Measuring object recognition ability: Reliability, validity, and the aggregate z-score approach

Area of Science:

• Cognitive psychology
• Neuroscience

Background:

• Accurate measurement of domain-general object recognition ability (o) is crucial but often requires extensive testing.
• Existing methods, like latent variable modeling, can be time-consuming and sample-intensive.

Purpose of the Study:

• To develop and validate an efficient and reliable measure of domain-general object recognition ability.
• To assess the test-retest reliability and stability of the proposed measure over time.

Main Methods:

• Utilized existing data to evaluate aggregate scores from object recognition tests as approximations of factor scores.
• Developed revised, time-efficient tests and an "odd one out" task with diverse stimuli.
• Administered revised tests to 163 participants on two occasions, one month apart, for reliability assessment.

Main Results:

• Aggregate scores from two object recognition tests correlated highly (r = .79 to .87) with factor scores from larger test batteries.
• The short aggregate measure demonstrated good test-retest reliability (r = .77).
• Structural equation modeling confirmed that the tests loaded onto a single latent variable, showing high stability (r = .93).

Conclusions:

• Score aggregation is an efficient method for estimating domain-general object recognition ability.
• The developed short measure is reliable and stable, facilitating future research on individual differences in object recognition.