Let's break this down, the samplesize was 4300, IMO way too small for an accurate result anyway, and of those 9% considered WP8 which is 387 people. Of these 197 prefer Samsung and 174 prefer Nokia, leaving 16 people for HTC
Hardly numbers Nokia will worry about. I am willing to bet that the question was presented as;
----
If you would consider buying a Windows Phone 8 device which brand would you prefer?
A. Samsung
B. Nokia
C. HTC
D. Other/Don't know
------
In this line of questioning the difference between A and B would be explained by people often simply choosing the first option anyway. Only when the sample size would be much larger (as in at least 3-4 time larger) would an accurate measurement of preference be available.