Table 2.

Advantages and disadvantages of different regression methods for modeling age of motor-onset.

Method	Advantages	Disadvantages
Correlation Analysis	• Simple and available in standard software. • Assesses strength of linear relationship between two variables (e.g., age-at-onset and subject-specific features).	• Missing observations are dropped from the calculation. • Results can be perturbed by outliers. • Does not reveal any nonlinear relationship between two variables.
Linear Regression	• Simple and available in standard software. • Quantifies the linear effect of subject-specific features (covariates) on age-at-onset (response). • Assesses how much of the variability of age-atonset is explained by the features (i.e., coefficient of determination).	• Missing observations are dropped from the analysis. • Results can be perturbed by outliers.
Logistic Regression	• Simple and available in standard software. • Assesses the linear effect that subject-specific features have on the log odds of onset occurring. • Extracts information from individuals who have experienced onset and from those who have not by incorporating a binary response variable in the model.	• Ignores time differences between subjects experiencing onset; that is, subjects who experience onset at different times are treated similarly. • Loss of time information can be remedied with multiple binary response variables but can result in numerically unstable estimation.