Interesting survey results on how web design affects users' confidence and decision to purchase: http://www.clickz.com/stats/sectors/software/article.php/3578366
With the exception of Amazon and other sites I know that I want to come back to, even before I make a purchase from them, I wont purchase if I am forced to register. Offering registration is fine, but not requiring it. Required registrations works for some sites, but especially for smaller online stores, I would definitely stay away from it. On the other hand it can be a very good deterrent of fraud, so for some, the money saved can definitely be worth the loss of customers.
Disturbing? No matter what you are doing online, every little barrier you put up is going to hack another percentage off of your conversions (be that buying something, signing up for a newsletter, registering for a forum, whatever.) Its just common sense.
Now, although this survey is excellent food for thought, I think we have to be a little careful with this data. Remember this is a survey of what people say they like / dislike, and what they say they would or would not do. Think about it this way: Look how polls go up and down before an election. Bush is ahead. Bush is behind. People both change their minds, and say things that they do not really want to follow through on to pollsters. The real issue is how do they actually vote on the day. Or do they even vote at all? Do they just stay at home? The same can be applied here: Like the pre-election opinion polls, this kind of survey gives us an *idea* about how people *might* behave in reality on your site. But, there is much better data for you out there - your own log files! Have a really high bounce-rate? Then your page might have a nasty design, or maybe it just isn't clear what the page is really about. Either way, people leave. Lose a lot of customers between when they put the product into their shopping cart and purchasing? It could be a long extended process, or maybe your site wants potential customers to sign up first before they can buy. Anyway, with this data in hand, you can then get some guinea pigs (your friends) who you can observe interacting with your site at these points. Video them and then get them to commentate on the video when you watch it together afterwards. At this point you will have an excellent idea of the changes you need to make. My opinion is that this survey is useful in the sense that it tells you the questions to ask of your site. But the answers can only come through looking at how real visitors really interact with your site - and you can see that in your log files and through usability testing with real people.