facts. Kumho Tire Co. v. Carmichael, 526 U.S. 137, 142 (1999). This inquiry is

highly “flexible.” Id. at 141.

In this matter, the district court considered both factors that Mr. Bulone

invokes. It simply determined that neither could overcome the many flaws it

identified in Dr. Zhang’s Exposure paper, a meta-analysis of the epidemiological

literature on glyphosate and NHL. Among the reasons given, the district court

explained that Dr. Zhang neglected studies postdating her analysis that tended to

undermine her results—studies Dr. Zhang admittedly would have included had she

2 See Luoping Zhang, et al., Exposure to glyphosate-based herbicides and risk for non-Hodgkin lymphoma: A meta-analysis and supporting evidence, 781 MUTATION RES. REV. IN MUTATION RES. 186 (2019) (Exposure); Iemaan Rana, et al. [including Dr. Zhang], Mapping the key characteristics of carcinogens for glyphosate and its formulations: A systematic review, 339 CHEMOSPHERE 139572 (2023).

3 24-2932 updated her analysis. Her paper also omitted large swaths of data from the Andreotti

(2018) study, which Dr. Zhang admitted was the highest quality study she examined.

The paper mixed data from studies that evaluated subjects’ exposure levels with

studies that did not, and combined studies that stratified data using “different

exposure cut-offs,” creating apples-to-oranges comparisons. And, of particular

concern to the district court, Dr. Zhang chose to include data from one study

(Eriksson (2008)) that failed to adjust for a rival cause of NHL—exposure to

pesticides other than glyphosate—despite the availability of adjusted data. The

paper then inaccurately implied that Dr. Zhang counted this study as one containing

adjusted data. In the district court’s view, the combination of these and other defects

rendered Dr. Zhang’s meta-analysis unreliable.

Mr. Bulone responds that certain (but not all) of these criticisms target mere

design limitations found in other scientific literature. Yet that does not rebut the

district court’s broader point that these flaws and limitations, taken together,

undermine the reliability of Dr. Zhang’s threadbare disclosures. On that point, we

cannot say the district court’s reasoning is “illogical, implausible, or without support

in inferences that may be drawn from the facts in the record.” United States v.

Redlightning, 624 F.3d 1090, 1110 (9th Cir. 2010).

Next, Mr. Bulone asserts that the district court improperly assumed a

factfinding role in the Daubert proceeding, but he is mistaken. This is not a case

4 24-2932 where the district court rejected the expert’s “ultimate conclusions” or doubted her

“credibility,” as in Elosu v. Middlefork Ranch Inc., 26 F.4th 1017, 1024–27 (9th Cir.

2022). Rather, the district court centered its analysis on the methods Dr. Zhang

employed and her application of those methods, just as Federal Rule of Evidence

702 contemplates.

Nor did the district court “create[] the problem[s] of which it complains,” as

Mr. Bulone asserts. As he concedes, district courts have discretion to question

putative experts in Daubert hearings. See United States v. Hankey, 203 F.3d 1160,

1168 (9th Cir. 2000) (affirming the admission of an expert where “the district court

conducted extensive voir dire” of expert’s testimony to assess its “reliability”). That

some of the deficiencies in Dr. Zhang’s methodology surfaced while the district

court exercised that discretion is immaterial. Dr. Zhang might not have expected the

district court to conduct such a searching inquiry into her methods, but that was

ultimately the district court’s prerogative. And given Daubert’s flexibility, we

cannot say the district court exceeded its discretion by doing so, especially in view

of the court’s near-decade of familiarity with the facts, the issues, and—

specifically—the scientific literature relevant to this Roundup litigation.

After excluding Dr. Zhang’s meta-analysis as unreliable, the district court

held that Monsanto was entitled to summary judgment because Mr. Bulone lacked a

cognizable expert-opinion on general causation. We affirm on this point as well.

5 24-2932 Mr. Bulone avers that the district court overlooked Dr. Zhang’s second paper—a

systematic review of the “mechanistic” evidence assessing the effect of glyphosate

exposure on characteristics of carcinogens. But that’s not the case. The district court

considered this paper, too, but found that it offered no “opinion about causation.”

We perceive no reversible error here either. On a previous occasion we had

to consider a case from the Roundup MDL, we affirmed the plaintiff’s admission of

three general-causation experts who opined on the relationship between glyphosate

and NHL. Hardeman v. Monsanto Co., 997 F.3d 941, 963–64 (9th Cir. 2021). But

there, we held that mechanistic evidence (or “cell studies”) could only help “show

causation so long as there is evidence of an association between glyphosate and NHL

in humans within the epidemiological literature.” Id. at 963. And yet, the only such

epidemiological evidence that Dr. Zhang offered was the meta-analysis, which, as

just discussed, the district court excluded. Under Hardeman’s reasoning, Dr.

Zhang’s mechanistic paper alone cannot support a cognizable inference of general

causation. See id. The district court was therefore justified in granting summary

judgment in Monsanto’s favor. See Engilis, 2025 WL 2315898, at *2.

AFFIRMED.

6 24-2932