Rifleman1776, in EVERY SINGLE case I found, in the two states I looked at, where a questionnaire grade (the NRA indicates grades based solely on the questionnaire by appending a "Q" to the end, EG "AQ" or "CQ") came up against a voting record grade ("A" "B" "F") that was within the same portion of the grading spectrum, the grade based on an actual voting record won.
In every case where two good voting records clashed, the incumbent won. (The NRA isn't stupid; given equal track records, the odds are the incumbent has more influence.)
If you can show cases where questionnaire results, alone, beat proven track records, please do. Otherwise, your claim does not match what I've found so far, in looking into this.