Significance testing for the coefficients of Quasi binomial or the quasi Poisson regression {Rfast}

Significance testing for the coefficients of Quasi binomial or the quasi Poisson regression

Description

Significance testing for the coefficients of Quasi binomial or the quasi Poisson regression.

Usage

anova_propreg(mod, poia = NULL)
anova_qpois.reg(mod, poia = NULL)

Arguments

mod

An object as returned by the "prop.reg" or the "qpois.reg" function.

poia

If you want to test the significance of a single coefficient this must be a number. In this case, the "prop.reg" or the "qpois.reg" function contains this information. If you want more coefficients to be testes simultaneously, e.g. for a categorical predictor, then this must contain the positions of the coefficients. If you want to see if all coefficients are zero, like an overall F-test, leave this NULL.

Details

Even though the name of this function starts with anova it is not an ANOVA type significance testing, but a Wald type.

Value

A vector with three elements, the test statistic value, its associated p-value and the relevant degrees of freedom.

Author(s)

Michail Tsagris

R implementation and documentation: Michail Tsagris <[e-mail]> and Manos Papadakis <[e-mail]>.

References

Papke L. E. & Wooldridge J. (1996). Econometric methods for fractional response variables with an application to 401(K) plan participation rates. Journal of Applied Econometrics, 11(6): 619-632.

McCullagh, Peter, and John A. Nelder. Generalized linear models. CRC press, USA, 2nd edition, 1989.

See Also

prop.reg, qpois.reg, univglms, score.glms, logistic_only

Examples

y <- rbeta(1000, 1, 4)
x <- matrix(rnorm(1000 * 3), ncol = 3)
a <- prop.reg(y, x)
## all coefficients are tested
res<-anova_propreg(a)  
## the first predictor variable is tested 
res<-anova_propreg(a, 2)  
a  ## this information is already included in the model output
## the first and the second predictor variables are tested 
res<-anova_propreg(a, 2:3)