PCM3 Model Design And Phylogenetic Regression

PCM3 is about model structure, not only model output. The lecture walks through how one primate response changes interpretation as the regression becomes more realistic.

flowchart LR
    A[Baseline GLS] --> B[Response transformation]
    B --> C[Categorical coding]
    C --> D[Interaction term]
    D --> E[Phylogenetic covariance]

Main Analytical Steps

• fit a baseline GLS with one continuous and one categorical predictor
• transform the response to improve residual behavior
• remove the intercept when group-specific estimates are the real target
• add an interaction term to ask whether one slope depends on the categorical grouping
• standardize the continuous predictor to change the intercept meaning
• refit the model with a phylogenetic covariance structure through corMartins

What Makes This Study Valuable

PCM3 forces the repository to be explicit about:

• how categorical levels are encoded
• which level is treated as the reference group
• how interaction coefficients should be read
• whether taxa were excluded before or after phylogenetic correction
• how diagnostic improvement is separated from biological interpretation

What Readers Should Watch Closely

• whether the intercept is still biologically meaningful after transformation or standardization
• whether categorical coefficients are being read against the correct reference level
• whether the interaction term changes the meaning of the main effects rather than merely adding one more coefficient
• whether the move into corMartins changes the scientific interpretation or only the residual structure

Current Repository Support

Nearby runtime surfaces already exist for categorical contrast reporting, interaction-coefficient ledgers, comparative model-selection ledgers, and residual review after phylogenetic regression.

Those surfaces make the study reviewable, but they do not yet turn PCM3 into a fully closed lecture-equivalent parity suite.