| pubmed-article:3296081 | pubmed:abstractText | Medical researchers and policy makers face decisions that require a choice from among two or more alternatives. Whereas traditional hypothesis tests cannot always serve the needs of the practitioner who needs to make a decision, a problem formulation that assigns losses to various incorrect decisions offers several advantages. With three possible decisions this approach offers a precise representation of the pragmatic and explanatory views of decision making. It enables the investigator to incorporate in the problem specification his attitudes about the seriousness of various errors by guiding him, before he sees the data, to a choice of asymmetric tail probabilities. It also suggests a reformulation of the P-value that can accommodate some of the difficulties practitioners face. | lld:pubmed |
