#1,4 are biased because the number range is too wide. (include #2 if the question allows)
#1: 31-20 = 11
#2: 29-21 = 8
#3: 30-23 = 7
#4: 50-15 = 35 (OMG that's so biased!)
so if the three surveys are eliminated, then the answer to the second question is 23%.
Hope that helps :)