH — The Introduction of a Multi Body System and its Validation,” was the first study of the use of a very simplistic human model in an updated version of PC-CRASH that provided a “multi-body” option for purposes of simulating the interaction between a pedestrian and the outside of a vehicle.
¶12 Heusser provided no validation studies that had been done on the use of the multi-body option of PC-CRASH to simulate the movement of a human body within the interior of a vehicle during a car accident. However, Heusser asserted that the principles used to predict interaction between a human body and the outside of a vehicle were the same as the principles used to predict interaction between a human body and the inside of the vehicle. Heusser testified that he was not aware of any significant debate in the accident-reconstruction community about the reliability of PC-CRASH for showing interior-occupant movement.
¶13 Heusser stated that to reconstruct the accident in the present case, he viewed King County survey data, a map and measurements of the scene, diagrams, the vehicle, digital photographs of the tire tracks, and friction tests on various surfaces involved in the accident. Heusser used the multi-body option of Version 6.2 of PC-CRASH to make his predictions. Heusser testified that he spent a lot of time “inputting” various speeds and turns until the PC-CRASH simulation followed the tire tracks documented by police at the scene. Heusser stated that he then entered the weight and body type of each occupant into the computer program to determine interior occupant movement. He asserted that *410these calculations had been validated by a Society of Automotive Engineers paper but did not identify the study.
¶[14 Sipin presented no witnesses at the Frye hearing and presented no literature. However, Sipin objected throughout the hearing that the foundation for Heusser’s conclusions was inadequate because PC-CRASH had neither been peer-validated for use as a simulation to predict interior-occupant movement during a multi-collision accident, nor admitted as evidence for such a purpose in Washington courts. The trial court overruled these objections. Nevertheless, during cross-examination, Heusser admitted that none of the prior cases in which he used PC-CRASH to simulate vehicle accidents involved interior occupant movement simulations. Heusser also admitted that the 1996 study showed that the results of PC-CRASH were not always entirely accurate and that no studies currently existed that validated PC-CRASH for use in simulating the interaction between a person and the interior surfaces of a vehicle during an accident. But he reasserted that the principles involving the exterior of a vehicle and a person were applicable to interior vehicle simulations.
¶15 Heusser also admitted that the 1999 study, which involved the use of PC-CRASH for simulation of pedestrian/ vehicle accidents, did not indicate that the program could be used to predict interior-occupant movement and that the study additionally concluded that further validation tests would be required to prove the results of the illustrated simulations. Heusser stated that such validation studies were undertaken in 2000. The State failed to provide the studies, but Heusser testified that the paper regarding the studies focused on validation of multi-body/person interaction with vehicle exterior surfaces, not interior surfaces.
¶16 The trial court ultimately allowed Heusser to testify as an expert witness at trial and ruled that he could rely on the results of his PC-CRASH simulation as the basis for his expert opinion. The court held that “a sufficient showing has been made that the PC-CRASH data is the type of data *411reasonably relied upon by experts in the accident reconstruction field in forming opinions or inferences upon the subject.” The court also stated that although an actual model for occupant movement within the vehicle was not presented, the mathematical principles that would support that model were illustrated by the studies validating simulations involving pedestrians and the exteriors of vehicles and noted that the testimony regarding this issue was unrebutted and unimpeached.
Trial Testimony and Evidence
¶17 At trial, the State presented evidence and expert testimony from accident reconstructionists, including testimony by Heusser and a video created by his inputs into the PC-CRASH program, to support its theory that Sipin was driving the vehicle when the accident occurred. Sipin likewise presented evidence and expert testimony from accident reconstructionists to support his defense that he was not driving the vehicle when the accident occurred.
¶18 During closing argument to the jury, the prosecutor described PC-CRASH as a computer program that
essentially takes the laws of physics and reduces them to mathematical calculations that can be done over and over again to generate an accurate picture of what happened during a collision based on the tire marks at the scene, based on the physical evidence in the case such as the damage to the car, as well as the conditions that can be observed at the scene.
¶19 Sipin was convicted as charged.
Sipin’s Motions for a New Trial
¶20 After his conviction, Sipin filed a motion for a new trial in which he again argued that PC-CRASH had not been validated for the use exercised by Heusser in reconstructing the accident. Sipin’s motion was supported by the declaration of Brian McHenry, an expert in accident reconstruction who was involved in the creation of mathematical *412models and computer programs for simulations of vehicle collisions. McHenry asserted that the multi-body version used by Heusser had not been validated to simulate or predict the movement of interior occupants and that no computer model currently existed that had been validated as a predictive model for detailed occupant kinematics.
¶21 Sipin also submitted three studies that he believed undermined the validity of Heusser’s use of the multi-body option in PC-CRASH as a predictive model for occupant movement in car crashes.3 A 1999 paper, “A New Approach to Occupant Simulation Through the Coupling of PC-Crash and MADYMO,” studied the evolving science of vehicle-occupant movement through the joined use of two computer programs. This study used PC-CRASH solely for predicting vehicle movements and “inputted” results of those simulations into MADYMO (MAthematical DYnamic MOdel), a computer program that was validated to simulate occupant movement. The study did not use the multi-body option in PC-CRASH. It also emphasized that the modeling dummy used to illustrate occupants was validated only for frontal collisions and not for multiple impacts and was to be used solely for the purpose of evaluating different restraint systems.
¶22 A 2002 paper, “Methods of Occupant Kinematics Analysis in Automobile Crashes,” analyzed various bases for evaluating free body motion within a vehicle, including plotting of vehicle motion to determine occupant motion. The examples utilizing PC-CRASH in the paper were simplistic, based on movement of one vehicle, and did not validate PC-CRASH to predict multiple-occupant movement in a multi-collision situation where the occupants are ejected from the vehicle. Another 2002 paper, “Occupant Kinematics in Forensics: Evaluating the Appropriateness *413and Applicability of an ATB Application,” was authored by McHenry and described the development of the Articulated Total Body (ATB) model computer program, which is used to simulate the dynamic motion of jointed systems of rigid bodies. This paper emphasized that although ATB was the most evolved model for vehicle occupants, it modeled a simple passive dummy occupant and had “never been validated as a general predictive occupant kinematics simulation model for any type of real-world accidents.” Clerk’s Papers at 162.
¶23 The trial court denied Sipin’s motion for a new trial, noting that the information provided by Sipin had been available when he made his first motion to exclude the evidence, prior to trial. The judge also stated that he was not certain that he would have rejected Heusser as an expert witness on the basis of McHenry’s opinions.
f 24 Sipin renewed the motion, attaching an assessment of Heusser’s PC-CRASH simulation from Boyd Allin of Maclnnis Engineering Associates, Inc., which is the distributor of PC-CRASH for North America. Allin emphasized that the multi-body model PC-CRASH program had not been validated for use in modeling the interaction of occupants within the vehicle interior, and that Heusser’s use represented “an overextension of the capabilities of the model.” Clerk’s Papers at 310.
¶25 The court denied Sipin’s renewed motion, finding it to be a collateral attack on the previous order denying the first motion for a new trial and that the evidence was of the same type offered at the first motion.
¶26 Sipin appeals the admission of the car crash simulation evidence generated by PC-CRASH and the accompanying testimony of Heusser, together with the trial court’s denial of his motions for a new trial.
DISCUSSION
¶27 In Washington, the Frye test is used to determine the admissibility of novel scientific evidence. State v. *414Copeland, 130 Wn.2d 244, 261, 922 P.2d 1304 (1996) (citing Frye v. United States, 293 F. 1013 (D.C. Cir. 1923)). While this court generally reviews the trial court’s decision on a motion for a new trial for abuse of discretion, Palmer v. Jensen, 132 Wn.2d 193, 197, 937 P.2d 597 (1997), review of admissibility of evidence under the Frye test is de novo. Copeland, 130 Wn.2d at 255. Moreover, we are not limited to the evidence that was before the trial court with respect to the Frye admissibility issues and may undertake a searching review of scientific literature as well as secondary legal authority before rendering a decision. Id. at 256. A key reason for this expanded scope of appellate review is that “it is impractical in many instances for a true cross-section of scientists to testify at a hearing.” Id. We may not properly sustain a trial court’s determination regarding admissibility on a mere finding that the record contains sufficient evidence of the reliability of the challenged scientific method. State v. Cauthron, 120 Wn.2d 879, 887, 846 P.2d 502 (1993).
¶28 Under the Frye test, scientific evidence is admissible if it is generally accepted in the relevant scientific community, but not admissible if there is a significant dispute among qualified experts as to its validity. Copeland, 130 Wn.2d at 255. Under Frye, novel scientific evidence is admissible where (1) the scientific theory or principle upon which the evidence is based has gained general acceptance in the relevant scientific community of which it is a part and (2) there are generally accepted methods of applying the theory or principle in a manner capable of producing reliable results. State v. Riker, 123 Wn.2d 351, 359, 869 P.2d 43 (1994).
The reliability of evidence derived from scientific methods depends upon three factors: “(1) the validity of the underlying principle, (2) the validity of the technique applying that principle, and (3) the proper application of the technique on a particular occasion.” (Footnotes omitted.) Gianelli, The Admissibility of Novel Scientific Evidence: Frye v. United States, a Half-Century Later, 80 Colum. L. Rev. 1197, 1201 (1980). The *415first two factors are critical with regard to the admissibility of evidence derived from a novel scientific technique.
State v. Huynh, 49 Wn. App. 192, 194-95, 742 P.2d 160 (1987).
¶29 However, if the proffered evidence does not involve new methods of proof or new scientific principles, then the Frye inquiry is not necessary. State v. Ortiz, 119 Wn.2d 294, 311, 831 P.2d 1060 (1992). This is because full acceptance of a process in the relevant scientific community obviates the need for a Frye hearing. State v. Russell, 125 Wn.2d 24, 41, 882 P.2d 747 (1994).
¶30 The State argues that no Frye hearing was necessary in this case because the laws of physics are well known, and all that Heusser did was to use PC-CRASH to apply mathematical equations and the laws of physics — as accident reconstructionists have always done — utilizing a computer program rather than a slide rule and a pocket calculator. We agree with the trial court that a Frye hearing was necessary. Jurisdictions that have addressed the issue uniformly hold that the admissibility of computer-generated models or simulations (as opposed to animations) as substantive proof or as the basis for expert testimony regarding matters of substantive proof is conditioned upon a sufficient showing that (1) the computer is functioning properly, (2) the input and underlying equations are sufficiently complete and accurate (and disclosed to the opposing party so that they can be challenged), and (3) the program is generally accepted by the appropriate community of scientists for use in the particular situation at hand. See Commercial Union Ins. Co. v. Boston Edison Co., 412 Mass. 545, 549, 591 N.E.2d 165 (1992); Commonwealth v. Fatalo, 346 Mass. 266, 269, 191 N.E.2d 479 (1963); Bray v. Bi-State Dev. Corp., 949 S.W.2d 93, 98 (Mo. Ct. App. 1997); Kudlacek v. Fiat S.p.A., 244 Neb. 822, 509 N.W.2d 603, 617 (1994); State v. Clark, 101 Ohio App. 3d 389, 655 N.E.2d 795, 812 (1995). Contrast Commonwealth v. Serge, 58 Pa. D. & C. 4th 52 (2001) (distinguishing between computer-generated animations and computer-generated simula*416tions, in terms of tests for admissibility). We agree with these courts, and hold that in Washington, computer-generated simulations used as substantive evidence or as the basis for expert testimony regarding matters of substantive proof must have been generated from computer programs that are generally accepted by the appropriate community of scientists to be valid for the purposes at issue in the case.
Our Independent Review
¶31 This court requested that the State provide the PC-CRASH manual for Version 6.2, the version used by Heusser. We appreciate that the State provided both the Technical Manual and the Users Manual for PC-CRASH version 6.2. We have examined these manuals to determine whether the program parameters include the uses described by Heusser at trial, that is, the prediction of movement of two occupants of a vehicle in a multi-collision accident. It is not clear that they do.
¶32 The Technical Manual primarily describes the mathematics involved in the computer program. This manual briefly describes the multi-body model, the model that Heusser used to make his simulation. While the manual states that this model may be used “to correlate pedestrian and occupant injuries to vehicle damage areas,” the primary focus of this section of the manual is the use of the model for purposes of pedestrian/vehicle impacts. (Tech. Man. at 51-62).
¶33 The User Manual for version 6.2 of PC-CRASH expands the description of the multi-body model and additionally states that a multi-body occupant option “allows impacts to be automatically calculated between the occupant’s multibody components and the vehicle’s interior or other occupant multibodies.” However, the User Manual warns that speeds over 25 mph “can cause the multibody shape to go through the vehicle outline and then rebound unrealistically due to a force much higher than could normally occur.” (User Man. at 57, 223).
*417¶34 The User Manual describes using a 3D drafting program, AutoCAD, to create surfaces in a vehicle for purposes of calculating a contact point, but does not mention its use for modeling a removable convertible rooftop that flexes and comes off during a collision. The User Manual also does not provide information on modeling interaction between the occupant and interior components, airbags, or removable convertible hardtops. And, while the User Manual provides information on how to change the default settings for a single impact to assist with simulation of secondary impacts for vehicles, the manual does not describe the use of the occupant model in a multiple or secondary impact situation like that at issue in the present case.
|35 The User Manual states that the MADYMO (MAthematical DYnamic MOdel) is a more comprehensive multi-body occupant option and that this additional option is provided in version 6.2 of PC-CRASH to carry out more detailed simulations of restrained or unrestrained occupant movement. The MADYMO model includes a seat with changeable parameters, a steering wheel, lap and torso seat belt options, and airbags. However, the User Manual cautions that only one MADYMO occupant can be modeled at a time and that the occupant cannot be out of a normal seated position at the start of the impact. The manual also warns that the occupant height is defaulted at five feet nine inches tall and cannot be changed. The steering wheel geometry, foot well, and seat cushion dimensions are fixed. The manual does not describe whether or how the MADYMO model could be used in a multi-impact collision.
¶36 Thus, the PC-CRASH manuals describe several limitations on the use of both the occupant model and the MADYMO model to simulate occupant movement in a vehicle accident. Additionally, there is no mention in either manual of the use of the occupant model to predict complicated interior occupant movement in a multi-collision situation, taking into consideration vehicle interior components, multiple impacts, and a removable convertible *418hardtop. Further, while the MADYMO model is the most comprehensive occupant model available to the program version used as evidence at trial, this model was not available for more than one occupant and was not used by Heusser in the present situation, in any event. Most importantly, neither manual indicates that either occupant model has been validated for use in accurately predicting or simulating occupant movement in multiple impact situations.
¶37 While this court may misunderstand the parameters of the PC-CRASH program based solely on these manuals, these possible program limitations are reflected in the declarations and papers provided by Sipin in his motions for a new trial. The declarations and other evidence indicate that the program might not be generally accepted in the relevant scientific community to predict or simulate complicated interior-occupant movement in a multiple-collision accident, the purpose to which the software was put to produce evidence for trial in this case.
¶38 Brian McHenry, who had assisted in the creation of several car crash simulation programs and was fully familiar with the limitations of PC-CRASH, stated that the program should not be used as the sole basis for accident reconstruction conclusions. McHenry asserted that the multi-body version used by Heusser had not been validated to simulate or predict the movement of interior occupants and that no computer model currently existed that had been validated as a predictive model for detailed occupant kinematics.
¶39 Boyd Allin of Maclnnis Engineering Associates, Inc., the distributor of PC-CRASH for North America, emphasized that the multi-body model PC-CRASH program had not been validated for use in modeling the interaction of occupants within the vehicle interior and that Heusser’s use represented “an overextension of the capabilities of the model.” Allin also stated that the PC-CRASH program could not calculate the speed change of a vehicle when it strikes a pole and pulls it out of the ground.
*419¶40 The three papers submitted by Sipin posttrial also showed the limitations of the PC-CRASH program. At the least, the 2002 paper authored by McHenry, “Occupant Kinematics in Forensics,” emphasized that the ATB model used for vehicle occupants in simulation programs had “never been validated as a general predictive occupant kinematics simulation model for any type of real-world accidents.” Clerk’s Papers at 162.
¶41 In sum, the two posttrial declarations provided by Sipin, the three papers that he submitted posttrial, and the PC-CRASH manuals themselves suggest that there is a lack of consensus in the relevant scientific community regarding the use of this PC-CRASH program for the purposes to which it was put by Heusser. We have examined other publications from the Society of Automotive Engineers (SAE) but have not been able to locate any documents that support the State’s position that PC-CRASH has been validated, or is universally accepted by the relevant scientific community, as an accurate predictive model for detailed occupant kinematics in a multi-impact accident.4
¶42 It is not our task to determine whether a scientific method or theory is correct. Such is beyond the expertise of courts. Instead, it is our task to determine whether the appropriate scientific community has generally reached *420consensus that the method or theory is reliable. Cauthron, 120 Wn.2d at 887.
|43 Although much of the evidence provided by McHenry and Allin clearly goes to the weight to be accorded Heusser’s testimony, much of it also relates to admissibility of the evidence under Frye, because it illustrates the limitations that two members of the relevant scientific community see in the computer program at issue — and explains why we think that the relevant group of scientists have not reached consensus that PC-CRASH is reliable for the uses that Heusser attempted to make of the program for purposes of this trial.
¶44 Notwithstanding Sipin’s failure to present expert testimony and supporting documentation that was available at the time of the Frye hearing, this court cannot properly affirm a conviction when postverdict developments so strongly indicate that the evidence relied upon by the State is not supported by the relevant scientific community. De novo appellate review sometimes requires that the court review scientific evidence that was not presented to the trial court even though it was available, as is the case here. Once an appellate court determines that the Frye test has been met as to a specific novel scientific theory, the trial courts generally may rely upon that ruling to govern admissibility of the same theory in subsequent cases. Cauthron, 120 Wn.2d at 888 n.3. Because the postverdict submissions cast doubt on the trial court’s conclusion that the challenged evidence in this case was admissible under Frye, we cannot countenance the State’s use of such evidence here or in future cases — and that will remain true until such time as the State demonstrates that the relevant scientific community has reached consensus with respect to the validity and reliability of the PC-CRASH program used by Heusser, for the specific purposes here at issue.
¶45 After this case was argued, another division of this court addressed the admissibility of evidence generated by a PC-CRASH program and affirmed the trial court’s admission of the evidence under Frye. See State v. Phillips, 123 *421Wn. App. 761, 98 P.3d 838 (2004), review denied, 154 Wn.3d 1014 (2005). There, the defendant in a vehicular homicide case admitted to driving the vehicle diming the accident, in which his car ran off the road and hit a pole. At issue was whether the evidence generated by PC-CRASH could be admitted to discount the driver’s version of events — that he had been driving at the speed limit while negotiating a curve but swerved to miss a deer and lost control of the car. In Phillips, the PC-CRASH program was used to predict movement of the vehicle in a single-impact crash, and the relevant scientific community of accident reconstructionists agreed that the computer program was reliable for that purpose. Our decision on the admissibility of the proffered evidence is different because here the State sought to use PC-CRASH for a purpose that is strongly disputed by the relevant scientific community: use of the multi-body version of PC-CRASH to predict interior occupant movement in a multi-impact accident.
Harmless Error
¶46 The State argues that even if the evidence was erroneously admitted, the error was harmless. See State v. Leuluaialii, 118 Wn. App. 780, 77 P.3d 1192 (2003) (erroneous admission of canine DNA (deoxyribonucleic acid) evidence despite failure of scientific process to meet Frye test required reversible error examination), review denied, 154 Wn.2d 1013 (2005). Sipin must show “within reasonable probabilities” that but for the alleged error the outcome of his trial would have been different. State v. Smith, 106 Wn.2d 772, 780, 725 P.2d 951 (1986) (erroneous admission of defendant’s prior burglaries in unrelated rape case required reversible error examination). However, the record compels the conclusion that the outcome of the trial might reasonably have been different if the trial court had excluded the challenged evidence. The primary issue at trial was the identity of the driver. Exclusive of the evidence generated by the PC-CRASH program, both Sipin and the State presented persuasive evidence regarding the identity of the driver. We conclude that within reasonable probabili*422ties, but for the error in admitting the computer-generated simulation evidence, the outcome of his trial might have been different.
¶47 On summary, the record developed in the trial court was shown by postverdict evidence and argument to be insufficient to sustain the reliability of the PC-CRASH evidence under Frye, and a new trial is required. Fairness to the State dictates that the State be permitted to try again to establish the reliability of the evidence. See, e.g., People v. Reilly, 196 Cal. App. 3d 1127, 1133, 242 Cal. Rptr. 496 (1987) (noting prior remand for new trial after previous appeal had found evidence did not meet Frye standard), cited with approval in Cauthron, 120 Wn.2d at 887.
¶48 Reversed and remanded for a new trial, with a new Frye hearing.
Baker, J., concurs.
Reconsideration denied November 21, 2005.
The initial opinion in this case was written by Judge Faye C. Kennedy who passed away on September 16, 2005. The court granted respondent’s Motion for Reconsideration and withdrew the opinion. The court now files this substitute opinion.
See Frye v. United States, 293 P. 1013 (D.C. Cir. 1923).
See William E. Cliff & Darcy T. Montgomery, Validation of PC-CRASH — A Momentum-Based Accident Reconstruction Program, SEA International (1996) (Ex. 8); Andreas Moser, Hermenn Steffan & Gustav Kasanicky, The Pedestrian *409Model in PC-CRASH — The Introduction of a Multi Body System and its Validation, Society of Automotive Engineers (1999) (Ex. 9).
See Hermenn Steffan and B.C. Geigl, A New Approach to Occupant Simulation Through the Coupling of PC-Crash and MADYMO, SAE International (1999); Jon. E. Bready et al., Methods of Occupant Kinematics Analysis in Automobile Crashes, SAE International (2002); Brian G. McHenry, Occupant Kinematics in Forensics: Evaluating the Appropriateness and Applicability of an ATB Application, ATB Users’ Group Conference (2002).
The publications include a study mentioned but not provided by Heusser during his Frye hearing testimony, which validates the PC-CRASH pedestrian model (SAE #2000-01-0847). The court also reviewed a study validating the use of PC-CRASH/MADYMO coupling in 2000 (SAE #2000-01-0471), which makes no mention of occupant modeling or occupant movement in PC-CRASH. Because Heusser did not use the MADYMO coupling, SAE #2000-01-0471 is not relevant to our discussion here. Other SAE publications include a study on stability analysis within PC-CRASH which discusses the development/application of an algorithm that can be used to determine uncertainty ranges for each input parameter in PC-CRASH, but which does not mention occupant kinematics (SAE #2003-01-0488); a paper addressing the validation of PC-CRASH/MADYMO program coupling for purposes of validating vehicle rollovers, which also does not discuss multiple occupant movement in multi-collision accidents (SAE #2001-01--0176); a study on uncertainty values in accident reconstruction using PC-CRASH, which made no mention of occupant kinematics or modeling of interior occupant movement in PC-CRASH (SAE #2003-01-0489); and a paper involving evidentiary considerations of computer generated exhibits that made no mention of PC-CRASH (SAE #1999-01-0101). None of these articles supports a conclusion that PC-CRASH has ever been validated for the purpose for which it was at trial here.