testfit |

Carry out the Hosmer-Lemeshow and/or Andrews goodness-of-fit tests for estimated binary models.

Syntax

binary_equation.testfit(options)

Options

h | Group by the predicted values of the estimated equation. |

s=series_name | Group by the specified series. |

integer (default=10) | Specify the number of quantile groups in which to classify observations. |

u | Unbalanced grouping. Default is to randomize ties to balance the number of observations in each group. |

v | Group according to the values of the reference series. |

l=integer (default=100) | Limit the number of values to use for grouping. Should be used with the “v” option. |

prompt | Force the dialog to appear from within a program. |

p | Print the result of the test. |

Examples

equation eq1.binary work c age edu

eq1.testfit(h,5,u)

estimates a probit specification, and tests goodness-of-fit by comparing five unbalanced groups of actual data to those estimated by the model.

Cross-references

See
“Goodness-of-Fit Tests” for a discussion of the Andrews and Hosmer-Lemeshow tests.