6.8.2.2.3. statsmodels.sandbox.stats.contrast_tools.TwoWay

class statsmodels.sandbox.stats.contrast_tools.TwoWay(endog, factor1, factor2, varnames=None)[source]

a wrapper class for two way anova type of analysis with OLS

currently mainly to bring things together

Notes

unclear: adding multiple test might assume block design or orthogonality

This estimates the full dummy version with OLS. The drop first dummy representation can be recovered through the transform method.

TODO: add more methods, tests, pairwise, multiple, marginal effects try out what can be added for userfriendly access.

missing: ANOVA table

__init__(endog, factor1, factor2, varnames=None)[source]

6.8.2.2.3.1. Methods

__init__(endog, factor1, factor2[, varnames])
ftest_interaction() ttests for no-interaction terms are zero
r_nointer() contrast/restriction matrix for no interaction
summary_coeff()
ttest_conditional_effect(factorind)
ttest_interaction() ttests for no-interaction terms are zero