validate(regularity.lrm,bw=T,B=200)
Irregular verbs also tends to be more frequent than regular verbs, and it is reasonable to assume that this high frequnecy protects irregular verbs through time against regularization.
"Each succesive row in a sequential ANOVA table evalues whether adding a new predicter is justifiled given the other predictors in the preceding rows. By contrasts, the summary() function evalutes whether the coefficients are signficantly diferrent from zero in a model containing all other predictors." .
"The second function of anova() is to allow us to evalute the overall significance of factors. When a factor has only two levels, the test for the (single) coefficient based on the Z-score in summary() is very similar to the test in the anova() function when relevant factor is entered last into the model equatation. But when a factor has more than two levels, the table of coefficients in summary() lists a t-values or a Z-score for each coefficients. In order to assess whether the factor as a whole is explanatory, the anova() table is essential"
data[abs(rstandard(data.glm )) < 5, ]