Cultural Dimensions of Software Help Usage
Contents
Click a link below to jump to a particular section; click any "CONTENTS" image following a section heading to jump back here.
The Origins of a Study 
When conducting usability testing, researchers and usability experts start by analyzing basic user demographic data. They often create personas to represent different user groups, and then use that data to select or classify the actual testers (the people participating in the usability study).
In two recent usability studies, I encountered tester behaviour that didn't fit my expectations based on demographics. These were minor anomalies that did not affect the overall study results, but I was curious. On examining the cases, I began to suspect that there was another influencer on user behavior; something besides users' domain knowledge, computer experience, product knowledge, and education level was influencing their use of product documentation.
Could it be culture?
Some Background on Culture
The very concept of culture is tricky. We generally think of culture as attitudes and behaviour that are characteristic of a particular social group or organization. But what about age, or gender, or country of residence (as distinct from country of origin), or mother-tongue language, or profession?
Some existing theory-based research provides us with clues about how culture affects user performance. Edward T. Hall's theory of high-context vs. low-context cultures is one example-usability data has confirmed that the HCC (high-context culture) Japanese user is far more likely to read documentation in order to get a better overall background and understanding of a product, while the LCC (low-context culture) American user tends to avoid reading at all costs.
Study Objectives
In early 2010, I returned to this idea of culture as a predictor of a user's online Help usage. Assuming that the general demographics of education, computer skill, and socio-economic level were constant, would other cultural "overlays" make a significant difference?
- Would we see real differences in usage based on gender? For example, are women more willing to "ask for directions" (that is, look at online Help) than men?
- Would age groups constitute unique cultural groups with their own behaviour? For example, were older users more likely or less likely to read Help?
- Would HCC users access or read online Help differently than LCC users?
- Would cultural indices, such as Hoftstede's IDV or UAI scores, effect how willing a user would be to use online Help? For example, would high-UAI cultures mean that users would be less likely to explore Help, or more likely to seek formal answers rather than just "play" with the product?
Methodology
Participants
The study consisted of 38 participants, all falling into the "normal" range (80% Bell Curve group) for computer experience (that is, no expert or novice users).
- 17 male, 21 female
- age 17 to 78
- mother-tongue languages: English, Hebrew, French, Arabic, German, Afrikaans
- cultural groups: gender, age, context (HCC and LCC), and MTL (mother-tongue language)
Tasks
Participants were paired with an application or task that they did not already know how to do. Tasks were described vaguely or in user terms, rather than in terms of the product GUI. Task scenarios included:
- crop a picture and convert it to sepia (PhotoPad Image Editor)
- insert a picture into a document so that the text wraps around it (MS Word)
- create a simple database (MS Access)
In all cases, users were requested to think out loud, and had an opportunity for an exit interview.
Challenges
Beyond finding testers and dealing with the false-positive effect of usability testing (that is, the tendency of testers to try harder and complain less because they are being observed), the biggest challenge proved to be finding scenarios that forced users to access online Help.
Results
Gender is tranparent
Despite the myth that women are more likely to stop and ask for directions, the women ignored online Help as much as the men! There was so statistically-significant difference at all. Ironically, several of the women partipants commented on how "men never look at Help" while blithely ignoring the application's Help menu themselves!
Experience is the equalizer
Low-end (less experienced) users showed a bigger difference, with women more likely to blame themselves for mistakes, rather than the application or unclear Help. But as the experience levels increased, this disappeared. High-end (more experienced) users were consistently more likely to blame the application and complain that the Help was useless.
Age is more significant than any other factor
Age alone was the strongest cultural influence on usage patterns. It made no difference how experienced a user was; every single user over 60 missed certain visual cues (Help buttons, links to more information, on-screen messages, etc.). Subtle visual cues such as the change of a cursor or the appearance of sizing handles, or even a line of Help text appearing above an object, was invisible to them. While I expected some of this, I was astonished at how significant the problem was.
Linguistic issues were more significant than expected
Non-native English speakers (but with very high-level English skills) had consistent problems with vocabulary in Help topics. This was expected. What surprised me, however, was the number of educated native English speakers who struggled with slightly technical vocabulary!
Inattentional blindness was significant in all groups
Over 50% of older users and 25% of others thought that they had succeeded despite massive errors. They ignored embedded warnings and on-screen text, and ended up overwriting existing files without noticing, saving to the wrong location, or wiping out their work. In some cases, users read error messages and then still clicked OK without appearing to understand the ramifications.
Learned distrust was far greater than expected
All high-end users showed great distrust of online Help. In fact, in 58% of the tests, no Help access was achieved. As one user said, "I never look at Help because it wastes my time."
Recommendations
Rebuild trust by providing meaningful answers
Users have learned to avoid Help because of useless self-referential definitions and topics that only restate the obvious. Provide good explanations and clear procedures.
Support the needs of older users
Make embedded Help text larger. Avoid side panel text. Make links to Help more obvious. Work with the GUI designers to make all visual cues more obvious and comfortable to older eye.
Counteract conditioned inattentional blindness
Make system messages shorter and stronger. Add color and an icon to improve signposting.
Flatten hierarchies
Users don't understand deep hierarchical structure in Help. Simplify.
Remove surprises
Take out jumps from Help topics (DHTML is OK). Add text to Help links that makes it clear when a browser window is about to be launched.
Use a normal user vocabulary
All users seem to benefit from Controlled English (also known as simplified English). Get rid of the jargon and describe functionality (features) and goals (tasks) in the users terms, rather than by referring to the GUI.
References
Leah Guren is the owner/operator of Cow TC. She has been active in the field of technical communication since 1980 as a writer, manager, Help author, and consultant. She now devotes her time to consulting and teaching courses and seminars in technical communication, primarily in Israel and Europe, and to usability consulting. Her usability work focuses heavily on cultural and linguistic issues, including her research on BDBL (bidirectional bilingual) website content. Her clients include some of the top hi-tech companies internationally, including Intel, IBM, and Microsoft.
If you have any questions please contact: leah@cowtc.com
