The pioneering Eyetrack III study was carried out by the Poynter Institute, the Estlow Centre for Journalism and New Media and technology firm Eyetools.
The team uses small cameras mounted just below the computer screen to monitor the movement of the viewer's eyes as they navigate the web page. Researchers then compile and analyse data from the cameras to identify patterns in viewing behaviour.
This information is extremely valuable to news sites because they can use the findings to improve navigation and 'stickiness' on their own websites.
The study found that fairly small differences in text size and headline format can be enough to keep a reader's attention.
"We found that on news home pages with a mix of headlines and blurbs, contrasting size between headline and blurb seemed to encourage more scanning - looking at the headlines but not the blurbs," said Mr Outing.
"When type size was smaller and headlines were smaller, they tended to be read more as blocks - more fixations on more words and more reading."
The team also found that subjects often tended to read only the first two or three words of a headline.
"Those first words were critical in deciding if the entire headline was read or not," Mr Outing told dotJournalism.
"It really points to the importance of headline writing, and putting the action words at the beginning to grab attention."
The Eyetrack team monitored 46 volunteers for an hour each while they viewed sample news sites and multimedia news features. They also studied 25 major news sites including Guardian.co.uk, FT.com and WashingtonPost.com, evaluating readers' responses to common layouts and design devices on screen.
Readers' eyes are drawn to the top left hand side of the screen, and they tend to scan the page in a z-shaped pattern. From the results, the team compiled a visual guide to the priority of each part of the screen.
Users are less likely to read standfirsts if there is a line or visual break between the headline and the standfirst.
Shorter paragraphs attracted twice the amount of eye fixations than longer paragraphs, which tended to discourage viewing.
Standard one-column news stories held the readers' attention better than those sites that mimicked newspaper with several smaller columns side-by-side.
Eyetrack also examined online adverts, and found that text adverts held readers' attention more than images. Adverts were also more likely to attract eye fixations if they are positioned close to popular editorial content, and if they occupy a space at the top left-hand side of the screen.
Eyetrack III also includes the first study of multimedia news features. Viewers were asked to recall key facts from the same story, one group using a text story and one using an interactive multimedia graphic.
Researchers found that information recall was better from those that had read the text story, although multimedia graphics proved more successful in illustrating more complicated concepts.
Full details of the research are available on the Eyetrack website, where results are presented alongside suggestions for effective homepage design, advertising formats and article layout.
The first Eyetrack study, from 1990-1991, investigated how readers interacted with print newspapers. Eyetrack II, from 1999-2000, looked at viewing patterns for some of the first news websites.
More news from dotJournalism:
New study tracks readers' eyeballs
Q&A: Steve Outing, senior editor at the Poynter Institute for Media Studies
Thousands attend NetMedia conference
See also:
Eyetrack III: http://www.poynterextra.org/eyetrack2004
Estlow Centre for Journalism and New Media: http://estlowcenter.du.edu
Eyetools: http://www.eyetools.com
Free daily newsletter
If you like our news and feature articles, you can sign up to receive our free daily (Mon-Fri) email newsletter (mobile friendly).
