"Given the screenshot you posted, I wouldn't really buy into your conclusions because you never really adjusted your designs to fit the varying widths. I think I can agree that have a landing page take up 100% of the available width is a bad thing, but it's not because the user is spending time turning their head to read content (that's a pretty poor argument); rather it's because users like to see visual boundaries -- where the content of the site begins and ends. Your widened versions appear to do a poor job of managing the space and as a result look more amateur than the thinner, more condensed versions. There are multiple problems that I can spot right away with your widescreen version -- for example, the user preferred width of a column of text is 3.5", and the right-hand utilities section of the header appears to be disconnected from the rest of the page. Problems such as those, stemming from the design of the page (independent of its width) are probably more than enough to explain the difference in bounce rates.
"Furthermore, are you arguing that the decision made by CNN, Digg, CNet (et al.) to switch to a native 1024x768 design was wrong? I think the conservative argument you make in your post is quite backward -- online store owners would do better to expand their site designs and provide an enhanced experience for customers."
While it is true CNN, Digg, CNet and others have chosen to switch to 1024x768, these sites target a different profile of users than the target demographics for our study.
SoftwareProjects is in the business of building websites that "sell". And by sell I am referring to the ability to convert an anonymous website visitor who is visiting a new website for the first time, into a qualified lead and then a paying customer.
Users frequenting Digg and CNet are in a completely different mindset, than someone looking to get a mortgage quote, or a user who is surfing the net looking for a new bookshelf.
For these type of websites, promoting a specific product/service, it's been shown you typically have 4 seconds to convince the user to stick around, before they hit the BACK button.
Our research conversion data shows users were more comfortable with buying when presented with a thin website.
If you examine popular retail sites targeting the average Internet user, you'll find they all follow a similar paradigm: Overstock.com, Target.com, BuyOwner.com, WalMart.com etc. - are all designed as "thin" sites, deliberately not utilizing the entire width of the page.
I feel like this is getting fairly pedantic but the question "what width should your site be?" is flawed -- there is no single correct answer. It's alot like asking the question: "which browser should I develop for?" Clearly, user environments vary. The developing team must take this into consideration and work accordingly.
I think the best answer you can give to someone asking your question is the following:
If you can only design a single layout, use whatever resolution allows you to maximize available screen real estate while supporting the largest subset of users. (Note that this does not necessarily mean you should design the site to take up the entire width of the browser.) However, if you have the resources to evaluate multiple user environments (and I'd make the argument that EVERYONE does), then do so and provide conditional formatting. So, for example, a user with an 800x600 display will see a design optimized for that resolution without compromising the design used for others with larger screens.
The scope of the experiment is so narrow that I wouldn't draw any conclusions based on this, unless I was making a cheaply-designed, SEO-driven, diet-pill selling website. Even then, the difference could be attributed to poor design (irrespective of width).
Sorry if I came across as overly critical - I wasn't suggesting that your conclusion is invalid, but that the experiment would only apply to a fairly narrow class of website (e.g. "mortgage-leads... business-leads... diet-pills... and ringtones"), and that you may not have varied the design significantly (screenshots would have been great!). The 'cheaply-designed' thing was uncalled for, and for which I apologise; I think I was trying to suggest that a different 1024 design style may make a big difference, but it came out snarky, probably because I'm working on something frustrating.
"Given the screenshot you posted, I wouldn't really buy into your conclusions because you never really adjusted your designs to fit the varying widths. I think I can agree that have a landing page take up 100% of the available width is a bad thing, but it's not because the user is spending time turning their head to read content (that's a pretty poor argument); rather it's because users like to see visual boundaries -- where the content of the site begins and ends. Your widened versions appear to do a poor job of managing the space and as a result look more amateur than the thinner, more condensed versions. There are multiple problems that I can spot right away with your widescreen version -- for example, the user preferred width of a column of text is 3.5", and the right-hand utilities section of the header appears to be disconnected from the rest of the page. Problems such as those, stemming from the design of the page (independent of its width) are probably more than enough to explain the difference in bounce rates.
"Furthermore, are you arguing that the decision made by CNN, Digg, CNet (et al.) to switch to a native 1024x768 design was wrong? I think the conservative argument you make in your post is quite backward -- online store owners would do better to expand their site designs and provide an enhanced experience for customers."