Do these numbers look right to you?

The Associated Press released a poll sponsored by AOL a few days ago (produced by Ipsos - PDF link). Out of 3024 adults 1046 played video games of any sort (PC or console, no questions were asked about mobile gaming). The margin of error is almost +/- 3% for the game section.

Of the ones who played video games (two different questions):

• 45% played on the net (471).
• 63% typically played by themselves (659)

That right there tells me there was a problem. More people do not play over the net (55%) than do (45%) but even more play by themselves, usually. However, there was no breakdown on how often people play on the net, there was only a question about how often they play games in total (with two peaks at 2 and 4 hrs., while the largest slice only played less than an hour).

Of the 471 who played online:

• 18% had formed relationships with others online (85)

So only 85 people out of 3026 had formed a relationship with someone they had met playing an online game. That is a small percentage (about 2.8%).

There's more info at the Ipsos page via the PDF link. It's unfortunate that cross question correlations cannot be made (ie. how many people who make more than $50 000/yr. play online for more than 2 hours a week and describe themselves as "white").

Some other oddities in the poll:

15% of those who say they play games also watch >25 hrs. of TV a week (where do they find the time - are these the gamers who only play 1 hr. a week or less?)
No questions were asked on how much they spent on hardware vs. software vs. online charges (just the totals are available).

It would be interesting to know how representative these numbers truly are. For example, 37% say they are from the south (over 1/3). Does that number represent the rest of America. Does 1/3 of the population live in the south?

I would really like to see the complete raw data.