Potential for bias in estimating human fecundability parameters: a comparison of statistical models.