Proposed methodology for testing aesthetics

Here are some thoughts on how one might be able to structure a cheap test to explore the usability implications of purely aesthetic variations. Maybe "usability implications" is a bad term since there is a commonly accepted dichotomy that places usability issues on one side and aesthetic issues on the other. I'm not sure what other word to use and I think placing usability on one side and aesthetics on the other is a "false dichotomy". 


The methodology is for a cheap test in the sense that I think it can be meaningful even with a handful of participants. Here's how it would work:

1. Have a user perform a "control" task like finding the lowest available priced KindleDX, including shipping, available to be purchased immediately on ebay.com. Time their performance of this task.
   
   
2. Then take them to one of two different versions of a similar site, like ubid.com. For some people, take them to the regular version of ubid.com (Figure A). For others, take them to a more aesthetically suspect version of the site (Figure B). Ask them to perform the same task (finding the lowest available priced KindleDX, including shipping, available to be purchased immediately) and time them again.

Figure A

There is a Google Chrome extension called Chrome Sylist that allows you to define local CSS files that override the CSS defined within the HTML code.  It's a pretty cool tool, allowing you to define a general override for all websites, an override for specific domains and an override for specific web pages on specific sites. This might be kind of a nice thing to install for a parent or grandparent that has trouble with readability on certain websites, maybe issues that aren't easily addressed simply through the browser's zooming function.

Figure B

I'm not entirely sure what it was designed for but it's certainly also very useful for testing usability with public websites whose styling you don't otherwise control.

I'm proposing that you could look the users' timed performance with ubid.com relative to their timed performance with ebay.com. Obviously, some users take longer than others to perform tasks but you can still compare users with varying levels of speed, computer facility, etc. with each by using this relative measure of performance.

I suspect that what you would see is that users who use the Figure B version of the ubid.com site will have longer tasks times relative to their ebay.com task times.

There are certainly non-task oriented sites where timing test are either difficult or impossible but I think that testing task oriented sites in this way can at least establish a few base level design principals that can perhaps be extended into areas where testing isn't possible.