Methodological Issues

The review of research in the field uncovered several methodological issues that may also be involved in masking effects along the aesthetic process.� One such aspect concerns evaluations that are more nuanced than overall aesthetic evaluations, which are quite common in the studies surveyed in this chapter. Studies that look for evaluations of aesthetic sub dimensions (e.g., Kim et al., 2003; Lavie and Tractinsky, 2004; Moshagen and Thielsch, 2010) can potentially yield richer accounts of the influence of design on aesthetic processes and on subsequent evaluations of the interactive system, attitudes towards it and interactions with it.

A related issue deals with the measurement of visual aesthetics evaluations or judgment. In the field of HCI, aesthetic evaluations were measured by a single item and by multiple-item scales. For example, Kurosu & Kashimura (1995) Tractinsky (1997), Schenkman & Jonsson (2000), Hassenzahl & Monk (2010), and Sonderegger and Sauer (2010) have used a single item asking about the beauty of the various applications and interactive products. Others, such as Schenkman & Jonsson (2000), Van der Heijden (2003),Moshagen et al., (2009) employed multiple-item scales to measure attractiveness. While multiple item scales are generally regarded as more reliable measures, single item scales have some practical advantages. In general, the main advantage if that single items make questionnaires shorter, reducing participants’ fatigue and tendency to skip some of the items. In particular, the use of single items in the study of aesthetics allows quicker responses to stimuli in studies that focus on swift aesthetic responses (Lindgaard et al., 2006; Lindgaard et al., 2011; Tractinsky et al., 2006). The tension between scientific directives and practical constraints may not be as severe as it first appears. Studies suggest that when dealing with a concrete object (e.g., the application or product to be evaluated) and a concrete attribute of the object then single item measures are as valid as multiple-item scales (Gardner et al., 1998; Bergkvist & Rossiter, 2007). While the scientific community may have a hard time defining what is meant by the concepts of “aesthetics” or “beauty” � perhaps due to the multiple disciplines that deal with these concepts and which attach different meanings to them, my experience is that ordinary people’s intuitive interpretation of the terms correspond closely to the dictionary definition provided above, which guides research on visual beauty in HCI.� This point may be worth further research for corroboration, but if correct, future scientific and practical studies would be able to safely use single item measures of visual aesthetics.

Research on visual aesthetics in HCI has employed a mix of experimental and correlational designs. Because some of the most interesting aspects of visual aesthetics research involve questions of cause and effect, experimental studies would appear to provide the most conclusive evidence. It is straightforward to study basic and relatively simple design effects (e.g., symmetry using basic patterns) on aesthetic perceptions using experimental designs (Bauerly & Liu, 2006; Winkielman et al., 2006). However, it becomes increasingly more difficult if we want to study the effects of aesthetic design using more complex and ecologically valid stimuli, like those used in correlational studies (e.g., Lindgaard et al., 2006; Hassenzahl & Monk, 2010). Thus, employing experimental designs using elaborated and realistic stimuli is a major challenge. Ideally, to test causal effects studies would manipulate design attributes independently of each other to separate aesthetic perceptions from perceptions of other system attributes. In practice, however, this is very difficult to accomplish due to the a priori association of these attributes (Moshagen et al., 2009). One frequent consequence of attempting to achieve this independent aesthetic manipulation is that it creates a relatively small variance in the manipulated stimuli (otherwise, strong aesthetic manipulations might also cause differences in other experimental factors). The danger is that small variance and the lack of strong aesthetic condition would in turn weaken the effects of visual aesthetics. Another challenge in manipulating or selecting aesthetic stimuli in experimental designs relate to whether the degree of aesthetic stimuli is defined “on average” (e.g., by a pilot study or manipulation check) or is defined separately for each individually (e.g., in a procedure described byTractinsky et al., 2000). The advantage of the latter approach is improved probability that individuals who are assigned to various aesthetic groups in the experiment indeed perceived the stimulus in a way that corresponds to their group (as opposed to a stimulus that may belong to that group on average, but which doesn’t match the participant’s aesthetic taste). This would increase the effect size of the aesthetic manipulation. On the other hand, such a procedure usually requires pre-experimental exposure to a set of potential aesthetic stimuli. This process may later interact with the experiment (e.g., by creating expectations towards the experiment), and may create undesirable noise.

Дата добавления: 2015-11-16