Premium
Considering ERP difference scores as individual difference measures: Issues with subtraction and alternative approaches
Author(s) -
Meyer Alexandria,
Lerner Matthew D.,
De Los Reyes Andres,
Laird Robert D.,
Hajcak Greg
Publication year - 2017
Publication title -
psychophysiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.661
H-Index - 156
eISSN - 1469-8986
pISSN - 0048-5772
DOI - 10.1111/psyp.12664
Subject(s) - psychology , subtraction , psychophysiology , anxiety , bivariate analysis , error related negativity , interpretability , psychopathology , developmental psychology , significant difference , negativity effect , personality , cognitive psychology , clinical psychology , cognition , statistics , social psychology , anterior cingulate cortex , artificial intelligence , mathematics , neuroscience , arithmetic , psychiatry , computer science
There is growing interest in psychophysiological and neural correlates of psychopathology, personality, and other individual differences. Many studies correlate a criterion individual difference variable (e.g., anxiety) with a psychophysiological measurement derived by subtracting scores taken from two within‐subject conditions. These subtraction‐based difference scores are intended to increase specificity by isolating variability of interest. Using data on the error‐related negativity (ERN) and correct response negativity (CRN) in relation to generalized anxiety disorder (GAD), we highlight several conceptual and practical issues with subtraction‐based difference scores and propose alternative approaches based on regression. We show that ERN and CRN are highly correlated, and that the ΔERN (i.e., ERN − CRN) is correlated in opposite directions both with ERN and CRN. Bivariate analyses indicate that GAD is related to ΔERN and ERN, but not CRN. We first show that, by using residualized scores, GAD relates both to a larger ERN and smaller CRN. Moreover, by probing the interaction of ERN and CRN, we show that the relationship between GAD and ERN varies by CRN. These latter findings are not evident when using traditional subtraction‐based difference scores. We then completed follow‐up analyses that suggested that an increased P300 in anxious individuals gave rise to the apparent anxiety/CRN relationship observed. These findings have important conceptual implications for facilitating the interpretability of results from individual difference studies of psychophysiology.