Abstract

The Forced-Choice Preferential Looking technique was used to obtain visual acuity data for 20 two-month-old infants. Two observers each collected 500 trials (50 trials at each of 10 stimulus levels) on 10 infants. Probit analysis¹ was used to estimate the 75% threshold and standard error of acuity and the standard deviation of the best-fitting cumulative normal curve. We used this unusually large data set to study the effects of changing stimulus parameters (number, placement, and spacing of stimuli) and varying the number of trials per stimulus. In addition, we estimated variability across infants and observers and test-retest reliability. Finally, we used the results of these analyses to create experimental designs that optimize the accuracy of an acuity estimate while minimizing its variability, given the small number of trials usually obtainable on human infants.

© 1986 Optical Society of America