Beautification

First, I would like to thank my supervisors, Dr Brenda Lobb, Dr Beryl Plimmer and Dr Doug Elliffe. Brenda, you helped open up my mind to see (and feel) answers from different angles and be a critical reader, thinker and doer; most importantly, you pulled me up and pushed me forward when I lost faith in my work and myself – thank you. Beryl, thank you for making me believe in the computer science side of me again. Without your encouragement and support at the beginning of this project, I would not have made it through the programming days in the lab. Thank you Doug, for helping me refine the experiment and data collection procedures; and for clearing up the grey clouds in my mind on results analyzes.

I would also like to thank my family: Mum and Dad, for giving me life in the first place and for your unconditional love and support, I love you; and Susan, my sister, my best-friend, for being a good listener and advisor, and thank you for being true, I love you too. Also my gym buddy and dining buddy, Iris, thanks for making me laugh when I don’t smile, for keeping me positive, and for your encouragement throughout the year.

Table of Contents

Chapter 1. Introduction

With this in mind, designers’ interaction with design tools/mediums and design formality (as a result of beautification) were examined in the present study. The main purposes of this study were to use an experimental approach to: 1) further explore the concept of beautification in the context of sketch-based design tools by examining the dimensionality of beautification; and also 2) to investigate levels of formality of designs, from rough, (hand-drawn, non-beautified) sketches to formal (beautified) diagrams, and their effects of on design performance during early stages of the design process.

1. 
   Design Research: The design process
   

• 
  Design epistemology – study of ways of knowing of designers
  
• 
  Design praxiology – study of practices and processes of design
  
• 
  Design phenomenology – study of the form and configuration of artifacts.
  

2. 
   Design process as problem-solving
   

1.2.1.1. Expert vs. novice designers

Expertise in design has been viewed by many (e.g. Cross, 1999) as an important factor that affects designers’ performance in the design process. One of the major approaches towards exploring the concept of expertise within the context of problem-solving (design) is to compare experts and novices. According to Cross (2004), novice behaviour is usually associated with a ‘depth-first’ approach to problem solving, i.e. sequentially identifying and exploring sub-solutions in depth, whereas the strategies of experts are usually regarded as being predominantly ‘top-down’ and ‘breadth-first’ approaches. Many of the classic studies of expertise have been based on examples of game-playing (e.g. chess), or on comparisons of experts versus novices in solving routine problems (e.g. mathematics and physics). For example, early chess studies carried out by De Groot (1946/1965) showed that instead of having superior information processing capabilities, ‘experts’ players (grand masters) ultimately chose better moves than ‘novice’ (highly skilled) players; and that skill level was linked to the amount of information remembered (i.e. better chunking). The results from Chase and Simon’s (1973) study, also conducted on chess players, suggested that experts not only processes more knowledge about their domain of expertise, but their knowledge was organized in more meaningful and readily accessible ways. These studies provided a strong the basis for later studies on problem solving as it showed that skill depended at least in part on the acquisition of domain knowledge and stimulated a vast amount of research on the nature of expert problem solving and the relationship between knowledge and skill. However, one must point out that those studies used well-defined problems, whereas designers characteristically deal with ill-defined problems (Cross, 2004; Romer, Leinert, & Sachse, 2000), therefore findings may not be generalizable to the design process.

There are some empirical studies specifically on design cognition, amongst which have been studies on expert, or experienced designers, and comparisons of the processes of novice and expert designers. The recent focus of such studies in design cognition has been through the use of protocol analysis studies (Cross, Christiaans & Dorst, 1996). Similar to the findings in Christiaans and Dorst’s (1992) analyses of the design protocol of junior and senior industrial design students, Atman et al. (1999) found from their in-depth protocol studies of twenty-six freshman engineering students (first year) and twenty-four (fourth year) engineering students as they designed a playground for a fictitious neighborhood, that there was a difference in terms of design performance and behaviour between senior students and freshman in the design process. The results showed that the seniors produced higher quality designs than freshman; in addition, the seniors gathered more information, considered more alternative solutions, moved more frequently between design steps and progressed further into the final steps of the design process. Atman et al. (1999) suggested that the design processes the seniors learned in their four years contributed to the differences found. Overall, it was an in-depth study, with carefully controlled variables, as well as observations, video recordings and meticulously defined elements in the protocol analysis, which in turn, enabled rich data to be collected and analyzed.

As problem solving occurs over time, and the notion that experts become expert through extensive practice – “practice makes perfect” (Green, & Gilhooly, 2005), follow-up and longitudinal studies provided a richer understanding of expertise in design (verses novice) over time. Adams, Turns and Atman (2003) conducted a longitudinal study on the behaviours of freshman verses senior students during design. As the authors commented, changes in individual students’ behaviours over time can be quite complex and variable – as one would predict. Definite change was noticeable for some, however, for some of the students, their behaviours did not appear to change at all and some simply spent more time on design projects but without any qualitative behavioural changes. It also appeared that students exhibited different behavioural changes for different types of design projects.

Similarly, Atman et al. (2005) conducted a follow-up study to their 1999 research, in which verbal protocols was, again, collected and freshman and senior engineering design processes to open-ended design problems was compared. The difference from the earlier study was the increase in the number of subjects who participated: sixty-one seniors (fourth year) engineering students and thirty-two freshman (first year) engineering students; as they worked on two design problems (as opposed to one design problem in the previous study). In addition, the study also included protocols for eighteen within-subjects participants, who participated in the study first as freshmen and later as seniors, which provided the rich data sources for the comparison of design process changes over time on the individual student level. In terms of between-subject differences in design process changes, results showed that seniors produced higher quality solutions, spent more time solving the problem, considered more alternative solutions and made more transitions between design steps than the freshmen to a greater extent, but comparable to results in Atman et al. (1999). Similar to the findings for the general population of freshmen and seniors, the within-subjects participants as a group showed differences between freshman and senior participation, however, some participants did not exhibit growth on all measures. Also, comparable to Adam, Tums and Atman’s findings in 2003, individual students’ design behaviour varied by problem (i.e. was task dependent). Again, like the study in 1999, this study was highly controlled and used coded variables to support the analysis and interpretation from the rich data obtained.

However, results from the longitudinal study should be interpreted with caution as the operational definition of change and the criteria for behavioural changes were ambiguous. As the measure of behaviours over time can be complex and variable, one must bear in mind that such study could not be conclusive with respect to expertise and behavioural changes over time. Hence, in addition, the comparison of results found in Adams, et al (2003) and Atman, et al. (2005) should be examined critically as the two studies measured different aspects (design process as a whole verses behaviours during design) and had different problem solving tasks over a different period of time.

1. 
   Individual differences in problem-solving
   

According to Lawson (1979) problem solving strategies can be categorized as either ‘problem focused’ or ‘solution focused’, in which in a recent empirical study on design protocol by Kruger and Cross (2006), were both found to be used by the nine experienced industrial designers performing the same task, along with some sub-variants (‘information driven’ and ‘knowledge driven’ respectively – see Kruger and Cross (2006) for a more detailed description of the different cognitive strategies). Clear individual differences between designers were found in most of the data relating to both design process solution outcome, even though the designers performed the same tasks under the same conditions. Kruger and Cross also found that most designers employed either problem driven or a solution driven design strategy, and each were equally prevalent. However, contrary to Lawson’s earlier claim in 1979, solution driven design was not used as the dominant strategy. Furthermore, the different strategies used were related to the overall solution quality (task outcomes) in a complex way. Designers who employed a solution driven strategy was found to have lower overall solution quality, but higher creativity scores. Where as, designers using a problem driven design strategy tended to produce the best results in terms of the balance of both overall solution quality and creativity. Overall, it was a rich descriptive study on design strategies and design performance and outcome. However, it must be pointed out that, similar to many other studies examining cognitive processes, the experiments were conducted as ‘think-aloud’ protocol studies (van Someren et al., 1994) i.e. designers were required to think aloud as they were solving the problem – this may have affected designers’ performance in detrimental manner, because problem-solving is a cognitive activity which requires attention, and by talking out loud (attention on talking), it may distract and disturb the flow of the problem-solving process.

Other studies examined individual differences in problem-solving with a slightly different focus – styles of problem solving (as opposed to strategies used) and their influence on the design process. For example, in Eisentraut’s (1999) empirical study, fifteen engineering design students with different levels of study (varying from two to 14 semesters) and hence, different knowledge of design methodology, worked on three problems – two computer simulated complex problems (non-design context) and the one adaptive design problem. However, the participants were allowed five hours to finish the allocated tasks, therefore, it was likely that participants’ concentration fluctuated, and became fatigued near the end of the experiment. Like Kruger and Cross (2006), the ‘think aloud’ approach was also used, which may have affected participants’ design performance, measured in terms of design quality, especially with such long experimental period. A single case approach was used for data analysis and it was found individual styles of problem solving determined the way in which designers organized their design processes. It was observed that subjects exhibited stable problem solving behaviour (e.g. amount of information gathered, speed of action). It was also noted that individuals who worked on the three situations in the same way may have been successful in one situation, but less successful in another, thus, Eisentraut concluded that the success in problem solving depends on whether a style of problem solving meets the demands of the situation. Eisentraut further suggested that diagnosing and training individual problem-solving behaviour may play an important role in optimizing individual design processes and therefore should be a part of design education

Similarly, Eisentraut and Gunther (1997) also empirically examined individual styles of problem solving, but with a specific focus on their relation to representations in the design process. Single case approach was chosen and analyses focused on the marks-on-paper which were created and used by the designers (fifteen male engineering students who worked on an adaptive design problem, as well as on two complex non-design problems). The results suggested that the course of the design process in general and the use of marks-on-paper in particular depended on an individual designer's style of problem solving. However, participants’ background was not adequately described, and furthermore, only four among fifteen cases were presented, therefore the generalizability of the findings is questionable, and thus, results must interpreted with caution. Nonetheless, Eisentraut and Gunther noted that drawings by hand and sketches turned out to be important to the subjects during their design processes. It was also noted that more than 70% of the documents produced in the experiments (illustration from the sketch in principle to the sketch roughly to scale) cannot be created by a CAD-system. Eisentraut and Gunther’s study yielded some important implications which are also relevant to the present study, for example, as suggested, CAD-training should not replace drawing and sketching in design education, and that the use of the ‘concreteness’ and ‘completeness’ (of sketches used by designers) as a tool in teaching the levels of abstraction in the design process. Furthermore, because of the importance of sketches – a notion also agreed by many other researchers (e.g. Goel, 1996; Goldschmidt, 2003) and supported by empirical evidence (McGown, Green & Rogers, 1998; Schenk, 1991 Ullman, Wood & Craig, 1990), it was suggested that CAD development should be directed towards systems that enable the designer not only to create formal technical drawings, but also to easily make sketches of different degrees of abstraction.

Although empirically studies have been conducted, more research is still needed to develop a more comprehensive model that explains on individual differences in problem solving in design more thoroughly.

1.1.2.2. Other factors affecting design performances

In addition to factors including expertise and individual difference in problem solving, there are other factors that also affect the design process in the context of problem solving. From the organizational and industrial psychology point of view, the physical environment, for example work station design (Grandjean, Hunting & Pidermann, 1983) and lighting (Stammerjohn, Smith & Cohen, 1981) in which an individual works in is an important determinant in work (design) performance – hence, impacts on the design process. The social context in which designers works in, for example individually or in a group, also influences the design process (e.g. Goldschmidt, 1995; Gross et al., 1998; Lahti, Seitamaa-Hakkarainen & Hakkarainen, 2004; Launis, Vuori & Lehtela, 1996; Stempfle & Badke-Schaub, 2002). Time and resource constraints in design situations are as recognized as major factors that shape the design process, in terms of designers’ motivation and decision making, and on design outcome (Burns & Vicente, 2000; Savage, 1998). In addition, research has also shown that the use of different design tools may influence a designer’s attitude towards the design process. For example, Hanna and Barber (2001) found that in addition to the difference in the design approach when using computer-aided design (CAD) tools only as opposed to using both sketching and computer-aided design tools, students’ attitudes towards the design process changed after using CAD. Furthermore, different (combination of) design tools used have been shown to affect the design process, namely, design activities (Holt, 1991; Lahti et al., 2004; Shniderman et al., 2006) and design performance (Black, 1990; Jonson, 2005; Sachse, Leinert & Hacker, 2001).

Overall, much research has been done on the design process and with so many pieces of information grouped differently by different researchers in various disciplines, one must bear in mind that to date, although attempts have been made to integrate existing research into a more comprehensive model of the design process (e.g. Elias, 2005 – cognitive model), there has not yet been a model that describes and explains the design process as a whole in a systematic way. Before such model can be developed and tested empirically, further understanding of the design process, particularly the designer within it and his/her interaction with artifacts (design tools/mediums), is required.

1. 
   Human-Computer-Interaction (HCI) and Design
   

Human-computer interaction is a multidisciplinary field that incorporates research and theories from computer science, psychology, engineering, anthropology, education, design, mathematics, and even physics. Researcher have identified the fundamental aims in HCI as to take a scientist-practitioner stance, to understand humans and their interactions with computers, and the roles and impact of computers on humans; and furthermore, to explore design techniques and methodologies with descriptive and/or prescriptive approaches, as well as usability testing and evaluation; and thus, to design effective products that are suitable for a wide range of human requirements and capacities (Benyon, et al., 1993; Czerwinski, 2003; May, 2001)

Much of the research in HCI to date is concerned with design guidelines and standards (e.g. Molich & Nielsen, 1990; Nielsen, 1994); effective design for diverse users (e.g. Czaja, 200; Joiner, et al., 1998; Sear, et al., 2001); understanding of psychological phenomena (cognition, perception, and action) in humans’ interaction with computers (e.g. Reason, 1990; Neisser, 1976; Pancer, George, & Gebotys, 1992); as well as the study of design methodologies including design tools techniques (e.g. Beaudouin-Lafon & Mackay, 2003); and usability studies (e.g. Dumas & Redish, 1999, Lewis, 1998) . However, in comparison to such research areas, it appears that the importance of research on the design process has been largely neglected within the HCI discipline.

With the rapid technological development in design tools such as computer-aided design (CAD), however, it was not until the last decade that design research became an increasing interest to researchers in the HCI field. In 1991, a whole issue in the Journal of Human-Computer Interaction was dedicated to the topic of design rationale i.e. the reasons and the reasoning processes behind the design and specification of artifacts (products), with a particular orientation to particular domain, namely computer and information systems in the issue (e.g. Carroll & Rosson, 1991; Maclean, et al., 1991). A few years later, in 1994, another issue of the journal was dedicated to design research, this time with a focus on design context as it could open up new approaches and ways of thinking for designers (Moran, 1994). However, it was only after more than a decade, in 2006, that another special issue on the topic of design appeared again in the Human-Computer Interaction (HCI) journal. It was pointed out by Carroll (2006) that design is appropriately one of the core topics of the HCI journal; however, as commented, there has always been a lack of papers on design in the journal. Carroll also argued strongly that design is the most fundamental topic in human-computer interaction because, “whatever understanding we may achieve of human capabilities and preferences, the social and cultural construction of activity, and the gamut of technological possibilities and constraints, we still have to put it together in order to have any effect on the world. The artifacts we design – infrastructures, systems and applications, policies and curricula – are the most important results of our endeavors”, (p. 2). In other words, it is the actual designers who design artifacts; and artifacts are all a result/product of design – by a designer. Furthermore, even with all the required resources (information and tools) available, whether a product is successful, in terms of quality and usability, is highly dependent on the designer him/herself (and the design team); hence – the importance of studying the designer-artifacts (i.e. design tools/media) interaction within the design process.

2. 
   Prototypes, Prototyping and Prototyping tools
   

Prototypes can be seen as a design tool and a prototype may be defined differently in different design disciplines (e.g. in engineering, architecture and fashion design). In the HCI context, in which this study focuses on, a prototype is defined and described, by Beaudouin-Lafon and Mackay (2003), as a “concrete representation of part or all of an interactive system. A prototype is a tangible artifact, not an abstract description that requires interpretation. Designers, as well as managers, developers, customers, and end users, can use these artifacts to envision and reflect on the final system” (p.1007). Not only is prototyping a design process itself, prototyping is a also a social process, as Helander (1988) discussed, and will provide the means by which that designer can make his or her ideas explicit to the community and by which these ideas can be evaluated. Studies have found that the use of prototypes can help refine a product’s functional requirements very early in the design process (e.g. Yang, 2005).

While research has shown the benefits and problems associated with using prototypes (see Alavi, 1984), prototypes have long been used in practical settings. In Houde and Hilll’s (1997) study, it was suggested that prototype increases creativity, allow early evaluation of design ideas help designers think through and solve design problems, and support communication within multidisciplinary design teams. Also, a few studies suggested that a development process based on prototyping during the early stages of a major design project can significantly reduce the cost of the final software, typically found in the early development phase. For example, in an early study, Boehm, Gray, and Seewaldt (1984) found in a comparative experiment using computer science students that a prototyping based design approach required approximately 45% less development time than a more structured approach. Furthermore, the use of prototype is important as gives an early picture of a design concept which can range from simplistic two-dimensional sketches that represent design thinking and ideas (Ullman et al., 1990; Goel, 1995; Suwa and Tversky, 1997) to more advance and sophisticated 3-dimensional mock-ups that are hardly distinguishable from the real manufactured product. Hance, prototypes can be different in nature and is dependent on the phase of design. Ullman (2003) described four categories of prototypes based on their function and stage in product development: 1) proof-of-concept prototype – used in the initial stages of design to better understand what approach to take in designing a product; 2) proof-of-product prototype – used later in design to clarify a design’s physical aspects and production feasibility; 3) proof-of-process prototype – which shows that the desired product could be successfully produced with the method and materials used; and 4) proof-of-production prototype – which demonstrates that the full manufacturing process is effective and operational.

Furthermore, according to Beaudouin-Lafon and Mackay (2003), prototypes and prototyping techniques can be analyzed in four dimensions: 1) representation – the form of the prototype (e.g. sets of paper sketches or computer simulations); 2) precision – the level of detail at which the prototype is to be evaluated (e.g. informal and rough or formal and highly polished); 3) interactivity – the extent to which the user can accurately interact with the prototype (e.g. watch only or fully interactive); and 4) evolution – expected life cycle of the prototype (e.g. throw away or iterative). For the purpose of the present study, the first two dimensions (representation and precision) of prototypes are more relevant and will be mainly discussed.

As noted, prototypes aid designers in creating concrete representations of design ideas and clarifying specific design directions, and also give designers (and others) an early glimpse into how the new system will look and feel (Beudouin-Lafon & Mackay, 2001). Moreover, prototypes serve different purposes, and thus take different forms, for example, series of quick rough sketches and a detailed computer simulation, all of which are of help to the designer in different ways. Hence, Houde and Hilll (1997) suggested that the designer must consider the purpose of the prototype at each stage of the design process and choose the representation that is best suited to the current design situation. Different levels of representations of prototypes can be achieved – for example, during the early stages in the software design process, paper prototype such as paper sketches are often used, as they can be created quickly, at a low cost, in comparison to using software prototypes such as interactive interfaces usually used in later stages in the design process, which is usually higher in cost and may require skilled programmers to implement advanced interaction and/or visualization techniques or to meet tight constraints (Beaudouin-Lafon & Mackay, 2003).

Moreover, prototypes vary in terms of precision – the relevance of details (content) with respect to the purpose of the prototype (Beaudouin-Lafon & Mackay, 2003). For example, when sketching a dialogue box (e.g. online (web) forms that requires user interaction by input), the designer specifies elements size, positions of each field, and titles of each label. The concept of precision is similar to the terms low-fidelity and high fidelity of prototypes used in the literature to refer to the degree of relationship to the final system. In addition to the theoretical and methodological research on prototypes, it would be beneficial for future research to also explore designers’ interaction with different levels of representation, (e.g. media, precision and formality) of prototypes, and whether such factors may influence design performance and design outcome.

There are different techniques for prototyping. Some examples of prototyping techniques include mock-ups, Wizard-of-Oz, storyboard (see Beudouin-Lafon & Mackay, 2001). It can be noted that such techniques uses different types of tools – for example, paper and pencil (e.g. for prototyping a web interface), cardboard mock-ups (scaled three-dimensional prototype) for a future building, video prototyping for story boards of a web site. On the other hand, other prototyping tools involve software – for example, Micromedia Director, Adobe Illustrator, Adobe Photoshop, Microsoft Publisher to create non-interactive stimulations; Visual Basic.Net, and other programming languages such as Java, C++, C#, to create interactive stimulations (e.g. user interface). However, as with any tools in general, every tool has its strength and weaknesses, even in the various design (prototyping) tools that are commonly used. Although one can say that design tools serve to aid designers in the design process, however, it is also important that a designer uses, in addition to the appropriate level of representation, appropriate design tools at different stages in the design process and nature of the design task. The combination of useful design tools employed at appropriate stages helps optimize the design process in terms of efficiency, design quality and outcome. Hence, in the following section, the two main tools/media used for prototyping (designing) and their effects on the designers during the design process are discussed and compared.
1.4.1. Traditional Design tools for Prototyping

1.4.1.1. Paper and Pen(cil)

Paper and pencil are one the most commonly recommended set of tools to use in the design process (e.g. Beaudouin-Lafon & Mackay, 2001; Brink, et al, 2002; Jacko & Sears, 2003). Not only are paper and pencil inexpensive in comparison to other tools such as computers and software, it also provides a medium for designers to freely explore different design ideas through sketching (Newman et al, 2003). According to many, sketching is one of the most important design activities as it facilitates reasoning (Akin, 1986; Goel, 1996; Goldschmidt, 1991, 1994; 2003; Tversky, 1999), problem-solving (Romer, Leinert, &Sachse, 2000), memory and thinking (see discussion by Scaife & Rogers, 1996), creativity (e.g. Goel, 1996, Goldschmidt, 2003; Hanna & Barber, 2001; Kavakli, Scrivener, & Ball, 1998; Nakakoji, Tanaka, Fallman, 2006; Schenk, 1991; Verstijnen, et al, 1998) and communication (Beudouin-Lafon & Mackay, 2001; Schenk, 1991; Van der Lugt, 2005); all of which are important aspects of design that affect design outcome (see review on drawing and the design process in Purcell & Gero, 1998). In the context of Web site design, Beudouin-Lafon & Mackay (2001) also agreed on the importance of sketching on paper as it assists exploration of layout, content and aesthetics of a design, and in addition, it is less likely that sketching on paper would constrain designers’ thinking compared to using development environments, software and programming languages. Such usefulness of paper and pencil as a design tool to support sketching is reflected in a recent Web design practice study, in which Newman et al (2003) found that paper prototypes (rougher, hand-drawn representation of the final product) are frequently employed during the design process, especially at the early stages.

Beudouin-Lafon and Mackay (2001) argued, however, that although paper prototypes have many advantages, they are not the answer to everything. In some situations, paper prototypes are insufficient to fully evaluate a particular design idea. For example, where interface requiring rapid feedback to users (e.g. Web site design), or complex, dynamic visualization (e.g. engineering and architectural design usually require software prototypes, created by using computer-aided design tools.

The technological advances which improved accessibility and lowered cost of the use of computers was one of the major forces for the dramatic increase in research on computer-aided design (CAD) tools in the past two decades. This had a major impact on design representation and the design process, for example, influencing tool choices at different stages, and faster and more complex design creation and evaluation in a dynamic way. CAD tools assist architects, engineers and other design professionals in their design activities, generally during the later stages in the design process, to create accurate precision drawings and technical illustrations, hence, increase in fidelity and formality of prototypes (a closer and more precise representation of the final design). As CAD enables version control, editing and easy distribution (e.g. emailing) of designs, which has its advantages over using paper and pen, the general argument for using CAD is to improve design productivity, lower product development cost and shorten design cycle (Jacko, & Sears, 2003). Additionally, Lahti, Seitamaa-Hakkarainen and Hakkarainen (2004) found that computer-supported environment helped facilitate collaboration between individuals in a group. In contrast, it is often claimed that the CAD system itself interferes with the design work (Landauer, 1996; Luczak & Springer, 1997), especially during conceptualization work (Lawson & Loke, 1997), as using CAD requires additional cognitive workload, for example, skills to manipulate the computer objects on the screen with the input devices and knowledge of commands to operate the system and input drawing data (Khalid, 2001). However, Hamade, Artail and Jaber, (2005) found that design behavior and performance using CAD tools can be increased through effective CAD training, especially in novice users. Such findings may suggest that manipulation and operation of the system (procedural knowledge) becomes automated, which in turn may help decrease mental workload. Nonetheless, research on CAD tools has shown the usefulness of such tools.

Despite of the opportunities and challenges that these CAD tools offer, their impact on the design process has not been fully examined through empirical experimentation. Nonetheless, it can be seen from a wider literature search, that theoretical and/or descriptive work on CAD (e.g. Grekas & Frangopoulos, 2000, Tovey, 1997) is beginning to bloom and mature, with an increasing number of empirically driven studies on user-interaction with CAD tools (e.g. Hamade, Artail & Jaber, 2005; Khalid, 2001, Sachse, Leinert, & Hacker, 2001).
1.4.1.3. Combination of Paper and Pen, and CAD

It is often that both pen and paper, and CAD tools are used in conjunction, and are viewed as important tools in the design process as their strengths and weaknesses are intertwined (Newman et al, 2003). According to Plimmer and Apperley (2001), designers traditionally engage in sketching on paper and pencil, and ultimately move to computer-based tools. Researchers have become increasingly interested in the non-traditional design process in the digital age, and recent studies have compared the use of CAD tools and sketching (drawing) on paper and pencil in different ways, including designers’ behaviors, design outcome, design quality, and the design process, by using common methodologies such as design protocol analysis – traditionally used studying human information processing (problem solving) mechanisms (Newell, 1968). For example, Bilda, Gero, and Purcell (2006) conducted think-aloud experiments with six expert architects (graduate students with similar design knowledge and experience) to examine whether sketching is essential for conceptual designing, based on a protocol analysis. Bilda et al. found no significant difference between sketching (using paper and pencil) and not sketching (using CAD) based on design outcome, cognitive activity and idea links, which led Bilda et al. to conclude that sketching is not an essential activity for expert architects in the early phases of conceptual designing. However, with extremely small number of subjects used in the study, the generalizability of the results to professional architectural design situations is questionable. Furthermore, thinking-aloud during the experiment may have disrupted the natural flow of design activities, thus, results may have been polarized (hence, the number of subjects).

In an experimental study with more subjects (seventy four undergraduates), Sachse, Leinert and Hacker (2001), too, examined the use of computer and sketches during design and found that sketching before and/or during the computer-aided design resulted in a significantly reduced time taken for more complex task analyzed, despite of the additional sketching time – this can be explained by the reduction of the number of processing steps needed when sketching was allowed. Sachse et al.’s findings further supported the perceived utility of sketching before and/or during the usage of CAD tools (e.g. Newman et al., 2003). Bilda and Demirkan (2003), on the other hand, took a slightly different stance, and aimed at gaining an insight on designers’ cognitive process while sketching in traditional verses digital media, by using retrospective protocol analysis. Their results showed that traditional media had advantages over the digital media, such as supporting the perception of visual-spatial features, and organizational aspects of the design, generation of alternative solutions and better conception of the design problem.

However, one remains concerned that the use of protocol analysis may not truly or fully represent information process in the design process, as it is highly task-specific, and may interfere with visual thinking, and distort the real design process (e.g. see discussions in Ericsson & Simon, 1984; Llyod, Lawson & Scott, 1995). Despite of critiques on protocol analysis, such studies yield valuable information about designers’ thought process that is for the development and improvement of computer aids in (architectural) design to support the conceptual phase of the design process.

Overall, although design education and design handbooks echoes the usefulness of sketching on paper during early stages of design along with CAD tools (for in later stages), they are often prescriptive in nature on the use of various design tools and derive from assumptions and experience of the authors, and many of them do not necessarily refer to such empirical evidence – which reflects the lack of conclusive research on the use of (a combination of) different design tools/ mediums and their effects on the design process. Moreover, it must be noted that although with their advantages and disadvantages, the use and effectiveness of various design tools are also dependent factors such as the design task (e.g. solution finding, development or refinement), design complexity (e.g. simple or complex), and design scope (e.g. a whole Web site or one Web page), as well as the designers’ very own personal preference and practice.
1.4.1.4. Paper prototypes verses Digital prototypes

In addition to the underlying effects of working with various design tools, one closely related factor that also influences design outcome, is design representation and its relative formality (on different media), created by using different design tools/media (e.g. paper and pencil, and computer-based).

According to Brinck et al. (2002), prototypes can vary from very course, fuzzy layouts of general page requirements done on paper (low-fidelity), to precise, highly elaborate, refined and polished digital versions of the web site (high-fidelity). This range provides the designer with levels of refinement and formality useful for testing and exploring varying details of a given design. For example, paper prototypes maybe used to gather feedbacks on basic functionality or visual layout in a quick and efficient way. In contrast, for example in Web design, digital prototypes may be the only accurate way to explore an online environment on issues such as colour and contrast.

There are some interesting differences among the perceptions conveyed via paper and digital mockups. According to Brinck et al. (2002), it is better to use paper mockups early in the design cycle for one major reason: clients tend to view paper mockups as a conceptual rather than a finished product; and that the strength of this cannot be overlooked. In addition, anecdotal evidence suggests that one of the biggest problems encountered with digital mockups is that clients tend to view them as final, unchangeable products, as opposed to paper mockups which are perceived as less polished and more conceptual – also voiced by design professionals interviewed by Newman, et al. (2003) in their study. Therefore, as these authors suggest, it can be seen that paper mockups tend to provoke more comments; where clients and users could become more open in their suggestions for change and the perceived inadequacies of the design (Beaudouin-Lafon & Mackay, 2003) – in this way, paper mockups can be viewed as an agent in generating useful feedback on boarder design issues. By contrast, clients viewing digital mockups tend to focus on the details of the layout and on issues such as font choice, exact spacing, label names, or colours; yet, it is often too soon to receive such advice in early stage in design (Brink, et al., 2002; Newman et al., 2003). Thus, digital mockups tend to be reserved for later in the prototyping process, nearer the finishing of the design. Hence, mixing both types of prototypes (low-fidelity and high-fidelity) at different times can be useful for revealing different types of problems (Brinck et al., 2002). This has an implication on the importance of design representation – in particular, formality level of designs (from informal, rough sketches to formal, computer-rendered representation of design) used at different stages during the design process.

Again, the descriptive and prescriptive use of paper prototypes and digital prototypes has been based mostly on anecdotal evidence from experts and professionals in the design field. Hence, more understanding of cognitive and perceptual processes when interacting with different types of prototypes may help the development of prototyping tools and techniques, and moreover, the understanding of design representation and its effects.

A deeper question yet to be answered, which arises, is whether designers will perceive designs differently when presented on paper compared digital media. In other words, whether the difference in media for representation would affect design performance and outcome (e.g. error detection and design quality). Another factor that further complicates such question is the introduction of design tools with sketch-based interfaces – a computer-aided tool with a pen and paper feel (i.e. ‘digital ink’).

3. 
   Current Trend in Design Tools Research: Informal Sketch-based interface
   

Along with strong anecdotal and empirical evidence on the effectiveness of traditional design tools (pen and paper, and computer-based tools), the improvements and popularity of tablet PC (pen-based input) also contributed to the significant increase in the demand and usage of such sketch-based interfaces in diverse software applications (Pomm & Werlen, 2004). This, in turn, has resulted in a major trend of research on sketching-based design tools with an emphasis on the advantages of sketch-based interfaces, as these “bridge the gap [between paper and pen, and computer-based design tools] by providing a design-friendly computer-supported sketching environment and add a useful new dimension to the design process” (p. 1337, Plimmer & Apperly, 2004). As opposed to “formal” design tools such as CAD and programming languages like Java, C++ and Visual Basic.Net, sketch-based design tools have been regarded as “informal” design tools (Landay & Myers, 2001) as they support the ambiguity and informality of sketching (Chung, Mirica, & Plimmer, 2005; Newman et al, 2003; Plimmer & Grundy, 2005). In the following sections, related research on computer-supported sketch system requirements is briefly described, followed by examples of current research on design, implementation and (further) development of such systems; and finally, ‘beautification’ – an important aspect (and functionality) of computer supported sketch-based tools – and its related research, is further discussed.

So far, overall, research has shown strong evidence supporting the notion that sketching (and its naturalness) plays an important role in design cognition and reasoning, and that designers typically work with paper and pencil first, then later, transfer the paper design onto formal computer-based tools (e.g. CAD software) – this formed the basis for research on the design and development of computer-based sketching interfaces to support the design process. For example, based on the importance of sketching in early stages of design, Dickinson et al. (2005) discussed the potential of pen-tablet interface in mechanical CAD modeling. Similarly, Plimmer and Apperly (2001) stressed on the importance of drawing and how current computer interfaces interferes with the sketching process; and further proposed requirements, in terms of functionality and usability, for an ideal computer-aided sketch system to capture preliminary design – i.e. an interface that facilitate direct, rapid drawing, while provide more functionality than paper or whiteboard. Based on previous experimental research, Mulet and Vidal (2006) also suggested similar functional requirements (i.e. sketch-supported interface) for improving on existing computer-based design support tools. On the other hand, Fitzmaurice et al. (1999) discussed design issues and requirements for rotating user interface (RUIs) to support artwork orientation, with a more specific focus on an integral part of drawing – the rotation of the piece of paper.

Many different types of ‘sketch systems’ have been implemented and developed as a result of various studies highlighting the benefits of building computer-supported sketch tools, and the associated design issues and requirements. Sketch systems that have been developed so far have different capacities ranging from two-dimensional pen-input to three-dimensional sketching and manipulation; as well as processing of sketched objects, for example, by (domain-specific) sketch recognition and beautification.

2. 
   2-Dimensional (2-D) sketch-based systems