Definitions

5

II. Literature review

In this section usability and culture are discussed, and relevant definitions and models are presented.

2.1 Usability

As a social construct, usability is a notion that is difficult to agree on. Definitions of usability typically attempt to identify the aspects of usability that encapsulate its meaning. Models of usability are based on these definitions but attempt to incorporate factors that may influence usability such as context variables or variables affecting the measurement of usability.

2.1.1 Definitions

One approach to defining usability is to examine how users themselves

understood the concept of usability. McGee et al. (2004) looked at how 42 users rated 64 usability characteristics. They found users defined usability as a combination of consistency, efficiency, ease of use, effectiveness, controllability, usefulness, expectability, and naturalness. Satisfaction and style were excluded. Similarly,

Hertzum et al. (2007) discovered that a user’s concept of usability also varies between cultures. Danish and Indian users described usability as including concepts such as

‘easy-to-use’, ‘intuitive’, and ‘liked’. On the other hand, Chinese users offered a range of issues related to security, task types, training, and system issues. Such studies have not resulted in accepted usability definitions, no doubt because by nature, such

usability definitions differ across cultures and between individuals, and cannot become a standard.

An additional attribute, aesthetics has been suggested. Tractinsky (1997) and Kuroso, M. and Kashimura, K. (1995) found significant correlations between user’s perceptions of beauty and perceptions of usability for Israeli (r=0.921) and Japanese (r

= 0.59) users respectively. Tractinksy therefore called for a more holistic

understanding of usability which includes aesthetics. This measure, however, is only intended as an addition to existing definitions of usability. In addition, it is possible that aesthetics may be similar to satisfaction, a commonly used attribute.

Shackel attempted to provide a standardized definition of usability. Shackel defined usability within a framework of user, task, tool and environment. (Shackel,

1991). Usability for individual users was determined by subjective ease of use and objective performance measures such as effectiveness, learnability, flexibility, ease of use, and user attitude. Shackel included a variety of possible measurement methods for these usability metrics such as number of errors, time taken to learn a task, and attitude measurement questionnaires. However, problems exist with implementation of some attributes. Learnability is not applicable to products that are used a single time only.

For example, many websites are encountered only once. In addition, the flexibility of a product may be difficult to measure.

Nielsen suggested similar attributes to describe usability. Nielsen’s definition of usability includes learnability, efficiency, memorability, errors, and satisfaction (Nielsen) However, Nielsen’s definition has been criticized for its blending of usability measurement methods, such as number of errors, and usability attributes, such as efficiency, which could be measured a number of ways (Winter, et al., 2008).

The Consolidated Model of Usability (Witold, et al., 2003) was based on a combination of usability attributes found in the ergonomics standard set by the

International Standards Organization - ISO9241 - and additional attributes considered important in other definitions. Witold et al. incorporate learnability from Shackel and Nielsens’ definitions, despite issues with single use products. In addition, the model incorporates security from the ISO9126 software quality standard. However, while security may be an important feature in software design, it is not essential in all products, while usability is.

ISO/IEC 9126-1 is a software quality standard which incorporates a definition of usability. The definition describes usability as, “the capability of the software product to be understood, learned, used and attractive to the user, when used under specified conditions” (ISO, 2001). This definition aims at defining user interface qualities to ensure overall software quality, and so is not intended to provide a standalone definition of usability. As an example, effectiveness is not included as an attribute of usability in ISO9126. This definition appears to be non-intuitive, as, to be usable, a product must at least do the job for which it was intended. However, effectiveness appears in ISO9126 as an attribute of software quality, supporting the aim of this standard, which is to define aspects of software quality, rather than usability.

On the other hand, ISO9241 is an ergonomics standard, widely accepted in the HCI (Human-Computer Interaction) community (Frandsen-Thorlacius, et al., 2009),

7

The standard defines usability as “the extent to which a product can be used by specified users to achieve specified goals with efficiency, effectiveness and user satisfaction in a specified context of use” (Bevan, 2001). ISO9241 further defines usability measures as follows.

 Effectiveness: the accuracy and completeness with which users achieve specified goals.

 Efficiency: the resources expended in relation to the accuracy and completeness with which users achieve goals.

 Satisfaction: the comfort and acceptability of use.

Of all usability definitions, ISO9241 most succinctly encapsulates actual product performance in terms of process and results, and perceptions of performance in terms of user satisfaction. For that reason and because of its widespread use in HCI

community it is adopted for use in this dissertation.

One problem with the ISO9241 definition is that there may be little correlation between usability attributes. Hornbæk and Law (2007) examined 73 published

usability studies and concluded that attributes showed little or no correlation with each other. This was supported by Frøkjær et al.’s work, which investigated the correlation between usability measures of a document retrieval system and found little evidence of a correlation between usability attributes (Frøkjær, et al., 2000). Walker et al. (Walker, et al., 1998) report a similar lack of correlation between usability measures in their study of a voice controlled email interface.

However, these findings have been challenged by the results of research by (Sauro and Lewis, 2009) who identified correlations between measures of usability in their analysis of the results of 90 unrelated usability tests. By examining actual

usability tests rather than academic studies, and by examining results at the task-level, rather than for a usability test as a whole, they were able to identify correlations ranging from -0.44 (errors vs. task satisfaction) to 0.6 (errors vs. time), all much higher than those identified by Hornbaek and Law or Frøkjær et al. They therefore argued that a usability construct exists, and that a single measure of usability is possible, based on usability metrics. However, their results also show such a measure would only account for 54% of variance in raw scores, which they acknowledge would result in the loss of a great deal of valuable information. Consequently, it is argued in this paper, that even if a single measure of usability were possible as Sauro and Lewis argue, the loss of information is too high to warrant combining measures, or to argue

that one usability measure sufficiently represents usability as a whole. Measurement of all usability attributes is still required to understand the effect of culture on usability.