Facial soft tissue thicknesses (FSTTs) hold an important role in craniofacial identification, forming the underlying quantitative basis of craniofacial superimposition and facial approximation methods. It is, therefore, important that patterns in FSTTs be correctly described and interpreted. In prior FSTT literature, small statistically significant differences have almost universally been overemphasized and misinterpreted to reflect sex and ancestry effects when they instead largely encode nuisance statistical noise. Here we examine FSTT data and give an overview of why P-values do not mean everything. Scientific inference, not mechanical evaluation of P, should be awarded higher priority and should form the basis of FSTT analysis. This hinges upon tempered consideration of many factors in addition to P, e.g., study design, sampling, measurement errors, repeatability, reproducibility, and effect size. While there are multiple lessons to be had, the underlying message is foundational: know enough statistics to avoid misinterpreting background noise for real biological effects.

No abstract

No abstract