Statistical inference in experimental neuroscience is routinely detached from the experimental record: analytic choices are reported in prose summaries that do not expose the code, assumptions, or decision pathways that produced the results. This detachment limits reproducibility and impairs peer review. Here, we describe the Companion Statistical Report (CSR), a structured, versioned document format designed to accompany empirical neuroscience manuscripts as peer-reviewed as a peer-reviewed resource. The CSR integrates data provenance, preprocessing decisions, exploratory analyses, model specifications, assumption diagnostics, inference with effect sizes, and sensitivity analyses into a single executable document, authored in Quarto 1.8 and supporting both R and Python workflows. We provide an open template hosted at on GitHub that implements this format with institutional branding, parameterization, and version tracking. The template was developed by the Bertrand Russell Research Excellence Group (NEC) at the School of Medicine, Rio de Janeiro State University. By making analytic choices auditable and reproducible by design, CSRs are designed to reduce the gap between what neuroscience experiments measure and what published statistics claim, offering a tractable and immediately implementable step toward greater transparency.
Ramos et al. (Wed,) studied this question.