Basically yeah
I'm not sure how your approach addresses this - unless I'm missing something it's currently only in terms of type dichotomies. If you expand to 4/2-groups and conditional probabilities then it could maybe compare all possible approaches.
Also, this kind of thing may be a bit overkill - for example you can just as easily take two traits that are conjectured to actually coincide, like Ti suggestive and Fe leading. For this all you have to do is check how well the operationalizations correlate.
E.g., have the question "do you have difficulty organizing your thought process?"
and "do you naturally and confidently express your emotional state to others?"
etc. If you find correlations then that would be decent evidence for socionics.
They may reconsider if someone makes significant progress on scientific questions using other means (e.g. rational/mathematical means).