Currently the assessments only seem to allow reporting based on a total value for each group (or perhaps the percentage of total possible score?).

It would be great if it were possible to construct assessments using a mean or other statistic (e.g. median or mode)

Even better would be to be able to create more complex calculations. Some applications might be to allow for differential weighting of questions (beyond simply assigning greater or lesser assessment values), transformation of values (e.g. combining a question with a 4-point scale and one with a 5-point scale), or calculation of new variables using question values (e.g. ratio of two question scores).

Make the infos like browser agent, javascriptversion, OS and additonal infos available in the results.

Some workarounds are not working with different browsers. E.g. you wonder why a hidden question was filled out with a invalid value (solution, very old browser wasn't hiding the question). If you have the additional infos at hand you can quickly pinpoint what might be the cause of that answer pattern.

The variables could also be made accessible for conditions.
But to save them into the results would be already a big help to give feedback about problems and weighting them in terms of how often they would occur in the field.