Modification of the triangle-taste-test model to allow for nonforced guessing

The modified triangle taste test proposed here allows for nonforced guessing. The maximum likelihood estimator (MLE) for the true proportion of distinguishers is derived. For the range of values examined, this estimator results in less bias and smaller variance than the MLE of the conventional (forced-guessing) model. For the modified (nonforced-guessing) model, an upper bound on the variance of the estimator is derived, along with an approximate 95% confidence interval for the proportion of distinguishers in the population.