|
Article Excerpt
INTRODUCTION
The term wearable has been defined by Gemperle, Kasabach, Stivoric, Bauer, and Martin (1998) as implying the use of the human body as a support for some product. Extending this definition, Gemperle et al. coined the term dynamic wearability, which includes the notion of the device being wearable while the body is in motion. After generating 13 design guidelines for wearable computers, Gemperle et al. tested their design of wearable forms by asking participants to carry out activities and rate their level of comfort. By doing so, they suggested that an assessment of wearable computers should include an element of comfort analysis.
Comfort has been assessed in ergonomics in numerous areas linked with workplace design, including chair comfort (Shackel, Chidsy, & Shipley, 1969), thermal comfort (Nicol, Humphreys, Sykes, & Roaf, 1995), and visual comfort (Saito, Taptagaporn, Sotoyama, & Suzuki, 1993).
Musculoskeletal discomfort has been ad dressed by a number of studies. Using a range of techniques, including the Borg CR-10 scale, visual analogue scales, and body maps, studies have related neck and arm posture to discomfort (Chaffin, 1973; Harms-Ringdahl & Ekholm, 1986; Wikel, Chaffin, & Langolf, 1989). Levels of postural discomfort have often been related to workplace design and to interaction with technologies such as desktop and laptop computers (Hunting, Laubli, & Grandjean, 1981; Sauter, Schleifer, & Knutson, 1991; Sommerich, Starr, Smith, & Shivers, 2002).
Adding load to the body has been addressed with respect to levels of discomfort induced by loading specific joint structures (Harms-Ringdahl et al., 1986), general load carriage using items such as backpacks (Knapik, Harman, & Reynolds, 1996), and using hand tools (Putz-Anderson & Galinsky, 1993: Wiker et al., 1989).
Specific to wearing items, comfort assessment has been carried out for personal protective equipment such as respirators, shoes, gloves, glasses, coveralls, and knee and elbow pads (Akbar-Khanzadeh, Bisesi, & Rivas, 1995). For wearable computer equipment, comfort assessment has been carried out for devices incorporated into helmets (Robinette, 1993; Robinette & Whitestone, 1994; Whitestone, 1993) and for arm-worn devices (Stein, Ferrero, Hetfield, Quinn, & Krichever, 1998). Indeed, Knight (2002) measured levels of pain and discomfort attributable to variations of head and arm posture and added load using the Borg CR-10 for areas of the neck, shoulder, upper arm, and forearm, with specific reference to the use of wearable computers.
A limitation of these studies is that they often scored comfort along one scale, implying that comfort is a one-dimensional construct. When wearing something, one's level of comfort can be affected by, a number of things, such as the device's size and weight, how it affects movement, and pain, whether direct (e.g., friction, knocking, heat) or indirect (e.g., muscle fatigue). In addition to physical factors, comfort may be affected by psychological responses such as embarrassment. Therefore, simply knowing that when wearing the device the wearer has a certain level of discomfort does not help in determining what aspect of the device makes the wearer feel uncomfortable. Consequently, it is proposed that comfort should be measured across a number of dimensions. The aim of this study was to determine the factors that affect the comfort of wearable devices and then to produce a tool to measure them. The rest of this paper is therefore split into two main sections. The first explains the process taken to develop a set of comfort rating scales. The second section presents two studies undertaken to test these scales.
DEVELOPMENT OF COMFORT RATING SCALES
Descriptors of Wearable Comfort
To develop an understanding of what constructs are inherent in wearable computer comfort, we generated a list of descriptor terms that could be applied to wearable computers. The terms were generated using a three-stage brainstorming process: First, the concept of "wearing" something was used to generate specific terms; second, each term generated in the first stage was used as the basis for generating further terms; third, the resultant set of terms was distributed among colleagues for comment and possible addition. Initially, the terms generated were not specific to wearing some electrical or mechanical device but instead referred to wearing any object on the body. Subsequently, terms were generated with reference to wearing an electrical or mechanical device (e.g., wristwatch, personal stereo, wearable computer). Finally, consideration was given to wearable devices that have the ability to record some aspect of the wearers' condition, such as physiological recordings (e.g., heart rate, blood pressure, temperature). Carrying out this procedure generated 92 wearable comfort terms.
The comfort terms are shown in the Appendix, and each term is given a definition explaining how it describes wearing something affecting the wearer. At an initial level, asking the wearer to score the device for each of the 92 terms could execute an analysis of the comfort of a wearable device. This was considered here to be inappropriate, however, as it would be too labor intensive and time consuming for the wearer. Therefore, having generated 92 comfort terms, in the next stage we needed to reduce the number to a more manageable level.
Study 1: Reduction of Comfort Descriptors
To reduce the number of descriptors it was necessary to group the terms into categories of similarity, in which the terms in the category defined a homogenous aspect of comfort.
Method for the reduction of comfort descriptors. Eight people (age 37 [+ or -] 16 years) participated in categorizing the descriptors. Five of the participants were postgraduate students at the University of Birmingham, 2 were retired schoolteachers, and I was a manager in the health service. Using participants with different occupations was considered to be appropriate to produce results that would not be biased because of any conditioning that might be present in a group of people with similar backgrounds. Each comfort descriptor term, along with its definition, was printed on a separate card to produce a pack of 92 cards. These cards were given to each participant, who was asked to group the terms with respect to having similar meanings or dealing with similar concepts by arranging them into piles on a desk. The participants were given the freedom to group the terms in any order they deemed appropriate, into as many groups as they thought necessary, and to take as long as they wished. It turned out that the categorizing sessions took between 30 and 120 min.
To get an overall picture of how the comfort descriptors relate to one another, we produced data from the categorizing tasks by generating a descriptor-by-descriptor similarity matrix. The matrix was filled in by counting the number of times each descriptor was placed in the same group as another descriptor. For example, 4 participants...
|
