President Donald Trump seeks to restore America’s “gold standard science”—a science that’s rooted in transparency, peer review, reproducibility, and freedom from political pressure. It’s time to apply that sort of science to climate.
“[W]here any type of ‘scientific evidence’ is used to support a government policy, a regulation, or legislation, the science behind it should … adhere to the elements outlined in President Trump’s May 23 Executive Order,” wrote environmental health expert Warren Kindzierski and statistician Stanley Young, a fellow of the American Statistical Association.
There’s no area that needs this more than that of the climate.
Climate policy has often relied on research that warrants stronger scrutiny and more rigorous validation. Agencies such as the Environmental Protection Agency sometimes rely on projections and models that cannot be tested or reproduced to justify sweeping regulations that affect everything from energy production to air quality standards.
Cooling the Climate Hysteria, edited by Kevin Dayaratna, dissects such unscientific projections. In a chapter written by Young and Kindzierski, they highlight the falsified link often drawn between climate change, ground-level ozone, and asthma.
According to Young and Kindzierski, ozone concentrations have declined significantly in recent decades, even as asthma rates have remained stable or increased. Thus, they conclude that the correlation between asthma and ozone concentrations is weak, if a correlation even exists at all.
But Young and Kindzierski highlight major inconsistencies in the EPA’s supporting evidence, an issue as important as the findings themselves. They write, “academic researchers today are rewarded in the university setting … for publishing research; but not for publishing ‘reproducible’ research.”
Specifically, they point to the EPA’s reliance on statistical associations of variables without the establishment of any plausible biological mechanisms. For example, its research simply identified a correlation between ozone levels and asthma outcomes, but didn’t account for other potentially significant variables like allergens, indoor air, quality, or socioeconomic factors.
When making statistical interpretations, it’s vitally important to establish biological plausibility. Without it, studies can mistake correlation for causation. Ignoring confounding variables can lead to an overstatement of health risks—and if such studies are used as the basis for regulation, the end result is regulatory action that imposes large economic costs for minimal or no actual health benefit.
Young and Kindzierski identify concerning discrepancies within EPA models. Many studies use proprietary or unpublished data, making it impossible for independent researchers to confirm the results under the same conditions. This lack of transparency prevents researchers from confirming that the EPA’s models are reproducible—a critical characteristic of the scientific method.
The EPA’s models also exhibit signs of multiple testing bias. The report found that for studies on the onset of asthma, the median number of possible hypotheses tested was about 13,824.
Assuming the tests were run at the 5 percent significance level and using the median number of hypotheses, the number of false positives (statistically significant results purely due to chance) would be approximately 691. This means the EPA’s studies could contain hundreds of false positives, results that may be nothing more than mathematical noise mistaken for meaningful science.
Without correcting for this statistical bias, these findings aren’t worthy of being considered in policy creation.
Environmental studies display one of the highest levels of effect size inflation in modern scientific journals. A 2023 meta-analysis found that nearly 30 percent of the “effects” reported in environmental science meta-analyses may be artifacts of selective publication, or publication bias.
That’s the phenomenon in which studies that produce positive or dramatic results are more frequently published, while studies with negative or inconclusive findings are often ignored or withheld. This creates a distorted picture of reality in which only “headline-worthy” science makes it into the literature, regardless of its accuracy.
This pattern feeds into the very problem the new executive order aims to fix. Policies based on inflated risks and non-reproducible studies have consequences that are felt by everyone in the U.S. Research used to shape energy policy needs a low margin for error and a higher standard of proof.
Young and Kindzierski’s analysis highlighted the discrepancies within current federal research, but their work also points to something deeper: a need for humility in science and restraint in policymaking. When science is used to justify sweeping authority, it must answer to more than the narrative—it must answer to the truth. The future of climate policy should depend not on louder claims but instead on sound evidence.