Slashdot Mirror
Women Now Outnumber Men Online
miller60 writes "There are now more American women than men using the Internet, according to a new study from the Pew Center on the Internet and American Life on gender and use of the Net. While a slightly larger percentage of men than women are online (68 percent vs 66 percent), the larger population of American women tips the balance. Other findings: younger women and black women outpace their male peers by larger margins than the wider population."
Before somebody says "68+66 does not add up to 100%" I suppose the submitter meant "While the percentage of men who are internet users (68%) is slightly larger than that of women (66%)"...
The filesystem is the package manager
Jesus, what a silly comment.
)
http://en.wikipedia.org/wiki/Statistical_survey
http://en.wikipedia.org/wiki/Sampling_(statistics
Ok, I'm not one to defend surveys much if at all given that many of them are biased and not done correctly, but the reasoning you mentioned is simply flawed. Surveys are not meant to be done on the entire population, because polling 280 million Americans would be an impossible and unfeasible task.
/rant
Surveys take a sample of the population to be polled and use that as a representative measure of the rest of the population. The sample size then is given a confidence interval of +/- percentage points (usually 5%) that indicate the accuracy of the poll, within a reasonable standard deviation. In English, this means that polls aren't 100% accurate, but a properly done survey should be accurate within 5% of the acutal figure the majority of the time.
Selecting a random sample from the population is often the hardest part of any survey, but can be done correctly. To flat out say that using samples means that the data is irrelevant is completely inaccurate.
Please read a little about statistical analysis.
Using your population number 280000000 and 6,403 people survayed that gives a 95% confidence (which is about the norm for this type of study) that the survey answers are correct and apply to people who were not surveyed. This is with an margin of error of 1.22% which means that survey results may vary by 1.22% in either direction. This is all provided that the people surveyed where a random sample of the US population.
Check this BBC article.
http://news.bbc.co.uk/1/hi/technology/4555370.stm
Apparently men and women use the Internet for different priorities.
Could this suggest that there is actually a difference in the genders?
The mystery thickens...
But you run into trouble when you are sampling a small part of the population, even if you have a fair sampling, say the 6000 mentioned. If half of those go online, then you only have a sampling of 3000, of those, half are women, down to 1500. And if you are working out what women online prefer, your sample is becoming woefully inaccurate, as there is no way to have an accurate representation of all demographics.
Besides, 97% of all statistics are simply made up.
Some say he is made with ascii, others that he is eyeballed daily by millions. All we know is, he is known as the Sig
The vast majority of men experience their primary arousal from visual sources.
While some women may peruse pornographic images, they are more likely to prefer text based accounts and descriptions.
I do not remember the source, but the use of the Internet to access pornography is growing across all age and gender demographics.
I use irony whenever I can, but my shirts are still wrinkled...
survey results may vary by 1.22% in either direction
The word is "may." It is possible. It is highly unlikely, however -- one chance in twenty, if the parent's analysis is correct. The odds are nineteen out of twenty that the results do not vary that much. Does any of you bother to know what he's talking about before posting a comment? Oh, wait, this is Slashdot. Some options: 1) Learn something about statistics other than stupid, misapplied cliches. 2) Propose an alternative to sampling. 3) Point out factual examples of how this report's sampling is wrong. Writing they might not have sampled properly or that the respondents might have answered wrongly is obvious and is a pointless waste of electrons and time.
You're right. The classic test case for nature vs. nurture was a horrible fraud. I highly recommend reading this if you might have any interest in the issue.