A more robust t-test

The Working Group on Risk - CREAR, with the support of the IDS dpt , Institut des Actuaires, LabEx MME-DII and the group BFA-SFdS, has the pleasure to invite you to the seminar by:

Prof. Ulrich K. MÜLLER, Princeton University, USA

“A more robust t-test”

Wednesday December 7, 1:00 pm (CET)

Dual format:
ESSEC Paris La Défense (CNIT), Room 237, and via Zoom, please click here (Password/Code : WGRisk)

Topics:
This paper combines extreme value theory for the smallest and largest k observations for some given k>1 with a normal approximation for the average of the remaining observations to construct a more robust alternative to the usual t-test. The new test is found to control size much more successfully in small samples compared to existing methods. This holds for the canonical inference for the mean problem based on an i.i.d. sample, but also when comparing two population means and when conducting inference about linear regression coefficients with clustered standard errors.