1 in 4 Americans don't read-CNN article

[BeerWench13]

What bothers me is people who don't read but watch TV.

Then again, if it wasn't for them, people wouldn't think of me as so smart.

Though I understand your meaning, there are some very educational channels and programs on TV. My husband, for instance, doesn't read, but you will find him watching The Discovery Channel, The History Channel, The Science Channel, The Learning Channel and Animal Planet all of the time. He just doesn't have the patience to read about these things.

He seems to gain a better understanding by watching documentaries and programs than he would ever glean from reading a book. TV is not evil and not all of the programming is mind-numbing fluff.

 

[Hollywood24]

I read in an article that Detroit is America's book city. Aparently they read the most books per capita then any other city.

I think its cause their all scared to go outside

 

[unKeMPt]

I don't much know (or care) if I'm right about that - I'm not a statistician, I used the expression 'statistically significant' in a general sense, not a specific one. If they only asked 1003 people about their reading habits, how do they know they asked a truly representative sample? Did they ask a lot more than 1003 people what their circumstances were before deciding which 1003 should participate in the actual survey? In which case, why not survey all the people they asked to begin with? Did they just use a sample of 1003 based on some set of criteria? Or maybe they DID ask gazillions of people but only 1003 bothered to reply? Which brings me back to my original point, unless they were very clever or very lucky, I don't see how 1003 out of 200-odd million can hope to be an accurate representation. But like I say, I'm not a statistician. From a 'man on the Clapham omnibus' common-sense point of view, it doesn't add up. Which is why I always distrust such surveys

There are a number of reasons to be wary of surveys, but this is not one of them. Populations tend follow something called a bell curve; that means two-thirds of the population fall within one standard deviation of the mean, 95% fall within two standard deviations, and 99.9% fall within three standard deviations. If subjects are randomly chosen from the population, as they were with this one, that means we should see results that approximate the actual population.

For example, if you ask one hundred people, you should expect roughly 67 to fall within one standard deviation, 95 within two, and all within three. By controlling the confidence level, one can adjust the margins of error: We can be 90% confident that all surveys taken would show results within a certain margin of error, or be 95% confident that all surveys taken would show results within a larger margin or error (thus the greater confidence in the results.) And, actually, asking any more people beyond 600 or so reduces the standard error so negligably that it's rather pointless to find more people to ask.