Purchase this article with an account.
S.B. Flynn, J. Milligan, J.B. Pittenger, T. Realini, T.B. Redens; Effect of Feedback on Cup–Disc Ratio Estimation . Invest. Ophthalmol. Vis. Sci. 2005;46(13):2493.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: To assess the impact of performance feedback on estimates of schematic cup–to–disc ratio (CDR) estimation. Methods: 54 medical students participated in one of two separate testing groups (A and B). Participants were presented 5 schematic stimuli, each composed of 2 concentric circles simulating the optic cup and disc. The outer circle’s size was kept constant, while the inner circle’s size varied across stimuli to produce inner–outer vertical ratios ranging from 0.31 to 0.67. For each stimulus, participants estimated the vertical CDR. After evaluating the 5 images, participants were given feedback: the images were re–presented, each with a numeric CDR. In both groups, the feedback was intentionally deceptive. In Group A, the feedback ratios derived from a regression slope that was too steep (slope = 1.5); i.e., for images with an actual CDR less than 0.50, the feedback ratio was too low, while for actual CDR greater than 0.50, it was too high. In Group B, the feedback ratios derived from a regression slope that was too flat (slope = 0.67); thus, for images with an actual CDR less than 0.50, the feedback ratio was too high, while for actual CDR greater than 0.50, it was too low. For Group A, the feedback ratios ranged from 0.26 to 0.74; for Group B, from 0.39 to 0.61. After receiving feedback, participants viewed a different set of stimuli and were asked to estimate VCDR for these. Results: For each participant, linear regression was performed separately with their pre– and post–feedback CDR estimates. The slopes of these regression functions were used in further analyses. The pre–feedback mean slopes were: Group A = 1.3, Group B = 1.41; these did not differ significantly (t = 1.34, p>.05). The post–feedback mean slopes were: Group A = 1.18, Group B = 1.04; these differed significantly (t = 4.04, p<.05). For each participant, the pre– to post–feedback change in slope was determined. The median decrease–in–slope for Group A = 0.16, and for Group B = 0.37; these differed significantly (Mann–Whitney Test, p<.003). Conclusions: Participants in both groups demonstrated a drop in regression slope after receiving misleading feedback. Participants in Group B had a greater drop; this is consistent with the flatter function from which their misleading feedback was derived. These results suggest that observers can be taught to interpret CDR inaccurately, which may have implications regarding resident education by teachers who estimate CDR poorly. Further study is warranted and is underway.
This PDF is available to Subscribers Only