Introduction

HFI's December newsletter reviews the findings of the research presented in our Putting Research into Practice course. In preparing this course, recent research from various disciplines (including Human Computer Interaction / Ergonomics, Cognitive & Social Psychology, Computer Science, Marketing, Economics...) that might have implications for usability professionals is systematically reviewed. The most interesting, important, and applicable papers are summarized for presentation in our 3 day seminar – essentially a "Cliff Notes" course for usability research, updated annually.

The list below differs slightly from that of previous years. Rather than presenting design "dos" and "don'ts", this year we present key findings of many of the papers presented in the 2003 PRP course. As such, in addition to providing design guidance, this list provides you recent research references to directly justify your analysis, design, and testing decisions.

By the way, if you only read one or two papers about usability this year, read these:

1. How Do People Evaluate a Web Site's Credibility? Results from a Large Study
2. Experts vs. Online Consumers: A Comparative Credibility Study of Health and Finance Web Sites

Index of Key Findings

Interface Definition: Methods
Interface Definition: Use Models
Interface Design: Branding / Credibility
Design: Navigation
Design: Presentation
Design: Interaction
Design: Content
Interface Assessment: Methods
Special Environments: Cell Phones
Special Environments: IVR
Special Environments: Globalization / Localization
Special Populations: Aging Users
Finally (!)... a few General Observations about Design Guidelines.

Interface Definition: Methods

Behavior reported in focus groups do not reflect behaviors observed during usability testing. (Eysenbach & Kohler, 2002)

When you are interviewing users to determine critical and desired product attributes, asking participants to generate rank-ordered lists yields a group of attributes that varies widely. If, instead, you ask participants to describe the ideal version of the product, you will end up with fewer attributes that will highlight attributes that participants use to evaluate the product. (Breivik & Suphellen, 2002)

Interface Definition: Use Models

Users demonstrate three primary interacting approaches on the Web:

• Browsing (non-specific surfing),
• Information Gathering (reconnaissance for future transactions / information finding), and
• Information Finding (seeking specific answers).

Under a task analysis, these three approaches to interacting with the Web have different flows, behavior characteristics, and determinants of success. Therefore, it is important to know how users approach your site. For example, it seems unlikely that Intranet users will do much true browsing – if they are really browsing, they are more likely to be doing it on external sites, instead. (Sellen, Murphy & Shaw, 2002)

Experienced users access the Web less often and more sporadically. (Cothey, 2002)

More experienced users access fewer sites. (Cothey, 2002)

More experienced users tend to browse to sites (either directly or via other trusted sites) rather than getting there via search. (Cothey, 2002)

Users take advantage of visual details (white space, text saliency) to plan their scan paths. (Hornoff and Halverson, 2003)

Interface Design: Branding/Credibility

Users do not evaluate credibility by checking site author or credentials. (Eysenbach & Kohler, 2002)

Users who do not have in-depth knowledge of the content domain evaluate a site's credibility based on its:

1. design look
2. information design / structure
3. information focus.
   (Fogg, Soohoo, Danielsen, Marable, Stanford and Tauber, 2002)

Domain experts evaluate the credibility of a site based on domain specific criteria.
Medical experts consider:

1. name / reputation / affiliation
2. information source
3. company motive.

Financial experts consider:

1. information focus
2. company motive
3. information bias.
   (Stanford, Tauber, Fogg & Marable, 2002)
   For discussion see: Web Credibility: What makes a Web site credible?

Design: Navigation

Although deeper sites tend to be more challenging for users to navigate, there is a tradeoff between depth and breadth (number of choices per level) in speed of finding. (Bernard, 2002)

Structures that have multiple levels should concentrate site navigation information at the first level and at the level closest to the ultimate content pages (or terminal nodes). (Bernard, 2002)
For discussion see: Depth vs. Breadth

Providing a visible sitemap facilitates site learning and encourages comprehensive exploration of a site. (Danielson, 2002)

Less than half of users take advantage of breadcrumbs (even when most report having noticed them). (Lida, Hull and Pilcher, 2002)

Under click-stream analysis, breadcrumbs are not more efficient than other approaches to navigation. (Lida, Hull and Pilcher, 2002)

Expandable menus are slower to navigate than sequential menus. This is particularly true when users are deeper in a site. (Zaphiris, Schneiderman & Norman, 2002)

Sequential menus helps users develop a better sense of orientation within a site. (Zaphiris, Schneiderman & Norman, 2002)

Users scan group labels within indexed content. (Hornoff and Halverson, 2003)

When users have expectations about where desired information will appear on the next screen, they move their eyes there in anticipation while the screen paints. (Hornoff and Halverson, 2003)

Indexed menus and vertical menus are preferred over horizontal menus (in absolute, not statistically significant, terms). (Bernard and Habmlin, 2002)

Users preference tends toward indexed menus over vertical or horizontal menus. (Bernard and Habmlin, 2002)

For searching tasks, pull-down menus provide fastest performance. For browsing tasks, combined global / local navigation provides fastest performance. (Yu and Roth, 2002)

Presenting short summaries with text links helps users understand and predict the content associated with the link. (Baker, Bernard & Riley, 2002)

Users only perceive / encode (change in) elements of the display that they are directly focused on. (Simon & Chabris, 1999)

Design: Presentation

Abrupt environmental changes (or edges) tend to capture attention. Examples of environmental "edges" include color transitions and changes from silence to sound, among many other things. (Olsen, 2002)

Presenting 10 to 50 search returns per results page optimizes both performance and preference. (Bernard, Baker and Fernandez, 2002)

Users find it difficult to ignore gestalt grouping principles of color similarity and common ground. (Beck and Palmer, 2002)

Color similarity has a stronger perceptual influence than common region, proximity, or grouping. (Beck and Palmer, 2002)

Proximity has a stronger perceptual influence than common ground for grouping. (Beck and Palmer, 2002)

Users expect a back-to-home link in the upper left corner of the page. Bernard (2002)

Expectations of footer links are fading. This reflects a change from Bernard (2001). Bernard (2002)

Users expect Help / Service to appear in the upper right corner of a page. Bernard (2002)

Highlighting with surrounding silence directs attention and enhances recall in short lists. Extrapolation: white space serves to direct attention and enhances recall. (Olsen, 2002)

Using animation in banners helps to capture attention but does not necessarily increase recall. (Bayles 2002)

Chart junk (additional visual information, often included with good intentions) gets in the way. (Blasio and Bisant, 2002)

Users can compare and evaluate absolute information (tables) more efficiently than relative information (graphs). Users prefer relative presentation. (Blasio and Bisant, 2002)

Banner animation does not significantly undermine speed of performance. It does, however, significantly increase perceived workload and frustration of users. (Burke and Hornoff, 2001)

Users favor fluid page presentations over centered or left-justified presentations. There is no performance difference across the three presentation styles. (Bernard and Larsen, 2001)

Users prefer sans-serif fonts to serif fonts on the Web though there is no significant performance change. (Bernard and multiple other authors in multiple papers)

Users read 12 point fonts faster than 10 point fonts. (Bernard and multiple other authors in multiple papers)

Users read Times and Arial faster than Courier, Schoolbook, and Georgia. (Bernard and multiple other authors in multiple papers)

At font size 10, users prefer Arial, Courier, Comic, Georgia, and Verdana to Times. At 12 point, Arial is reliably preferred to Times. (Bernard and multiple other authors in multiple papers)

Design: Interaction

Users' mental model of how the back button works is so fragile that you can change it experimentally and most won't notice. (Cockburn, McKenzie and Jason-Smith, 2002)

Poorly titled pages make using the Back Button / Browser History harder. (Cockburn, McKenzie and Jason-Smith, 2002)

Users report that scrolling on content pages is not worse than paging. (Baker, 2002)

Users typically peruse the first page of search results (93% of links visited were within first 10 results). (Eysenbach & Kohler, 2002)

The specific navigation approach that users demonstrate (click-navigation vs. search behavior) tends to depend on the task at hand. (Spool, 2001)

The more times users try to search, the less likely it is that they will find what they are seeking (if users search 3 or more times, they won't find it). (Spool, 2001)

Providing category information helps users group and effectively evaluate search returns. (Dumais, Cutrell and Chen, 2001)

Delays greater than 100ms effectively "hold the perceptual system hostage". (Hornoff and Halverson, 2003)
For discussion see: Are We There Yet? Effects of delays on users' perception of Web sites

Design: Content

Presentation style influences how users encode and interact with content. (Jedetski, Adelman & Yeo, 2002)
For discussion see: Decisions, Decisions. What's a poor designer to do?

Presentation can influence user decision-making strategies (compensatory vs. non-compensatory). (Jedetski, Adelman & Yeo, 2002)

Content RE-written for presentation on the Web reduces time to task (80%), improves memory for content (100%), and increases satisfaction-with-use (37%). (Morkes & Nielsen, 1998)

Users' eyes are drawn to pictures of human faces the first time they see a page. (Riegelsberger, Sasse & McCarthy, 2002)

Photographs do not increase the trustworthiness of already credible sites. They do, however, improve the credibility of sites that are not generally perceived as trustworthy. (Riegelsberger , Sasse & McCarthy, 2003)
For discussion see: From Bricks to Clicks: Building customer trust in the on-line environment

Interface Assessment: Methods

If you conduct both usability testing and expert / heuristic review on the same site, there will be some overlap in the usability problems that are identified. However, there will also be a specific set of problems that is uniquely identified by expert / heuristic review, and an orthogonal set of problems that is uniquely identified by usability testing. (Fu, Salvendy and Turley, 2002)

Heuristic review tends to uncover usability issues related to presentation (skills- and rules-based user performance). (Fu, Salvendy and Turley, 2002)

Usability testing tends to uncover issues related to domain-specific knowledge and interaction (knowledge-based user performance). (Fu, Salvendy and Turley, 2002)

For discussion see: Pitting Usability Testing against Expert Review

Special Environments: Cell Phones

Talking on a cell phone undermines attention to driving. (Trbovich & Harbluk, 2003)

Special Environments: IVR

IVRs with 3 or fewer levels, and 4 or fewer choices per level, work best. (Dulude, 2002)
For discussion see: Press 8 for Natural Language

Special Environments: Globalization / Localization

Comprehension and recall is enhanced when information / arguments presentation (inductive vs. deductive) is culturally appropriate. (Spyridakis & Fukuoka, 2002)

Special Populations: Aging Users

IVR Study: Older users have a harder time recovering from errors. (Dulude, 2002)

Older users have a harder time understanding / adjusting to internal jargon. (Dulude, 2002)

Finally (!)... a few General Observations about Design Guidelines

Design guidelines can be helpful, but they should not be construed as rules. Users can (and do) learn and adapt to new environments – even when things are not quite as expected. (Spool, 2002)

Design guidelines should be specific. Their effectiveness should be measurable and testable. That is, they should be "operationalized". (Spool, 2002)

Know HOW and WHY the guidelines you are applying were created. Were they created in the same task-design context that you want to apply them in (e.g., If you are designing a data-mining interface, look for guidelines that speak to designing for that specific task environment.) How were they / will they be tested? Are they derived from sound research? (Bosley & Straub, 2002)