Recent Research on Minimalism


June 2004

Recent Research on Minimalism

CIDMIconNewsletter Hans van der Meij, Faculty of Educational Design, Management, and Media, Twente University

John Carroll and I first met when we attended the Document Design conference in 1992 at Twente University. By that time I had been engaged in research on minimalism for almost two years, after having moved from Leiden University to Twente University, where I started a new research program on the design of user support.

Right from the beginning of my work in this area, I was attracted by the basic ideas of minimalism as laid out in Carroll’s 1988 paper. I shared his conviction that use and user-centeredness are key factors in designing effective user support. My PhD student at that time, Ard Lazonder, and I read all available papers on minimalism, as well as Carroll’s book The Nurnberg Funnel: Designing Minimalist Instruction For Practical Computer Skill (MIT 1990). These texts further reinforced my motivation to adopt a minimalist design approach in the research program.

But I also had uneasy feelings about being capable of replicating Carroll’s work, which consisted primarily of descriptions of learning strategies and some design principles. What I needed were guidelines or more detailed descriptions to set up a systematic research program for selecting and manipulating the instructional strategies that constitute minimalism as a design approach. This need prompted a close examination of Carroll’s original minimal manual. Ard and I spelled it out from cover to cover. Our reverse engineering effort led to the rediscovery of four main design principles and a considerable number of heuristics.

When we finally pitted our minimal manual against a best-selling commercial version, the results were astounding. Being used to finding only small effects of experimental manipulations, I could not believe the huge effects favoring the minimal manual. We even checked the data twice to be sure that there had been no mistakes in our calculations. It was then that Carroll and I began our conversation, which has continued ever since and which has led to joint publications on minimalist design characteristics and misconceptions about minimalism. In this discussion, I will use Carroll’s 1998 volume as my point of departure for a brief exposition on minimalism.

Beyond the Nurnberg Funnel

The first book on minimalism was mainly based on the work of Carroll and his colleagues at IBM. In contrast, the second book, Minimalism Beyond the Nurnberg Funnel, was a volume edited by Carroll (MIT 1998), with contributions from researchers and practitioners who had not participated in the initial takeoff of minimalism. The book addresses many important questions about minimalism, including the three questions posed here.

Question 1: Minimalism lacks clear guidelines for the practitioner. Shouldn’t minimalism be more precise and more prescriptive?
The minimalist principles have been stated in various ways and with varying degrees of precision, but this has not deterred people from making minimalist manuals, even in minimalism’s early days. As I can attest from the work of my own students, and as Steve Draper also indicates in Minimalism Beyond the Nurnberg Funnel, the principles have been sufficiently clear to be successfully applied in practice. The work described in the book provides a great number of refinements compared to Carroll’s earlier papers. But, mind you, minimalism never aspired to be prescriptive. Just as it takes seriously the notions that learners need support but are responsible for their own learning and that learners will activate their prior knowledge and make mistakes as a consequence, so it takes seriously the idea that designers cannot and should not have their hands held every step of the way. Designers should be empowered; they should be informed and encouraged to give creating effective user support their own best shot.

Question 2: Minimalism has concentrated on user instructions for relatively simple domains. Can it be upscaled to more complex domains?
It definitely can, as several chapters in the book attest. The work of Hackos and Mirel, and Carroll’s own work on Smalltalk programming, is illustrative. But upscaling is a tricky word. It is not simply a matter of doing more of the same thing. More complex domains require different solutions to the new design challenges that inevitably emerge.

Question 3: Minimalism pays a considerable amount of attention to error handling. Is that really necessary? Or, conversely, should people be especially trained to handle problems?
People invariably encounter problems as they activate prior knowledge and skills to deal with the demands of a new set of tasks (for example, a new software program). Supporting error handling is, therefore, a design issue that any instructional designer should take very, very seriously, in my view. Indeed, one of the leading researchers in the field of instructional design (Merrill 2004) has only recently acknowledged this important point made by minimalism. The impact on user documentation has been disappointing, however. Minimalism has not sufficiently convinced the design community of the importance of error handling for designers to act accordingly and regularly present support for error handling in user documentation. Two inventory studies that I conducted, one in 1996 involving 60 manuals and one in 2003 on 104 manuals, showed that users are not informed adequately in this respect.

The answer to the second question is an emphatic “No.” Minimalism does not make a case for separate training in troubleshooting. People are not motivated to learn about troubleshooting outside the context of learning to use a software program.

Beyond Beyond the Nurnberg Funnel

Several interesting studies on minimalism have been completed after the appearance of Carroll’s 1998 book. One strand of research concerns the issue of transfer in minimalist documentation. The design issues studied here deal with screen captures, invitations to explore, and online help.

Research on screen captures initially began as an effort to support users in the joint management of an input device, a computer interface, and a manual (Van der Meij 2000, Gellevij 2002). It progressed into research on the roles of screen captures in supporting specific cognitive functions, such as the development of a mental model (which is assumed to afford transfer).

Research on invitations to explore has been conducted by several researchers (for example, Glasbeek 2001, 2004; Wiedenbeck, Zavala, and Nawyn 2000). Some of this research originates in the question about designing for individual differences in learning styles.

Yet another recent research effort examines the qualities of minimalism for designing online help systems (for example, Knabe 2000, Pratt 2000). In her research, Jean Pratt tested various minimalist design variations in a well-crafted empirical study.

Another strand of research concerns the application of minimalism to scaffold the use of existing interfaces. These studies have further examined the potential qualities of Carroll’s Training Wheels technology (for example, Bannert 2000, Leutner 2000).

In this article I focus on using screen captures and invitations to explore to enhance transfer (for a discussion on some of the other topics, see Van der Meij 2003).

Transfer in Minimalist Documentation

When designing training materials from a minimalist perspective, it is a priority to provide users with an immediate opportunity to act. This means giving less conceptual information and focusing more on getting users to practice basic tasks. This view is now widely accepted, but the important question of how to achieve knowledge transfer remains as crucial as ever. Presenting only instructions to act is insufficient to prepare users for coping with new tasks. They need to develop deeper insights, a mental model, for a successful transfer of knowledge. Recent research on minimalist documentation has examined the question of designing for such transfer in two ways. One design variation concerns the functions of screen captures. Another variation deals with the contribution of user explorations for developing deeper insights.

On screen captures
Studies on screen captures were extremely rare when I started my first explorations into their design and effects. Indeed, I found only one empirical study in the literature. My PhD student Mark Gellevij and I set up a research program on the topic. We started bottom-up by carefully examining screen captures in existing documentation, especially in visual manuals. This effort led to the creation of a taxonomy on screen captures, which formed the basis for a variety of empirical studies.

In one of these studies we examined the question of whether or not screen captures contribute to the development of a mental model. There are two distinct ways in which they are likely to do so (see Mayer and Gallini 1990), namely by explaining the structural layout of a user interface (system topology) and by illustrating visual flow (component behavior).

Screen captures can strengthen the presentation of conceptual information by showing and explaining important screen elements. Such pictures of the layout of a screen are fairly common in various kinds of user documentation. What seems to be important is that these pictures do more than label screen elements by their names. Their effectiveness is greatly enhanced when an explanation of their function is added. In other words, along with a label, there should be a brief description of what the user can do with a particular screen element.

Screen captures can also contribute to the development of a mental model when they are coupled with procedural information. User action and system reaction are coupled in this mode of presentation. The user gets a visual impression of how screens follow one after the other as he or she uses the software. In the documentation for Apple Macintosh, presenting screen captures along with user instructions has been used for many years (see Figure 1). During the last couple of years, the use of screen captures in manuals has increased dramatically, as a result of more companies following Apple’s example.

Hans screen cap b-w15

Figure 1. Screen Captures with Procedural Information

Apple was one of the first companies to exploit the potential of presenting a series of screen captures along with user instructions. Many things are good about the type of design shown in Figure 1. However, our research with Western users suggests that the placement of the two columns should be the other way around. That is, it works better to have the instructions presented in the left-hand column and the screen captures in the right-hand column (following the regular reading order for Western languages).

The main findings from our empirical study are that screen captures positively affect the users’ skills development and facilitate their completion of new tasks. In addition, we found a substantial effect on knowledge development. Not surprisingly, users did not develop a mental model quickly. They needed to work with the software and documentation for some time before we could detect these effects.

On user explorations
One of the reasons it is desirable to couple direct instructions to act with invitations to explore is that users are likely to explore anyway. Incorporating sections that invite users to explore a program is a way to keep users aboard who are prone to exploration. In addition, such sections can play an important role in the development of a mental model or schema.

User explorations should follow rather than precede or replace related instructions. In minimalist manuals, most invitations to explore appear as “on-your-own” sections at the end of a chapter. The idea is that users are likely to benefit from explorations only after they have developed some prior knowledge about a program or set of options. Only then is exploration likely to advance the user’s insights about a program.

A critical issue about invitations to explore is the question of how often users process these invitations. And when they do, does it advance their knowledge of a program? Wiedenbeck, Zavala, and Nawyn (2000) addressed these two questions in their study on explorations in minimal manuals.

Users received invitations to explore a program after processing related instructions. The explorations offered users a chance to acquaint themselves with additional features of the software. Three types of explorations were examined: exercises, “on your own” sections, and a combination of the two. Exercises gave specific instructions on what users could do. Placement and formatting distinguished these exercises from regular instructions. “On your own” sections consisted of questions prompting users to explore certain features of the software (for example, buttons, fields, and graphics). These sections also stood out from the instructions. The combined approach presented users with both types of invitations to explore.

Exercises had a much stronger impact on user actions than “on your own” sections. User compliance was very high. All users engaged in all or nearly all of the actions described in these exercises. In addition, they completed 88 percent of the exercises successfully. Findings for the combined approach were essentially the same as for the exercises. Both conditions contrasted sharply with the “on your own” sections, for which compliance was much lower. About 7 percent of the “on your own” sections were ignored completely, and, for another 34 percent, users explored only a part of the described features.

The findings from the study nicely complement Wiedenbeck’s earlier research on the learning effects of the three design variations (Wiedenbeck, Zila, and McConnell 1995). Users who completed exercises were superior in speed of execution and were more successful in completing transfer tasks than users left to explore on their own. Users working with a combined design had scores similar to those in the exercise group.

Another study on user explorations in minimalist manuals was conducted by Glasbeek (2001, 2004), who examined variations in task orientations and in invitations to explore. Task orientations were induced by informing the participants of a goal for consulting a manual about a spreadsheet program. Two types of task orientations were induced. Participants in the global task orientation were instructed to study the manual as best they could. In the specified task orientation, participants were to study the manual in order to create a gymnastics competition registration program.

Invitations to explore were given at the end of various chapters. Two variants of explorations were studied: exercises and “on your own” sections (see Figure 2). Exercises were designed around built-in training files. Users were asked to modify specific parts of a file (for example, to change the order of data presentation for group A, or to add names to class B). “On your own” sections also involved the use of built-in files. However, these sections merely encouraged users to try out general options (for example, try different formulas and functions).

Figure 2. Two Types of Explorations Used by Glasbeek

Exercise: Inserting Text and Data

The display below illustrates a part of a membership administration. List these data in a worksheet, and number the names as shown in the example.

Save the file under a name of your own choice.

On Your Own: Inserting Text and Data

Explore the possibilities of inserting text and data. Use the Delete key to correct mistakes. Use the arrow keys to move the cell selector.

Save the file under a name of your own choice.



To be paid

1. John Baz

2. Mike Smith

3. Kai Low

4. Erin Karg









Just as in Wiedenbeck’s study, users were found to engage significantly more often in exercises than in “on your own” sections. No main effects on users’ skills development were found, but there was an interaction effect of the experimental manipulations on user knowledge. Users appeared to learn more effectively with the “on your own” sections than with the exercises. The study showed that effects of task orientation and exploration types were intertwined. Only the right mixture of the two had a beneficial effect on the user. The effective combinations are: global task instructions and exercises, and specific task instructions followed by “on-your-own” sections.

The conclusion that can be drawn from these studies is that invitations to explore affect user actions. Users do not ignore these sections; rather, these sections stimulate users’ explorations of a system. As to the question about which format works best for the user, I think it is too early to tell. I think a better understanding is needed to fully appreciate the differences between design options and the conditions in which these options are to function.


According to some people, minimalism has become less controversial over the years. This assumes that minimalism has ever been controversial at all. For some of its ideas, perhaps this was the case (for example, don’t provide indexes). But controversial is, in my view, not an adequate description if one considers how well the core principles of minimalism have been received by the design community. It does seem to be true, however, that many of the once novel ideas of minimalism have gradually turned into an accepted practice that effectively supports the user. CIDMIconNewsletter

About the Author

Hans b-w13

June 0419