CNN Snap Poll Has Poor Sample Group

The CNN poll presented following the presidential debates last night lacked an adequate sample group, focusing entirely on the opinions of white Southerners aged 50+.

According to the breakout of the poll, which can be found in pdf format directly from CNN, 430 registered voters were polled, with opinions from non-whites falling well below the sampling error, indicating that the majority of the people polled were white, 50 or older, and located in the Southern United States. The breakout shows columns referring to nonwhites, those under 50, and those living in geographical regions other than the Southern US listed as “N/A.” This does not conclusively mean that no one was polled who fell under a N/A category, but that if people were polled, their numbers fell below the sample error cut off — for example, one minority 20 year old would be considered a statistical error.


An adequate sample group requires variation in terms of age, gender, race, and geographical location. While this poll is accurate, it is not accurate for all registered voters as CNN has presented it, but does indicate the opinions of older white Southerners. Claiming the poll represented the views of all registered voters is the same as saying that just because golden retrievers bite more children than bears, that bears make better pets. It draws an inaccurate conclusion based on insufficient data and flawed methodology.

Unfortunately, the ramifications from this poll, inaccurate as it is, are yet to be seen. It could potentially affect Obama’s standing negatively through a sort of “follow the leader” where people vote based on public opinion, and not the facts. The ramifications of CNN’s bad approach to the debate polls remain to be seen as more debates ensue.