Internal Inconsistencies vs Root Mean Square as Measures of Threshold Variability

Abstract

Although differences in response are permitted at the final threshold, any dimmer spots which are seen and any brighter ones not reported are considered internal inconsistencies, here called boo-boos. The root mean square (RMS) used by Octopus as a measure of intratest variability requires duplicate threshold testing in ten spots. In an effort to eliminate the time tax this implies, boo-boos were recorded during two double reversal threshold tests (4, 2, 2) at ten selected loci in eight surgically controlled glaucomatous eyes. The same 80 loci were then tested to 50% thresholds, with 2 minute rest periods after each five minutes of testing. Despite this precaution, a progressive deterioration of threshold occurred, preventing psychometric analysis in 19 of 71 valid loci. These data showed important differences from the 52 places where results were fitted to ogives. The RMS was approximately twice as effective in predicting the variable loci as the boo-boos. Since additional data for analysis of boo-boos are free of time penalty and precision increases as the square root of increase in data, this preliminary evaluation suggests boo-boo quantification may help to identify variable test loci.