# Puzzling Gallup Poll on Internet Use

Gallup has released a poll that shows that the percentage of Americans who use the internet for more than an hour has increased to 48 percent. But some of the claims of change in use by demographic group puzzle me--I don't think they're significant.

First, the data (percent who use the internet for more than an hour per day):

Some of the changes don't really make sense to me. Does anyone really think that the number of 'heavy' userpost-graduates (Ph.Ds, masters, laywers, M.Ds) increased by twelve percent over the last year? Or that unmarried users increased by ten percent? I'm thinking that the statistical caveat that

Results are based on telephone interviews with 1,009 national adults, aged 18 and older, conducted Dec. 4-7, 2008. For results based on the total sample of national adults, one can say with 95% confidence that the maximum margin of sampling error is Â±3 percentage points.

Isn't that three percent sampling based on the total survey numbers? In other words, that three percent error estimate only applies to the general and universal question, "Do you (regardless of demographic group) use the internet for more than an hour per day?" Once you start breaking out demographic groups, don't you have to use the sample size of the group? In other words, if your survey only speaks to 100 post-graduates, then you have to estimate sampling error as if there were 100 respondents.

Also, the 'real' frequency of use should matter--the error rate will be determined by the underlying frequency of the phenomenon.

At the very least, Gallup should release the number of respondents in each class.

Tags

### More like this

##### Lindgren comments on Lott's weighting scheme
Kevin Drum comments on Lott's weighting scheme. He also links to a January posting which has this explanation from Lott explaining how he might have got a weight of 1/8 from his weighting procedure (my emphasis). Whether it is possible depends upon how finely you do the weighting. If…
##### Note to Dems: If You Weren't So Loud, We Wouldn't Have to...
...get drunk and hit you. Digby thinks that's the meaning of this poll by Gallup: A new USA Today/Gallup poll out this morning shows that more Americans blame the Democrats more than any other group when it comes to the inciting the violence and vandalism that have spread across the country in the…
##### Flypaper for innumerates: WSJ edition
It never ceases to amaze me the way the Wall Street Journal combines superb news coverage with a completely clueless editorial page. To balance an excellent news article by Carl Bialik on the first Lancet study, we have an innumerate article on the editorial page by Steven E. Moore. Moore claims…
##### Look at These Poll Numbers
...And tell me what the poll means. These are from a poll conducted on July 5, 2007, by the href="http://www.americanresearchgroup.com/" rel="tag">American Research Group (sorry, no permalink): cellspacing="0" width="90%">Question: Do you approve or disapprove of President George W. Bush…

"I'm thinking that the statistical caveat that