Tim Napier-Munn and Graeme Jenkinson have published a paper on errors in measuring software in the Journal of Double Star Observers.
This paper reports quantification and explanation of some sources of error in using the Reduc software to measure the PA and separation of double star images. A comparison of measures of the images of two pairs, one close and one wider, made by the two authors independently has shown that the two individuals produce measurements that are numerically nearly identical and are not significantly different statistically. Analysis of the repeated measures of 38 pairs has shown that the standard deviation of PA increases with decrease in separation due mainly to the geometry of the measurement. The standard deviation of separation increases with separation due mainly to propagation of the error in determining the separation calibration constant. All these errors can be controlled to any desired precision by conducting sufficient repeats.
Napier-Munn, T. and Jenkinsin, G. (2014). Journal of Double Star Observations Analysis of Errors in the Measurement of Double Stars Using Imaging and the Reduc Software. Retrieved from http://www.jdso.org/volume10/number3/July2014B.pdf