Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 249 - Multivariate Methods for Neuroimaging Data
Type: Contributed
Date/Time: Tuesday, August 9, 2022 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistics in Imaging
Abstract #322336
Title: Conditional Correlation Models with Association Size
Author(s): Danni Tu* and Bridget Mahony and Tyler M. Moore and Maxwell A. Bertolero and Aaron F. Alexander-Bloch and Ruben Gur and Danielle S. Bassett and Theodore D. Satterthwaite and Armin Raznahan and Russell Shinohara
Companies: University of Pennsylvania and National Institutes of Mental Health and University of Pennsylvania and University of Pennsylvania and University of Pennsylvania and University of Pennsylvania and University of Pennsylvania and University of Pennsylvania and National Institutes of Mental Health and University of Pennsylvania
Keywords: correlation; association; regression; effect size
Abstract:

In tests of cognitive and physical performance, the trade-off between speed and accuracy requires that the two be studied together. A natural question is whether speed-accuracy coupling depends on other variables, such as sustained attention. We propose CoCoA (Conditional Correlation Model with Association Size), a likelihood-based statistical framework to estimate the conditional correlation between speed and accuracy as a function of additional variables. We also propose novel measures of the association size, which are analogous to effect sizes on the correlation scale, while adjusting for confounding. In simulation studies, we identify and compare likelihood-based estimators of conditional correlation to semi-parametric estimators adapted from genome association studies, and find that the former fare better under both ideal settings and misspecification. Using neurocognitive data from the Philadelphia Neurodevelopmental Cohort, we demonstrate that greater sustained attention is associated with stronger speed-accuracy coupling in a complex reasoning task while controlling for age.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2022 program