Table 3.

Norms and Reliability for ASTEROID and for Some Current Near Clinical Stereotests

	Mean Threshold in Non-Stereoblind Adults, arcsec	Population SD, log10 arcsec	Bland-Altman 95% Limits of Agreement in log10 arcsec = 1.96 SD of Differences Between Test and Retest	Spearman Correlation Between Test and Retest
ASTEROID	57	0.35	±0.64 (raw)	0.63
			±0.56 (quantized)
TNO	5250	0.2150	±0.4550	0.5731
	6031		±0.4831
Randot Preschool	51	0.24	±0.5944
Near Frisby	2150	0.0650	±0.2444
			±0.2750
Randot	2950	0.1550	±0.3450
Titmus	4150	0.0550	±0.1350

Values reported without references are from the present study. The “quantized” Limits of Agreement reported for ASTEROID is after mapping the original thresholds to the closest available Randot Preschool stereotest score. This is essential for a fair comparison, since test-retest differences are naturally smaller where only a few discrete scores are available. Antona et al.50 reported values only in arcseconds. To convert to log arcseconds, we divided the range (standard deviation or limits of agreement) reported in arcseconds by the population mean threshold in arcseconds and by the natural logarithm of 10. For the means, we have reported the value in arcseconds corresponding to the mean of log₁₀ arcsecond values. Antona et al.50 reported only the mean of the arcsecond, so we have used those values. Bosten et al.31 reported mean and median in arcseconds; we have taken the median of the values in arcseconds in the expectation that this will be closer to the mean of the log values.