How do [I] design [a] test for measurement of [3D rotation] improvement?
The way to start is to define what you wish to measure.
What needs to be mentally rotated?
Why does it need mental rotation?
What goal needs to be reached by the task?
Are the tasks a matter of correct or incorrect, or is it a graded measure?
Only then can you start thinking about ...
I think what we call ideologies may be too varied across time and locations to capture in a questionnaire format. If you go one level more specific, to something like support of religion, free market, or moral situations, these may be more reliably and universally measurable.
One feature that feels universal is what constitutes a moral violation. Schwartz, ...