Considering ERP difference scores as individual difference measures: Issues with subtraction and alternative
Alexandria Meyer1, Matthew D Lerner1, Andres De Los Reyes2
1Department of Psychology, Stony Brook University, Stony Brook, New York, USA.
Psychophysiology
|December 22, 2016
Summary
Subtraction-based difference scores in psychophysiology research, like the error-related negativity (ERN) minus correct response negativity (CRN), can obscure findings. Regression-based methods offer a clearer view of individual differences in psychopathology, such as generalized anxiety disorder (GAD).
Area of Science:
- Cognitive Neuroscience
- Psychophysiology
- Clinical Psychology
Background:
- Growing interest in psychophysiological and neural correlates of psychopathology, personality, and individual differences.
- Prevalence of subtraction-based difference scores in correlating psychophysiological measures with individual difference variables.
- Conceptual and practical issues associated with subtraction-based difference scores.
Purpose of the Study:
- To highlight issues with subtraction-based difference scores using error-related negativity (ERN) and correct response negativity (CRN) in relation to generalized anxiety disorder (GAD).
- To propose and demonstrate alternative regression-based approaches for analyzing psychophysiological data.
- To improve the interpretability of results in individual difference studies of psychophysiology.
Main Methods:
- Analysis of error-related negativity (ERN) and correct response negativity (CRN) data in individuals with generalized anxiety disorder (GAD).
- Comparison of traditional subtraction-based difference scores (ΔERN = ERN - CRN) with regression-based methods (residualized scores and interaction analyses).
- Inclusion of P300 amplitude in follow-up analyses to explain observed relationships.
Main Results:
- ERN and CRN are highly correlated, and ΔERN shows opposing correlations with ERN and CRN.
- Bivariate analyses suggest GAD relates to ΔERN and ERN, but not CRN.
- Regression-based analyses reveal GAD relates to both larger ERN and smaller CRN, with the ERN-GAD relationship varying by CRN.
- An increased P300 in anxious individuals may explain the apparent anxiety/CRN relationship found with simpler methods.
Conclusions:
- Subtraction-based difference scores can mask or distort relationships between psychophysiological measures and individual differences like GAD.
- Regression-based approaches, including residualized scores and interaction analyses, provide a more nuanced and accurate understanding of these relationships.
- Findings underscore the importance of employing appropriate statistical methods for accurate interpretation in psychophysiological research on individual differences and psychopathology.
