Variance components affecting the repeatability of the alternating cover test.

In within-subject and within-examiner repeated measures designs, measures of heterophoria with the manual prism cover test achieve standard deviations between 0.5 and 0.8 deg. We addressed the question how this total noise is composed of variable errors related to the examiner (measurement noise), to the size of the heterophoria (heterophoria noise), and to the availability of sensory vergence cues (stimulus noise). We developed an automated alternating cover test (based on a combination of VOG and shutter glasses) which minimizes stimulus noise and has a defined measurement noise (sd=0.06 deg). In a within-subject design, 19 measures were taken within 1.5 min and multiple such blocks were repeated either across days or across 45 min. Blocks were separated by periods of binocular viewing. The standard deviation of the heterophoria across blocks from different days or from the same day (sd=0.33 deg) was 6 times larger than expected based on the standard deviation within the block. The results show that about 42% of the inter-block variance with the manual prism cover test was related to variability of the heterophoria and not to measurement noise or stimulus noise. The heterophoria noise across blocks was predominantly induced during the intermediate binocular viewing periods.