Abstract: The invention provides improved methods for selecting items most likely to be the best items in a population of items. In accordance with certain preferred embodiments the probability that an item is among the best items in the population is determined by calculating the misclassification probability of each item and designating as the best items those with a misclassification index above a first threshold value from among the values thus determined and as the not best items those with a misclassification index below a second threshold value from among the values thus determined. In further preferred methods of the invention, the misclassification probabilities are determined by rankings of the items by respondents. In certain preferred embodiments the selection methodology of the invention is used to involve consumers in the screening stage of the idea generation process. The methodology is faster and more accurate in selecting the best items in a population than other methods.