Table 5.
Method | Effect of mismeasured onset age |
---|---|
Correlation Analysis | • Pearson correlation is biased downward. |
Linear Regression | • Biased estimated effects of subject-specific features (covariates) if the measurement error in age-at-onset is correlated with the features. |
Logistic Regression | • Highly sensitive to response error (i.e., misclassification). Ignoring misclassification leads to estimated model parameters being severely biased and highly variable. |
Survival Models | • Evidence of bias but exact effects need to be explored, especially for genetic mixture models where genetic information is not always available. |