Table 3.

Bootstrap method results evaluating regression model with outcome of vaginal delivery.

Number of variables removed	Variables removed	Percent of samples
0	None of variables were removed (full model contains six predictors: dilation, effacement, station, consistency, position and site)	5.5
1	Consistency	17.3
	Position	10.7
	Effacement	6.0
2	Consistency + Position	37.0
	Position + Effacement	14.6
	Consistency + Effacement	2.9
3	Consistency + Position + Effacement	6.0

Bootstrap samples (n = 1,000) of the same size as the original dataset were used. Logistic regression was performed with P<.01 significance level of the Wald chi-square for an effect to stay in the model in a backward elimination step using the resample.