{"id":"arxiv/2606.07847","url":"https://arxiv.org/abs/2606.07847","tags":["math.ST"],"score":0,"title":"Revisiting the Behrens-Fisher Problem: Validity-First Optimality","domain":"math.ST","source":"arxiv","authors":["Xiao Wang","Chuanhai Liu"],"reviews":0,"subject":"math.ST","summary":"The Behrens--Fisher problem concerns inference on the difference of two normal means when both variances are unknown and unequal. It is a classical example in which nuisance parameters prevent ordinary exact fixed-sample inference, and it has long served as a benchmark for the foundations of inference. We revisit it through the inferential model (IM) framework of Martin and Liu. After conditioning and regular marginalization, the exact association is two-dimensional, with one coordinate for the standardized mean contrast and one for the variance ratio. Their one-dimensional generalized…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-06-05","reactions":{}}