OrganScaleR - scripts

Organ weights are often divided by body weight to report "relative" organ size. Yet ratios usually stay size-dependent and become misleading. Liver and body-weight data from a mouse nutrition study were used 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. Liver-to-body-weight ratios were compared to size-adjusted regression when the body-size overlap allowed comparisons, and with causal Bayesian mediation when it did not.

The analysis is decomposed in three parts:
* ratio vs ANCOVA vs log-log allometry model comparison based on liver and body-weight data
* ratio vs ANCOVA vs log-log allometry model comparison based on simulated data
* Bayesian causal mediation based on liver and body-weight data