Always baffles me that them using such a small sample size makes them believe it can reflect the entire population. You can make numbers say anything you want (believe I know - I studied mathematics), but it really is just a way to make your article work for you. Let's not forget that woman are more likely to complete these studies, men are more likely to watch ESPN, moreover, one could argue the type of people that watch ESPN are less likely to do one of these surveys. Not saying ESPN isn't losing subscribers, but these surveys don't accurately reflect much. Both Disney and Comcast will have much more comprehensive research to "sell" or "negotiate" with. All these articles due is throw stocks for a loop.
Not that it would really be much value, but I would be curious to know the % of the survey for men vs woman. sports viewers vs reality tv viewers (I would argue a reality tv viewer is more likely to fill out a survey then a sports viewer), and what the percentage of households cable decisions are made by men vs woman, sports viewers vs reality tv viewers etc. but it would be impossible to get those actual numbers and any numbers you do get would again "assume" the sample they have accurately reflects a population. Without a far greater sample size it is impossible to accurately get a picture of the population views (think exit polls - greater sample size, vs pre election polls - often wrong and poor reflection)