Karlson-Holm-Breen method for comparing probit coefficients

Significance test for confounding; that is, the difference between regression coefficients from same-sample nested logit and probit models. The test procedure follows Karlson et al (2012), Section 3.4.

khb(X, y, z)

Arguments

X	data frame comprising independent variables including confounding variable.
y	vector of dependent variable.
z	character string giving the name of the confounding variable in X.

References

Karlson, K.B., A. Holm and R. Breen (2012). Comparing regression coefficients between same-sample nested models using logit and probit: A new method. Sociological Methodology, 42(1):286--313.

Examples

## 1. load results from Klein (2015a)
 data(klein15a)

## 2. apply KHB method
with(klein15a$variables, khb(X=X, y=Y, z="eta"))
#> Warning: glm.fit: fitted probabilities numerically 0 or 1 occurred
#> Warning: glm.fit: fitted probabilities numerically 0 or 1 occurred
#> 
#> Karlson-Holm-Breen method
#> Null hypothesis: Change in coefficient is not attributable to confounding by z.
#> 
#>                  p.value
#> pi.inv            0.5581
#> wst.ieq           0.0480
#> loan_size.add     0.9034
#> loan_size2.add    0.0442
#> lngroup_agei.add  0.0412
#> `0`               0.9535
#> `1`               0.5665
#> `2`               0.2517
#> `3`               0.4388
#> `4`               0.3487
#> `5`               0.5749
#> `6`               0.7760
#> `7`               0.3651
#> `8`               0.3345
#> `9`               0.6832
#> `10`              0.4322
#> `11`              0.4884
#> `12`              0.6036
#> `13`              0.6391