Acceptability of an experimental intracervical device: results of a study controlling for selection bias.