Proof paradoxes refer to situations where naked statistical evidence supports the conviction of a defendant, yet the resulting judgment appears counterintuitive when considered solely from this evidence. The prevailing approach to addressing proof paradoxes involves distinguishing naked statistical evidence from other types of evidence. In this framework, Thomson (1986) proposes that the existence of a causal relationship between the evidence and its source can serve as a criterion for this distinction. Conversely, Redmayne (2008) contends that Thomson's proposal is unhelpful, arguing that even with naked statistical evidence, a causal relationship can be established in accordance with Thomson's view. In this study, we demonstrate that, by leveraging Pollock's nomic probability theory and Woodward's causal model, Thomson's proposed criterion can be refined to effectively address Redmayne's criticisms.
Keyword: Pollock's Nomic probability, legal standards of proof, Proof Paradoxes, Woodward's causal model, Thomson, Redmayne, Statistical evidence.

Introduction:
The proof paradoxes refer to situations where the foundational claim supporting a verdict relies solely ‎on naked statistical evidence. These examples are constructed so that the decision-maker’s degree of ‎belief exceeds the probabilistic thresholds set by relevant standards of proof, yet many intuitively feel ‎that this is insufficient to justify a conviction. This raises a critical question: what explains this intuitive ‎preference for other forms of evidence? Two main theoretical frameworks address this: the ‎‎“revisionary” approach, which argues that such intuitions are unwarranted and should be reconsidered, ‎and the “non-revisionary” stance, which seeks to justify the distinction between valid evidentiary proof ‎and naked statistical evidence through concepts such as "defendant-dependency," "knowledge," and ‎‎"normic support.
Materials & Methods:
This research employs a critical analytical method to evaluate the debate between Thomson and ‎Redmayne. The methodology consists of three steps:
‎1. Clarification: Analyzing Thomson’s criterion for valid evidence (causal connection) and Redmayne’s ‎subsequent critiques regarding the ambiguity of causality.
‎2. Deconstruction: Identifying the core assumptions of Redmayne’s critique, specifically the validity of ‎inferring accident rates from ownership rates and the existence of a causal relationship between ‎ownership and liability.
‎3. Theoretical Integration: Applying Pollock’s nomic probability framework to challenge epistemic ‎justifications of statistical inference and utilizing Woodward’s causal model (specifically the ‎reproducibility criterion) to provide a precise definition of a "causal link".

Discussion & Results:
The analysis demonstrates that Redmayne's critique of Thomson is well-founded regarding the lack of ‎precision in the definition of causality. However, by applying Pollock’s framework, this study reveals ‎that inferring accident rates solely from ownership data fails to satisfy the criteria of nomic probability, ‎meaning such inferences are not epistemically justified. Furthermore, by integrating Woodward’s causal ‎model, a clear distinction is established: valid evidence is distinguished from naked statistical evidence by ‎its ability to satisfy the reproducibility criterion. This explains why convictions based on individualized ‎evidence are intuitively superior to those based on statistical data.
Conclusion:
The primary aim of this research—to refine Thomson’s criterion to address Redmayne’s objections—is ‎achieved by transitioning from a vague notion of causality to a rigorous model based on Woodward’s ‎reproducibility. The study concludes that the intuitive rejection of naked statistical evidence is grounded ‎in the absence of a genuine causal link between the statistical data and the specific liability of the ‎defendant. This refinement provides a robust epistemological basis for distinguishing between a naked ‎statistical evidence from other forms of legally valid evidence. ‎