When scoring algorithms matter: effects of working memory load on different IAT scores

Area of Science:

• Cognitive Psychology
• Psychometric Measurement

Background:

• The Implicit Association Test (IAT) is widely used to measure implicit biases.
• Process models assume varying cognitive demands (working memory, executive functions) for compatible and incompatible IAT blocks.

Purpose of the Study:

• To investigate the impact of experimentally manipulated cognitive load on IAT performance and scoring.
• To examine how different IAT scoring algorithms are affected by increased cognitive load.

Main Methods:

• A secondary task (random number generation) was used to manipulate cognitive load in 120 participants completing an IAT.
• Performance metrics (latency, errors) and IAT scores (conventional, D-scores, log-transformed) were analyzed.

Main Results:

• Increased cognitive load resulted in slower reaction times and more errors, particularly in incompatible IAT blocks.
• Raw performance-based IAT scores (latency, errors) increased under load, while transformed scores (log-latency, D-scores) decreased.
• External correlations of IAT scores with other measures were not significantly altered by the cognitive load manipulation.

Conclusions:

• Cognitive load differentially affects raw versus transformed IAT scores, highlighting the importance of scoring algorithm choice.
• The dissociation suggests that different scoring methods capture distinct aspects of IAT performance under varying cognitive demands.
• Despite scoring discrepancies, the external validity of the IAT appears robust to cognitive load manipulations.