Reading On The Web: A Study
Right-hand navigation seems to be the status-quo of weblog layout. Why is that?
Right-hand navigation seems to create an easier-to-read layout for reading large amounts of text. It is much easier to read content where the navigation is on the “ragged” edge (right) of the text than if the “ragged” edge of the navigation was closest to the left-justified edge of the text (left). The left-hand navigation distracts the eye when a new line is started.
But what if there was some kind of wall or guide seperating the left-hand navigation and the left-justified edge of the text? Would there be enough of a performance increase to be comparable to right-hand navigation?
I aim to find out just that.
Goals
The goal of this study is two fold: 1) to establish that right-hand navigation is easier to read than left-hand navigation and 2) discover if some kind of wall or guide between left-hand navigation and the content would make the content easier to read.
Definitions
“Easy” in this context is performance: the amount of time necessary to read text.
I have chosen four publicly available articles. They are of similar difficulty and length and deal with various topics which are neutral enough to not effect the score of someone who has extensive knowledge in the subject.
I am testing four different layouts.
- Control - content is left-justified and no navigation is present
- Right-navigation - content is left-justified and navigation is left-justified; navigation is aligned to the right of the content
- Left-navigation - content is left-justified and navigation is left-justified; navigation is aligned to the left of the content
- Left-navigation with “guide” - content is left-justified and navigation is left-justified; navigation is aligned to the left of the content; a thick “guide” seperates the navgiation and the content
A 12 point sans-serif font is used for the layout which will give the greatest efficiency. Content is displayed in a 550 pixel column which displays approximately 70-80 characters per line which will provide the best layout for comprehension.
Methods
The four articles will be randomly rotated through the four layouts for twenty four possible data sets. The user will be instructed to comfortably read each of the four articles in a randomly selected layout while being timed. The same layout will not be used more than once in a testing instance. Times will be recorded along with the correlating article and layout, and a survey of information about the user. The user survey is not data and will be used to draw conclusions based on interesting correlations between the results and the survey statistics.
Remote testing poses a mutitude of issues and apprpriate data screening will take place in order to identify the integrity of each record. Data which does not have a verifiable email address (to identify duplicate test users) will be thrown out. Records which have incomplete times will be thrown out. Records which have abnormally high or low times will be identified as irregular and may be disqualified from final results.
Results
The statistical formulas have not yet been developed to analyze the data.
The data collection has not been completed.
The full report will be posted at a later date.
The Test
If you would like to participate in this study, please follow the link to the Reading On The Web welcome page. Feel free to distribute this link to anyone who may be interested in participating.
If you were involved in the beta test of the survey, please do no re-take the test. You may distribute link www.obso1337.org/usability/reading_on_the_web if you would like to help recruit participants.
Notes
This entry is a draft of the final presentation and is incomplete.
[…] actual test is available here: http://www.obso1337.org/usability/reading_on_the_web/ Her weblog entry about the launch of the test and additional information about the methods used.
[…]
[…] reated a usability study about reading on the web, and she needs help getting data. Please read about it and take time out of your schedule to try this test out. Thank you. This […]
How will you rely on an e-test? You say “Records which have abnormally high or low times will be identified as irregular and may be disqualified from final results.”, but is this enough? Can you guarantee its validity?
Distance testing is a good way to get a larger sample size at the price of control over the system.
“Records which have abnormally high or low times will be identified as irregular and may be disqualified from final results.”
By identifying independent factors, normalizing the data, I should be able to identify trends. Thats what statistics is for. There are pretty solid numbers which can identify reading levels and differentiate between reading cognitivly and speed reading. The data wont be as solid as a controlled usability study, but it will be able to judge if the differences are significant enough to matter.
Keep in mind this is only a quanitative test. There have been numerous studies on emotional design which have found that regardless of usability, some users feel more comfortable using something that is “pretty” or “normal”. Good human-centered design is finding the medium where things are both usable and aesthetically pleasing in order to have the optimum design for user comfort and performance.
[…] ork is pretty slow today so I have been gathering my data and working on statistics for my Reading on the Web study. Over the past two weeks I have had a little over 90 participants. I think participa […]