The effect of misclassification on sample size: one-sample proportion test