Document Type

Pre-Print

Publication Date

7-16-2025

Subjects

discarding data, sufficient statistic, tournament, sequential trial

Abstract

It seems intuitively obvious that no part of a random sample should be discarded. However, for most students, the first formal justification of this idea is when sufficiency is introduced in an introductory statistical theory course. Students are often left with the idea that conditioning on the sufficient statistic always improves inference, even though the formal theory only says that for any statistical procedure there is one depending on the sufficient statistic that is no worse. Here we present an example on comparing two coins based on two consecutive tosses, where the optimal probability of guessing which coin has the higher heads probability is not improved by taking the second observation. We discuss some ramifications and introduce some generalizations, focussing on tournaments.

Rights

Copyright 2025 The Authors

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

Persistent Identifier

https://archives.pdx.edu/ds/psu/44001

Share

COinS