This is an appeal from a decision by the reviewing board of the Department of Industrial Accidents (board) affirming a decision of an administrative judge finding that Theresa Canavan (plaintiff or employee) is temporarily unable to work and that her medical treatment is reasonable and necessary. The self-insurer, Brigham and Women’s Hospital (hospital), appealed from the board’s decision to the Appeals Court pursuant to G. L. c. 152, § 12,1 arguing that the medical testimony provided through deposition by the plaintiff’s doctor on diagnosis, disability, and causation was not based on reliable methodology pursuant to the standard set forth in Commonwealth v. Lanigan, 419 Mass. 15, 25-26 (1994), and should not have been admitted by the judge. The Appeals Court held that the medical evidence was properly admitted and affirmed the plaintiff’s award. Canavan’s Case, 48 Mass. App. Ct. 297 (1999). We granted the hospital’s application for further appellate review and reverse the board’s decision.
1. Background. We set forth the factual background. The employee received her bachelor’s degree in nursing from Boston State College in 1980. In September, 1983, she began a full-time position as a nurse in the hospital’s recovery room where she worked until June, 1990, when she began working as a nurse in the hospital’s operating room. As an operating room *306nurse, she was responsible for the safe care of patients during surgery, including preparing the room for surgery, caring for the surgical instruments, and assisting the surgeons. During the workers’ compensation hearing, the employee testified that, while in the operating room, she was subjected to various chemicals including ethylene oxide, formaldehyde, and diesel fuel. On August 6, 1993, at the conclusion of a ten-hour day in the operating room, she experienced a severe headache, nasal congestion, and dizziness. She returned to work on August 9, 1993, but still suffered symptoms, including a fever, headache, and swelling of her nose and right cheek. After arriving at the hospital for work, she was referred to Dr. Arthur M. Laurentano who confirmed her symptoms and prescribed a course of antibiotics. At that time she was diagnosed as having chronic sinusitis and was determined to be disabled. The hospital accepted her medical condition and paid her workers’ compensation benefits.
Dr. Laurentano, whose antibiotic treatment had proved only marginally effective for the employee, referred her to Dr. N. Thomas LaCava. The employee first met with Dr. LaCava in June, 1994. He became her treating physician and provided, by deposition, the expert medical testimony on her behalf at the workers’ compensation hearing. Dr. LaCava is a private practitioner, an instructor in pediatrics at the University of Massachusetts Medical School, a staff pediatrician at Holden District Hospital, St. Vincent Hospital, Worcester Hahnemann Hospital, and The Memorial Hospital, and is on the medical staff at University of Massachusetts Hospital. He is certified in pediatrics by the American Board of Medical Specialties and is certified in environmental medicine by the American Board of Environmental Medicine, a field not recognized by the American Board of Medical Specialties.
In response to the employee’s complaints, Dr. LaCava conducted an extensive medical examination. He took her medical history, performed an examination, and conducted a number of diagnostic tests. Dr. LaCava concluded that the employee suffered from arthritis, paresthesias, organic brain syndrome, chemical induced headaches, immunodeficiency, and multiple chemical sensitivities (MCS) secondary to chemical poisoning, which Dr. LaCava believed was caused by exposure during her employment at the hospital. He testified that MCS “is a systemic reaction of the body with multiple symptoms to multiple kinds *307of chemicals, which may be chemically unrelated, which are commonly present in the every day working and living environment where that environment has not been meticulously cleaned up and had the chemical sources removed.” Dr. LaCava stated that laboratory tests that he conducted provided evidence that the employee suffered from MCS. Dr. LaCava testified that the employee’s injury was caused by chemical poisoning at her work environment and that MCS rendered her totally disabled. Dr. LaCava prescribed a course of treatment that required intravenous infusion of vitamins, in particular high doses of vitamin C, oral nutrient supplements, antibiotics, and heat and sauna therapy.
Dr. Donald D. Accetta testified by deposition for the hospital. Dr. Accetta’s testimony directly contradicted the conclusions reached by Dr. LaCava. Dr. Accetta is certified by the American Board of Allergy and Immunology. He is a private practitioner, a consultant in allergy at New England Medical Center, and the secretary of the New England Society of Allergy, and has served on the board of directors for both the New England Society of Allergy and the Massachusetts Allergy Society. Dr. Accetta examined the employee on two occasions. He testified that the employee’s condition was not caused by chemicals present in her work environment and that MCS is “not accepted as a diagnostic disease by mainstream allergists/immunologists and occupational medicine physicians.” Also, he testified that the course of treatment prescribed by Dr. LaCava is not accepted and not appropriate for the symptoms exhibited by the employee. Dr. Accetta diagnosed the employee as suffering from chronic nonallergic rhinitis caused by nonspecific stimuli that exist in the every day environment. He also stated that the employee’s symptoms have a psychogenic component.
After hearing testimony from the employee and a claims adjustor and reviewing the depositions of Dr. LaCava and Dr. Accetta, the judge concluded that the employee was totally unable to work as the continuing result of an injury (“being highly reactive to low levels of environmental chemicals”) that arose in the course of her employment and that the employee’s medical treatment, including intravenous vitamin C treatment, was reasonable and necessary. The judge made these determinations “based on the opinions of Dr. LaCava regarding disability and causal relationship.”
The primary issue presented by this appeal is whether the *308judge properly admitted Dr. LaCava’s testimony regarding the diagnosis and the cause of the employee’s disability, on which the judge relied for his ultimate determinations. The hospital maintains that the methodology used by Dr. LaCava to reach his opinion regarding MCS and its cause is not generally accepted in the medical field and is not otherwise reliable, and therefore his opinion is inadmissible under the principles set forth in Commonwealth v. Lanigan, supra.
2. Preliminary issues. Before reaching the merits of the hospital’s claim, we address two arguments of the employee. She contends that the hospital’s claim that MCS is not a scientifically reliable diagnosis is barred by the doctrine of judicial estoppel: the principle that “[a] party who has successfully maintained a certain position at a trial cannot in a subsequent trial between the same parties be permitted to assume a position relative to the same subject that is directly contrary to that taken at the first trial.” Paixao v. Paixao, 429 Mass. 307, 309 (1999), quoting East Cambridge Sav. Bank v. Wheeler, 422 Mass. 621, 623 (1996). The purpose of the doctrine is to prevent the manipulation of the judicial process by litigants. See Blanchette v. School Comm. of Westwood, 427 Mass. 176, 184, (1998).
In a lawsuit separate from this action, the employee and others sued the hospital seeking common-law tort damages because of her affliction with MCS. See Niles-Robinson vs. Brigham & Women’s Hosp., Inc., Suffolk Superior Court. Civ. A. No. 96-4260-D (Jan. 2, 1997).2 The defendant filed a motion to dismiss the suit pursuant to Mass. R. Civ. R 12 (b) (6), 365 Mass. 754 (1974), contending that the employee could not maintain a common-law tort action because of the exclusivity provisions of the Workers’ Compensation Act. A judge in the Superior Court agreed and dismissed the suit. Id. The employee contends that the hospital’s argument in its rule 12 (b) (6) motion estops the *309hospital from now arguing that MCS is not a disease that is compensable under the Workers’ Compensation Act.
The employee’s argument, however, misconstrues the function of a motion to dismiss pursuant to rule 12 (b) (6). The purpose of a rule 12 (b) (6) motion is to determine whether, under any set of facts, a plaintiff would be entitled to relief. Nader v. Citron, 372 Mass. 96, 98 (1977). Thus, all factual inferences must be made in favor of the plaintiff. Eyal v. Helen Broadcasting Corp., 411 Mass. 426, 429 (1991). For purposes of its motion to dismiss, the hospital was required to assume that the employee had acquired MCS while at the workplace. By filing a motion to dismiss, the hospital was not admitting that the employee acquired MCS at work; it was only stating that, even if she had acquired MCS at work, she could not maintain an action for common-law tort. The hospital took no position in its rule 12 (b) (6) motion on the nature of the employee’s disease or its cause. Thus, no bar prevents the hospital from asserting in this suit that MCS is not a compensable disease under the Workers’ Compensation Act.
The employee also maintains that the hospital did not properly preserve its objection to the admission of Dr. LaCava’s testimony at the hearing. All the expert physician testimony was taken by deposition pursuant to 452 Code Mass. Regs. § 1.12(5) (1993). At each point during the deposition prior to Dr. LaCava’s offering a conclusion regarding the nature of the employee’s condition or the cause of the condition, the hospital’s attorney objected to his conclusion as lacking a foundation. The hospital’s attorney reiterated these objections in his written closing submission to the judge in which he explicitly referenced the reliability standards established in Commonwealth v. Lanigan, supra at 26. In our view the steps taken by the hospital’s attorney, in the context of a workers’ compensation hearing, were sufficient to preserve the admissibility of Dr. LaCava’s testimony for appellate review.
3. Lanigan analysis. Having determined that the hospital’s claim is properly before the court, we turn to whether the judge properly admitted Dr. LaCava’s testimony. As workers’ compensation proceedings are governed by the rules of evidence applicable in the courts of this Commonwealth, 452 Code Mass. Regs. § 1.11(5) (1993), our decision must be made pursuant to our rules concerning the admission of expert scientific testimony.*3103 Thus, before examining the judge’s decision to admit Dr. LaCava’s diagnosis and causation testimony, we review the basic principles for admission of scientific evidence. See Commonwealth v. Vao Sok, 425 Mass. 787, 796 (1997).
Prior to our decision in Commonwealth v. Lanigan, supra, we required that in most circumstances “the community of scientists involved [must] generally accept[] the theory or process” for it to be admitted in evidence. Commonwealth v. Curnin, 409 Mass. 218, 222 (1991), citing Frye v. United States, 293 F. 1013 (D.C. Cir. 1923). The general acceptance test, or Frye test, often proved to be useful because, if there is general acceptance of a theory or process in the relevant scientific community, the theory or process in question is likely reliable. Commonwealth v. Lanigan, supra at 24. However, we recognized that “strict adherence to the Frye test” could result in reliable evidence being kept from the finder of fact. Id. For example, a new theory or process might be “so logically reliable” that it should be admissible, even though its novelty prevents it from having attained general acceptance in the relevant scientific community. Id.
In order to account for this circumstance, we adopted in part the United States Supreme Court’s reasoning in Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), and held that “a proponent of scientific opinion evidence may demonstrate the reliability or validity of the underlying scientific theory or process by some other means, that is, without establishing general acceptance.” Commonwealth v. Lanigan, supra at 26. We noted, however, that in most cases general acceptance will be the significant, and “often the only, issue.” Id. Thus, we have concluded that a party seeking to introduce scientific evidence may lay an adequate foundation either by establishing general acceptance in the scientific community or by showing that the evidence is reliable or valid through an alternate means. Commonwealth v. Sands, 424 Mass. 184, 185-186 (1997).
In Commonwealth v. Vao Sok, supra at 797, we held that our review of a judge’s Lanigan decision is de novo. We recognized *311that adopting a de novo standard of review was contrary to the position taken by certain Federal appellate courts, which generally adopted an abuse of discretion standard. Id. at 797-798 n.14. We noted that the United States Supreme Court had not yet ruled on this question. Id. In General Elec. Co. v. Joiner, 522 U.S. 136, 141-143 (1997), the Court held that Federal appellate courts reviewing a Daubert determination must apply an abuse of discretion standard. In light of the United States Supreme Court’s decision in Joiner and our commitment in Lanigan to follow the general proposition set forth in Daubert, we again address the proper appellate standard of review for a judge’s Lanigan determination.
The Court in Joiner held that a judge’s determination on the reliability of scientific testimony is no different from other evi-dentiary decisions by a trial judge that are reviewed on appeal under an abuse of discretion standard of review. Id. We agree with this conclusion. The advantage of applying a de novo standard of review for the admission of scientific testimony is that an appellate court may conduct a thorough review of the relevant scientific literature on a particular subject and reach a well-informed decision that will serve as precedent to guide trial judges in future cases. The shortcoming in this approach, however, is that it assumes that “the application of the particular scientific method would not vary from case to case and thus would be worthy of a judicial stamp of approval or rejection as a matter of law from an appellate court.” State v. Alberico, 116 N.M. 156, 169 (1993). In reality, the validity of a scientific theory is not always certain. The process of constant scientific testing, over time, validates certain theories and discredits others. A decision by an appellate court that as a matter of law a certain scientific theory or method is reliable or unreliable may freeze perceptions concerning the evidentiary usefulness of the theory or method without accounting for the evolving state of scientific knowledge.
Moreover, when considering novel scientific testimony there is often limited literature for an appellate court to examine to determine whether a scientific theory or method is reliable. This is part of the rationale for abandoning exclusive reliance on general acceptance under the Frye test. We recognized in the Lanigan case that a scientific method or theory could be so novel that it has not yet had time to appear in the scientific literature and thus to have gained general acceptance in the *312relevant scientific community, but could nonetheless be sufficiently reliable to be the basis of admissible expert testimony. Commonwealth v. Lanigan, supra at 25. Thus, primary reliance by a reviewing court on the scientific literature is inconsistent with the principle in the Lanigan case that reliability can be shown through factors other than general acceptance. Id. at 26. Determining whether novel scientific testimony is reliable often will hinge on the presentations made by the parties in a particular case. A trial judge is required to assess the credibility of various expert witnesses in determining whether proposed scientific testimony is reliable; these determinations are not readily susceptible to de novo appellate review and these determinations may vary appropriately on a case-by-case basis. On balance, we conclude, consistent with the United States Supreme Court, that applying an abuse of discretion standard on appellate review will allow trial judges the needed discretion to conduct the inherently fact-intensive and flexible Lanigan analysis, while preserving a sufficient degree of appellate review to assure that Lanigan determinations are consistent with the law and supported by a sufficient factual basis in the particular case. See Kumho Tire Co. v. Carmichael, 526 U.S. 137, 152 (1999) (noting that abuse of discretion standard provides trial judges needed discretion to apply flexible Daubert standard).
4. Dr. LaCava’s testimony. The hospital challenges the reliability of Dr. LaCava’s opinions that (a) the employee suffered from MCS and (b) her MCS condition was caused by chemicals present in the hospital. The judge admitted both these opinions. We address the challenge to each opinion separately.
a. Diagnosis. The judge in a written opinion adopted Dr. LaCava’s opinion that “the employee is totally disabled due to being highly reactive to low levels of environmental chemicals,” i.e., she suffers from MCS. The Appeals Court concluded that the judge’s decision to admit Dr. LaCava’s diagnosis testimony was proper either under a Lanigan analysis or on an alternative basis. The court stated that “certain expert testimony based on personal observations, clinical experience, or generally accepted scientific techniques need not be subject to the Lanigan analysis.” Canavan’s Case, 48 Mass. App. Ct. 297, 301 (1999). Because Dr. LaCava based his opinion on his clinical experience and personal observations of the employee, the court concluded that the expert evidence was admissible without a Lanigan analysis. We disagree with this conclusion.
*313In Vassallo v. Baxter Healthcare Corp. 428 Mass. 1, 15 & n.15 (1998), we explicitly reserved the question whether expert testimony based on, inter alia, personal observations and clinical experience would be admissible without application of the Lanigan analysis. We noted that the Federal courts that had considered this question had taken the view that such testimony would be subject to scrutiny regarding reliability pursuant to the Daubert decision. Id. at 15. Since the Vassallo case, the United States Supreme Court addressed this question in Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999). In the Kumho Tire case, the Court considered the contention that an engineer’s testimony regarding the cause of a tire explosion should be admissible simply because it was based on the observations of an experienced engineer who specialized in tire analysis. Id. at 145. The Court in the Kumho Tire case held that the engineer’s observations were subject to a Daubert analysis as the question before the judge was whether the expert’s specific observations were sufficiently reliable to support the expert’s ultimate conclusion regarding the cause of the explosion. Id. at 157.
We agree with this conclusion. There is no logical reason why conclusions based on personal observations or clinical experience should not be subject to the Lanigan analysis. “That a person qualifies as an expert does not endow his testimony with magic qualities.” Boston Gas Co. v. Assessors of Boston, 334 Mass. 549, 579 (1956). Observation informed by experience is but one scientific technique that is no less susceptible to Lanigan analysis than other types of scientific methodology. The gatekeeping function pursuant to Lanigan is the same regardless of the nature of the methodology used: to determine whether “the process or theory underlying a scientific expert’s opinion lacks reliability [such] that [the] opinion should not reach the trier of fact.”4 Commonwealth v. Lanigan, 419 Mass. 15, 26 (1994). Of course, even though personal observations are not excepted from Lanigan analysis, in many cases personal observation will be a reliable methodology to justify an expert’s *314conclusion. If the proponent can show that the method of personal observation is either generally accepted by the relevant scientific community or otherwise reliable to support a scientific conclusion relevant to the case, such expert testimony is admissible.5 See Commonwealth v. Sands, 424 Mass. 184, 185-186 (1997).
The Appeals Court held that the judge implicitly conducted a Lanigan analysis and properly admitted Dr. LaCava’s testimony that the employee suffered from MCS. Canavan’s Case, supra at 300-301. The court pointed to the judge’s findings that the diagnostic tests performed by Dr. LaCava were “generally accepted in the community of doctors who understand toxicity,” id. at 300, and that Dr. LaCava had knowledge, training, and experience and conducted diagnostic and laboratory tests on the employee to support its conclusion that Dr. LaCava’s MCS diagnosis was admissible. Id. at 301.
In his deposition testimony, Dr. LaCava stated that the tests he performed were accepted by doctors familiar with environmental toxicity6 to determine whether the employee had been poisoned by chemical exposure. However, Dr. LaCava also testified that not every person who suffers chemical exposure poisoning suffers from MCS. The tests that Dr. LaCava testified were generally accepted showed that the employee had suffered chemical exposure but did not show that she suffered from MCS. There was no evidence in the record to show that Dr. LaCava used a reliable methodology to transform his general finding of chemical exposure to his more specific diagnosis of MCS.
Dr. LaCava referred to “the literature” as confirming the *315existence of MCS and asserted that based on the examination and tests he conducted the employee had symptoms “highly suggestive of . . . environmental sensitivity.” However, Dr. LaCava did not identify any specific studies that showed the existence of MCS based on specific symptoms and did not identify tests that can be performed to prove that a patient suffers from MCS. On cross-examination, he admitted that there was a dispute in the medical community regarding the existence of MCS.7 Thus, the only evidence on this record tending to show that the employee suffered from MCS was Dr. LaCava’s assertion. The purpose of the Lanigan test is to prevent an expert from offering testimony to a fact finder that is not based on reliable methodology. Commonwealth v. Lanigan, supra at 26. We cannot conclude that the expert’s mere assertion that a methodology is reliable is sufficient to pass the Lanigan test absent any other evidence showing its reliability.8 See Kumho Tire Co. v. Carmichael, supra at 157, quoting General Elec. Co. v. Joiner, 522 U.S. 136, 146 (1997) (“nothing in either Daubert or the Federal Rules of Evidence requires a district court to admit opinion evidence that is connected to existing data only by the ipse dixit of the expert”). We conclude that on this record it was an abuse of discretion for the judge to admit Dr. LaCava’s diagnosis testimony.
b. Causation. Our conclusion that the diagnosis testimony was improperly admitted disposes of this case; however, because of the importance of the question, we address the hospital’s contention that it was an error to admit the causation testimony. The judge concluded not only that the employee suffered from MCS but that it was caused by her work at the hospital. Specifically, Dr. LaCava agreed that “to within [szc] a reasonable degree of medical certainty” it was his opinion that her condition was “caused by the exposures of multiple chemicals that she has been exposed to at the [hospital] during the course of her employment.” Dr. LaCava admitted during cross-examination that there is medical uncertainty as to the cause of MCS. He acknowledged that MCS may be caused by many factors including genetics, metabolism, physical stress, or toxic exposures, and that causation may differ depending on the *316person. The Appeals Court concluded that Dr. LaCava’s causation opinion was properly admitted because he was aware of the chemicals present at the hospital and diagnostic tests confirmed that the employee had been exposed to these chemicals. Cana-van’s Case, supra at 300-301. There is no suggestion that the judge conducted a Lanigan analysis to determine whether Dr. LaCava used a reliable methodology to conclude that the chemical exposures to which the employee was subjected caused her to suffer from MCS. This was error.
Because understanding medical causation is “beyond the . . . knowledge of the ordinary layman . . . proof of if it must rest upon expert medical testimony.” Hachadourian’s Case, 340 Mass. 81, 85 (1959). Opinions regarding medical causation are classic examples of testimony that must be subject to Lanigan analysis. Both the Daubert and Joiner cases involved the admissibility of medical causation testimony. In Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), the issue was whether there was reliable methodology to show that the drug Bendectin caused birth defects; in General Elec. Co. v. Joiner, supra, the Court addressed whether there was reliable methodology to show that polychlorinated biphenyls (PCBs) caused small cell lung cancer. Similarly, the causation question in this case is whether exposure to a variety of chemicals including ethylene, diesel fuel, and formaldehyde caused the employee to be “highly reactive to low levels of environmental chemicals.” Because the judge below did not conduct a Lanigan analysis to determine whether Dr. LaCava relied on a reliable methodology to determine that chemical exposures caused the employee’s MCS, it was an error to admit the opinion.9
The decision of the Industrial Accident Board is reversed.10
So ordered.
(concurring). I accept the court’s decision to adopt an abuse of discretion standard for analyzing Lanigan decisions by trial judges. See Commonwealth v. Lanigan, 419 Mass. 15 (1994). I understand the words “abuse of discretion” also to encompass an error of law. When I wrote the opinion in Commonwealth v. Vao Sok, 425 Mass. 787 (1997), the jury “was still out” on whether the standard for these decisions should be de novo or one of abuse of discretion-error of law. Since Vao Sok was decided, the United States Supreme Court in General Elec. Co. v. Joiner, 522 U.S. 136 (1997), and numerous other courts, have adopted an abuse of discretion-error of law standard. The clear weight of authority is now on the side of that test. I agree that we should follow the weight of authority on the issue, particularly because most evidentiary rulings by trial judges are governed by the abuse of discretion-error of law standard. Further, and perhaps more importantly, the application of the standard will make little difference in how an appellate court decides a hard science Lanigan-type case. The goal of Lanigan, as was the goal of the preceding tests, is to keep unreliable (or so-called “junk”) science from fact finders, thereby reducing the prospect of the return of verdicts or the rendition of decisions of dubious validity. Where new hard science is involved, an appellate court will always take a hard look at the trial judge’s decision to admit or exclude the evidence. And, as the cases illustrate, the appellate court will not hesitate to substitute its judgment for that of the trial judge, if the judge has erred in ruling on the evidence by finding it either reliable or unreliable. Thus, the Vao Sok case (reversing a motion judge’s improper exclusion of DNA evidence), and this case (reversing the improper admission by an administrative judge of expert medical testimony), would be decided the same under either a de novo or an abuse of discretion-error of law standard.
Finally, I agree with the court that the acceptance of expert testimony based on, among other things, personal observations or clinical experience, should, as a general proposition, be subject to a Lanigan analysis. I add two observations. First, in the absence of specific, concrete evidence suggesting unreliability, Lanigan should not be used to revisit areas where we have validated expert testimony based on properly conducted personal observations and clinical testing applying generally accepted scientific techniques. See generally Vassallo v. Baxter Healthcare Corp., 428 Mass. 1, 15 n.15 (1998). Second, I expect that *318Lanigan will have little application to expert testimony in the so-called “soft” sciences, such as psychology and sociology, which are highly dependent on information derived from such sources as personal observations, clinical assessments, and statistical data. It is here, more than anywhere else, that an appellate court will defer to a trial judge’s exercise of discretion, once the judge makes a decision as to the reliability of the process or theory underlying the proffered opinions and the relevance of the opinion to a matter in issue.