Accuracy in astronomical calculations is a tricky thing. There are a lot of competing factors and it's often difficult to know how right or wrong you are, because you can't go back in time to test your calculations. We've done our best to be accurate in our calculations, but things in the heavens aren't as clockwork as you might believe.
So, the best thing we can do is to try to match up what is displayed in Starry Night with actual recoded astronomical events, like eclipses, which seem pretty good.
Starry Night has been used for "Archeoastronomy" before and has been found by many people to be accurate, matching up with e.g. precisely measured eclipses and other events far into the past. That being said, nothing can be %100 accurate, computers are only as good as the models they use and we're doing our position calculations in "real time" in order to present a smooth, "movie like" quality. If you want extremely precise calculations for a given planet, I'd encourage you to use Starry Night as a primary source, then use NASAs JPL Planetary Ephemerides calculator to get precise positions (and also check USNO). It's more complicated, requiring a better understanding of Astronomy, but you'll get more precise (and perhaps more accurate) results.
Previous to SN 6.2 Starry Night used a Delta-T calculation based primarily on equations provided by Greame Waddington of Oxford, along with a correction needed to match up several historical eclipses with the other calculations SN was making. (Code available, if necessary.)
SN 6.2 and later uses the NASA calculation seen here:
Note that between 6.2 and 6.4.3 for events between 500BC and 500AD there was a bug that would have thrown off the calculation of Delta-T somewhat. This has been fixed.
To support your conclusions always check other sources.