Purchase this article with an account.
Steven M Silverstein, Brian P Keane, Deanna M Barch, Cameron S Carter, James M Gold, Ilona Kovács, Angus MacDonald, III, J. Daniel Ragland, Milton E Strauss; Test-Retest Reliability of a Contour Integration Test in Samples of Healthy Control and Schizophrenia Subjects. Journal of Vision 2012;12(9):891. doi: 10.1167/12.9.891.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Introduction: Contour integration tests have clarified mechanisms of perceptual organization in humans and monkeys, in children and the elderly, and in disorders such as amblyopia and schizophrenia. However, the test-retest reliability of the contour integration paradigm is unexplored. Establishing this aspect of the paradigm is useful for assessing the effects of aging, pathology, or response to treatment. Methods: Participants with schizophrenia and healthy age-matched controls completed 3 contour integration assessments, 7 and then 14 days apart. On each trial, subjects judged whether a subset of otherwise randomly oriented Gabor elements formed a leftward or rightward pointing shape. Task difficulty depended on the amount of orientation jitter added to the target elements: 0°, 7°, 9°, 11°, 13°, 15°. The number of controls and patients for the three sessions were: 1) 122, 93; 2) 120, 93; 3) 118, 93. Test-retest reliability was calculated for accuracy and fitted-Weibull threshold estimates, using both intraclass correlation (ICC) and Spearman rho methods. Results: ICC estimates for accuracy varied with jitter condition. For conditions in which controls could reliably perceive contours, the ICC values for accuracy scores ranged from .62-.80 across Times 1-2, and from .56-.86 across Times 2-3. For schizophrenia patients, value ranges were .59-.86 and .71-.87, respectively. ICC values of threshold estimates were .56 and .46 for controls, and .28 and .64 for patients. Spearman rho values for threshold estimates for controls were .46 and .50 for Times 1-2 and 2-3, and .47 and .49 for patients. Additional analyses examined reliability across Times 1-3, and differences when subjects with poor acuity were excluded. Conclusion: This version of the contour integration test has fair-very good test-retest reliability. Reliability varied as a function of subject group (patient vs. control), calculation method (ICC vs. correlation), and unit of analysis (accuracy level vs. threshold estimate).
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only