This article is a powerful rebuttal to recent arguments minimizing the importance of statistical inference in science. It challenges claims by Hubbard et al., Tong, and Amrhein et al. that question the validity and usefulness of inferential statistics in fostering scientific insight. Through an extensive, evidence-based defense, the paper underscores statistical inference as the backbone of rigorous, reproducible, and meaningful scientific discovery.
The work spans disciplines—from physics and genetics to economics and environmental science—highlighting how inference informs experimental design, identifies causal patterns, and supports predictive modeling. The author presents a vivid thought experiment envisioning a world without statistical inference: one of scientific chaos, interpretive darkness, and irreproducible results. Without inference, causality collapses into anecdote, and uncertainty becomes paralyzing.
Across six branches of science—physical, biological, social, formal, applied, and interdisciplinary—the article details how statistical inference enables progress, from confirming the Higgs boson to modeling climate change and training AI systems. It directly addresses and dismantles criticisms, asserting that although statistical choices involve subjectivity, modern methodologies such as Bayesian models, bootstrapping, and machine learning accommodate this complexity.
The piece concludes with a call to preserve and promote statistical literacy. It affirms that inference is not just a tool but a conceptual pillar—vital for interpretation, generalization, and the cumulative progress of science.
For comprehensive discussion, field-by-field analysis, and responses to recent critiques, see the full encyclopedia article in the International Encyclopedia of Statistical Science.