In defence of SAM
SAM by nature can't be entirely accurate even for the year given.
As an example:
In 2003 student A and B get:
80/100 in 3U Maths, 4U Maths, 2U English, 3U English, 4U English, 2U Modern History.
One student could get 90.45 and another could get 91.4. How is SAM meant to get both right? You can only guess that you got somewhere in between. [79.5 and 80.4 will both show up as 80]
This doesn't even begin to think about different distribution of HSC marks for different years and changing candiditures.
Then be nice and give SAM a small margin of error.
So if SAM is 1.5 off, don't whinge about it.
SAM is a nice fun little toy.
Edit: Plugging what would appear to be the same marks of 2 students across 2 years in SAM, I managed to generate UAI's 4.5 apart. This is by no means the maximum range of error, and this is assuming SAM is 100% accurate.