Purpose: To evaluate the reliability of measuring CS with uncalibrated iPads.
Methods: Six random iPads with retina display were calibrated with a colorimeter and the correlation between Luminance (L) and pixel level (y) was computed according to an exponential function. The mean and confidence interval (±2SD) obtained from the six iPads were calculated and the bit-stealing technique was applied for expanding y from 256 to 2540 possible values. The L of the optotype was computed for the selected contrast values (logC) represented in log units, using 0.1 log and 0.05 log steps. At each particular y, the contrast was considered reliable when the mean L plus 2SD was less than half the difference of luminance between two consecutive levels of contrast. Differences between the iPads for the Experimental logC were evaluated with the Friedman test.
Results: Luminance properties vary between devices, which were reflected in the computed Experimental logC (p < 0.0005). The contrast was found to be reliable for 0.1 log steps in the range from 0 to -2.2 log. On the other hand, for steps of 0.05 log, the contrast was only reliable for values ranging from 0 to -1.7 log.
Discussion: Both luminance and contrast steps differed between iPads with the same retina display, making it necessary to calibrate each display to achieve accurate luminance and contrast steps of 0.05 log units or less. However, for screening purposes utilizing contrast steps of 0.1 log unit or greater for a validated psychophysical test, calibration is not required to achieve accurate results across the displays described herein.