Background: Organ weights are often divided by body weight to report "relative" organ size. Yet ratios usually stay size-dependent and become misleading. We built a simple decision path for size adjustment and wrapped it in a Shiny application so physiologists can get correct answers without coding. Methods: We reused liver and body-weight data from a mouse nutrition study for confirmatory examples in two common cases: when diet groups shared a wide body-size range, and when diet produced much smaller animals with little overlap. We compared liver-to-body-weight ratios with size-adjusted regression when the body-size overlap allowed comparisons, and with causal Bayesian mediation when it did not. The full workflow was implemented in OrganScaleR, a guided R Shiny application. Results: The ratio-normalized organ weight is still associated with body weight, leading to misleading comparisons. Modeling liver weight against body weight gave more cautious, size-adjusted effects when groups shared a common size range. When diets shifted body size strongly and overlap was limited, causal mediation showed that most organ differences followed the change in body weight rather than an organ-specific action. Simulations confirmed that ratios can generate false positives and biased estimates under allometric scaling, while model-based approaches remained reliable. OrganScaleR implements this decision workflow in a guided Shiny application that returns interpretable effects. Conclusions: OrganScaleR selects scale, enforces common support, and routes the analysis to size-adjusted ANCOVA or causal Bayesian mediation depending on body-weight overlap. It reports adjusted effects in original units through a point-and-click workflow, removing the statistical barrier to abandoning ratio normalization.
Rebiffe, L., Gilquin, L., Leulier, F., De Vadder, F.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 5
- Comments 0