Session 3

User Diversity and Interaction Styles


Environment, Technology, and Society

Exercise: User Diversity and Interaction Styles


You have begun to explore how users can be frustrated by the design of everyday things and how they form a conceptual model of the things they use. You have read through Page 17 of The Design of Everyday Things.


  1. To consider the unique and common features brought to a task by different kinds of users.
  2. To consider the relative benefits and costs of different interaction styles.
  3. To begin to explore the interaction between user diversity and interaction style.


Break off into teams of three or four people based on the number in the righthand corner of this page.

  1. A well-known book on design for users classifies users into three generic categories: (1) novice, first-time users, (2) knowledgeable intermittent users, and (3) expert, frequent users.

    Pick something people use. Describe a user of this item for each category above. For example, you might consider a car and think of examples of novice, knowledgeable, and expert users of cars. Next, think of a kind of user who does not fit neatly into any of these categories. Define a fourth category of user based on your idea and list some features of this kind of user.

  2. The same well-known book classifies how users use a thing into several generic categories: (1) direct manipulation of the item, (2) menu selection, (3) data entry, and (4) command language. We use a car by direct manipulation, many microwaves by menu selection, an ATM by a mixture of menu selection and data entry, and many computer systems by command language. These categories are called interaction styles.

    Discuss the similarities and differences among these interaction styles. Identify at least two examples of each style from your experiences as users.

  3. Discuss the appropriateness of each interaction style from Task 2 for each of the four categories of user from Task 1. For example, do you think that novice, first-time users are able to use menu systems effectively? Be sure to list reasons that one style is or is not appropriate for a particular kind of user.

    Hint: There are 16 different combinations to consider, so you may want to divide them up among the three or four of you and then discuss your solutions tgether at the end.

At the End

  1. You possess a page or two that summarizes your group's conclusions for all three tasks.
  2. Each group will present some of its conclusions.

Summary from the Exercise

The well-known book referred to in Tasks 1 and 2 is Designing the User Interface, by Ben Shneiderman. Shneiderman's focus is on the design of computer-based systems, but the theory that he presents draws on studies of users and usability more generally. I will occasionally refer to his ideas throughout the semester.

Shneiderman's categories of users represent points in a multi-dimensional space. What are the dimensions? Can you identify other dimensions relevant to the problem of classifying software users? Answering these questions may help you identify other relevant classes of user.

We could look at his categories as three items from a 2x2 classification. One dimension is frequency of use (infrequent or frequent), and a second is level of knowledge (novice or expert). The "missing" category is a frequent user who remains a novice. Is that even possible? Well, some users (children, perhaps older folks, people with some illness) may have insufficient enough long-term memories that they remain in this category, despite frequent use. Might we want to treat them differently?

The 'knowledge' dimension could actually refer to two kinds of knowledge: knowledge of the tool or knowledge of the task domain. I think that Shneiderman is referring to the former, but what role does the second kind of knowledge play in user interaction with a program? How might a domain expert/tool novice interact differently with a system?

The notion of task domain is important. Think about the relationship between a writer and a word processor. Whenever I ask my computer science students to design an interface for a word processor, many invariably comment that the assignment is difficult because the the tool is so "simple". But I think the real problem is that, as writers and frequent users themselves, they have great knowledge of the domain -- and the creators of word processors have already satisfied most of their users' needs. Doing something different than Microsoft Word is difficult. (This leads to a later topic this semester: standardization of tools, whether by rule or by inertia.)

Would it make sense for us to have as a design goal to create a tool that a domain expert/tool novice can use immediately?

A common fourth category suggested by students is "drone", the application user "expert" in a very narrow set of uses but otherwise "clueless". How does this person differ from a novice in ways that affect how we might want to design the user interface?

Does any interaction style work well for every class of user? Direct manipulation may well. But then why don't we use DM as the primary interaction style for every system? (Good metaphors are hard to come by...)

A fifth interaction style is "natural language", a style in which users speak their 'natural' language -- say, English or Spanish -- to control the system. We do not know how to create natural language interfaces very well yet. Off the top of my head, I can think of only a few systems that use natural language interaction, those telephone-based, and then only with *very* narrow parameters. For example, when making a call from a pay phone and charging the call somewhere else, many companies now ask the caller to "say 'collect', 'calling card, or 'credit card'". Can you think of any?

We can contrast "natural language" with the artificial languages that we ordinarily use to communicate with computers (command languages for operating systems, programming languages, etc.) What trade-offs do we make between the use of natural and artificial languages? Natural language opens up effective communication to many more people, but ...

Students Summaries

All four of these summaries have something good to say. I especially like the second summary -- not for its length, but for its thoughtfulness and discussion of how the topic relates to the student's life.

Eugene Wallingford ==== ==== September 5, 2003