Expert Admissibility in State Courts
California never adopted Daubert, and the reason traces to a voiceprint. People v. Kelly, 17 Cal. 3d 24 (1976), reversed an extortion conviction built on spectrographic voice identification after finding the prosecution had proven general scientific acceptance of the technique through a single expert too closely identified with the technique's own development rather than through evidence of what the broader relevant field actually believed. The rule the court adopted that day, borrowed from the federal Frye line and known ever since in California as Kelly-Frye, asks whether a new scientific technique has gained general acceptance in its own field, a question the court assigns to the scientific community itself rather than to the trial judge's own independent assessment of the underlying science.
That rule still governs any genuinely novel scientific technique offered in a California courtroom, and nothing the state's supreme court has done since has touched it directly. What changed is what happens to the much larger category of expert testimony that never involves a novel technique at all, an accountant's lost profits calculation, an engineer's opinion about a defective part, a real estate appraiser's method for valuing an unusual parcel. Sargon Enterprises, Inc. v. University of Southern California, 55 Cal. 4th 747 (2012), gave trial judges an affirmative gatekeeping duty over that whole remaining category, grounded not in Frye but in Evidence Code sections 801 and 802, to exclude opinion testimony that rests on speculation, on assumptions the record does not support, or on a methodology no reasonable expert would use. A unanimous bench cited the federal Daubert trilogy approvingly along the way, grounding the duty in California's own evidentiary text rather than adopting the federal test outright, and was careful to add that nothing in the opinion touched Kelly's own holding on novel scientific techniques, which continues to require general acceptance regardless of what Sargon asks of everything else.
The seam between the two tracks is where the harder litigation now sits, because Sargon's own opinion was careful to say the gatekeeper's focus belongs on an expert's principles and methodology, not on the conclusions those principles produce, which is precisely the line Kelly-Frye also polices for novel techniques. An expert whose underlying method is well established but whose application of it to a particular set of facts is aggressive or thin does not obviously trigger either doctrine's most searching version of review, and California's own courts have not drawn a bright boundary between a Sargon-level speculation problem, which a judge can resolve before trial, and a garden variety weight-of-the-evidence dispute that state law still leaves for the jury to sort out.
North Carolina drew that boundary differently, and did so on purpose. Its legislature amended Rule 702(a) of the state's evidence code in 2011 to track the language of the federal rule as it existed after Daubert, and the state supreme court confirmed what the amendment meant in State v. McGrady, 368 N.C. 880 (2016), holding that the amended rule incorporates the federal reliability standard rather than the state's own older general-acceptance test. The court applied that standard to affirm the exclusion of a law enforcement trainer's testimony about the unconscious psychological responses of someone using force in self-defense, reasoning that whatever expertise the witness held in training officers did not establish that his opinions about a defendant's own mental state rested on a methodology courts could treat as reliable. Unlike Sargon, McGrady's gatekeeping reaches the fit between a method and the facts it is applied to, not only the method standing on its own.
Pennsylvania kept Frye and drew the line in the opposite direction, and it took a fractured opinion to say so plainly. Walsh v. BASF Corp., 234 A.3d 446 (Pa. 2020), arose from a forty year groundskeeper's claim that decades of pesticide exposure caused the acute myelogenous leukemia that killed him, and a trial court that excluded his expert's testimony not because the underlying epidemiological methods were novel or unaccepted, but because the judge concluded the studies those methods drew from did not actually support the expert's own causation opinion. A four to three majority held that the trial court had exceeded a Frye inquiry's proper scope by doing that. Frye asks only whether the relevant scientific community generally accepts the methodology an expert used, the majority held; whether the studies that methodology produced are strong enough to support the conclusion drawn from them is a question about the weight of the evidence, and weight belongs to the jury once the methodology itself clears the gate. A report resting on a generally accepted epidemiological method but a debatable causation conclusion clears that gate and goes to a Pennsylvania jury; the identical report, offered against North Carolina's post-2011 rule, faces a judge who may ask further whether the method was reliably applied to reach this particular conclusion, a question Walsh's own majority held Pennsylvania law forbids a trial judge from asking at all.
None of this shows which regime actually gets the underlying science right more often, or protects a jury from bad expert testimony more effectively than the alternative does. It also does not show that any one state's judges are applying their own chosen standard incorrectly; every court discussed here was reading its own governing text the way that text told it to be read. Whether Pennsylvania's jury-trusting model or North Carolina's judge-gatekeeping model produces fewer wrongful verdicts, or lets more bad science reach a jury, is an empirical question about outcomes, not a question a comparison of doctrinal text can answer, and this account of the split does not attempt to answer it.