It's quite easy to actually...
1. Location HEAVILY skews the data. People will look at unis in the north with lower average grad salaries and come to the conclusion that they suck, when it's a case of more grads choosing to stay in that area (with suppressed wages due to lower cost of living) than moving down to London or the South. Likewise, salaries at London unis are inflated for the same reason.
2. Students choose to go into varying jobs after university, some of these jobs happen to be insanely well-paying and thus skew the 'average' into a ridiculous area. LSE Econ, UCL CompSci etc are culprits for this. The vast majority of grads sit in a numerical region that is lower than that average.
3. Data for some courses can be dodgy because the graduates of said courses tend to opt for lower paying (even if they are graduate level) jobs than average. Nothing is stopping a grad from that degree course opting for a higher paying job
4. UniStats salary data for courses has a laughable sample size. When graduating classes are 100-300+ per year and the representation on UniStats only includes 10-50 data points, you can bet your bottom dollar that you aren't getting the full picture
Personally, I think salary distributions and most popular job titles + company destinations would be a lot more effective than the way UniStats currently does it.
Posted from TSR Mobile