Compare two logistic/probit regression coefficient using different methods.

Description

Compare two logistic/probit regression coefficient using different methods.

Usage

1
2
3
coefChange(reduced, full, method = "ystand", ratio.type = "perc", R = 1)

compareModels(reduced, full, method = "ystand", ratio.type = "perc")

Arguments

R

Integer. If larger than 1, confidence interval of the change are computed using bootstrap.

reduced

The reduced model.

full

The full model

method

Character. The method used to compare coefficients (see details).

ratio.type

Character specifying how to compare the coefficients. This should be one of "perc" (percentage change between the two models), "ratio" (coefficient of the reduced model divided the coefficient of the full one) or "diff" (difference between the two models).

Details

The compareModels function compares the coefficients of two logistic/probit regressions using different methods.

  • If method="naive", the coefficients are compared without any transformation. This method is mainly provided to highlight the effect of other methods.

  • If method="ystand", y-standardization is used to compare the coefficients.

  • If method="ape", Coefficients change are computed using average partial effects.

  • Finally, if method="khb", the KHB method is used to compare the coefficients accross models.

The coefChange function only compute the change of coefficients using either the percentage of change or the ratio of change. It can also compute confidence intervals of this change using bootstrapping.

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.