What's new in version 1.2.4

Version 1.2.4 strengthens the longitudinal workflow with a treatment-by-time example, conditionally IID residual errors, pre-specified visit contrasts, small-sample inference checks, and random-slope and categorical-visit sensitivity analyses. The underlying fitting engines are unchanged; the emphasis of this patch is independently validated examples, regression protection, documentation, and reproducibility.

Basic Examples (1) 

Fit repeated measurements with a patient random intercept using ordinary Wolfram expressions. The synthetic patientData is a smooth deterministic function of time and patient index, so the output is exactly reproducible; the fitted variance components quantify deviation from a linear-in-time model rather than measurement noise:

In[1]:=

Out[1]=

Nested and crossed designs (2) 

A list of random specifications adds grouping factors; NestedGrouping makes a clinic/patient hierarchy explicit:

In[2]:=

Out[2]=

Crossed subject and item factors use one random specification per grouping variable:

In[3]:=

Out[3]=

Generalized and nonlinear models (2) 

Use an explicit response family for events or counts:

In[4]:=

Out[4]=

Or give a symbolic nonlinear mean expression and starting values:

In[5]:=

Out[5]=

Small-sample inference (1) 

Satterthwaite degrees of freedom for the repeated-measures model:

In[6]:=

Out[6]=

Inspect fit["ConvergenceReport"], fit["SingularFit"], fit["Warnings"], the grouping levels, residual structure, and prediction target before interpreting a result.