UI Design Newsletter – October, 2005

One of these things is not like the others

If you're a human factors/usability practitioner like me, you were probably "raised" on the core attributes of usability:

• effectiveness
• efficiency
• ease of learning
• memorability
• error handling
• aesthetic satisfaction

And if you're like me, you may have treated that last one – aesthetics – as somewhat of an afterthought, covered quickly with a few preference-oriented, rating-scale questions, or dismissed with a comment like, "We'll get to the visual design later. First let’s focus on how it works."

Hold that response

There is a lot of talk about beauty in the HCI community (e.g., Norman 2004). However, research tying aesthetics to design is still rather new. (e.g., Kurosu and Kashimura, 1995; Lavie and Tractinsky, 1997; Tractinsky, Katz and Ikar, 2000; Wilson, 2002). These studies demonstrate the "what is beautiful is good" stereotype. That is, they show that the beauty of a product can influence the users' overall impression or general user satisfaction of the product. Think iPod. But how do you measure that?

One of the challenges in measuring the product-emotion relationship is that users are not good at articulating their emotional responses, at least not in a way that's consistent across users. To deal with this challenge design researchers have developed standard measures to help users express their range of emotion by selecting pictures rather than picking words. One such example is Desmet's (in press) Product Emotion Measurement (PrEmo). When using PrEmo, participants' pick the cartoon-like expression that matches their experience. The cartoons dramatically simplify the user's task of representing an abstract emotional response.

There are other similar approaches – such as mood boards – sometimes used in the design space.

Being beautiful helps

Hassenzahl (2002) suggests that users develop an assessment of a product by bundling impressions about various attributes to make up a "product character." The attributes (and the weights of attributes) in the bundle include impressions about the product interaction, the content, the consumers' personal expectations or standards for quality, and, of course, impressions about the physical characteristics of the product. "Product characters" evaluations are very personal. For example, while one user may construe a Web site layout as new or novel, another user may perceive it as amateurish.

Hassenzahl's model separates attributes into two different types: Pragmatic Attributes and Hedonic Attributes. Usability professionals are familiar with the concept of pragmatic attributes. Pragmatic attributes are linked to getting things done. Hedonic attributes are related to how the object reflects the user's (perception of his/her) self. For instance, the Hedonic quality of Identification describes how important it is for users to express themselves through the objects around them and how much energy they focus on selecting objects to promote their "personal brand."

Try this question:

Would you prefer to ride:
(a) a Harley Davidson
(b) a Japanese Racing Motorcycle
(c) it doesn't really matter

If you answered either A or B – adamantly – Hedonic Identification qualities are important to you – at least with respect to motorcycles. (We are oversimplifying a little to make a point.)

Hassenzahl identifies a second hedonic quality – Stimulation – that describes the feeling of novelty and challenge within an experience. Stimulation (in slightly different terms) has previously been identified as important in evolutionary psychology, aesthetic philosophy, and theories of experiential flow. The oversimplified distinction here is: Would you prefer to do a puzzle you have done a hundred times and can solve easily, or one that challenges you to think in new and novel ways?

Hassenzahl captures users' perception of the relation of Pragmatic, Hedonic and Overall Impressions for products using a familiar, bi-polar anchor Likert rating scale approach, as shown below in both English translation and the original German anchors:

In early studies using these rating scales, Hassenzahl found that both hedonic attributes and perceived pragmatic attributes correlate with users' ratings of overall appeal of a product.

When beauty IS only skin deep...

In subsequent studies, Hassenzahl (2004) explored how beauty continues to influence perceived "Goodness" (or not) when people have direct experience with an interface. Here Hassenzahl presented a variety of MP3 skins for evaluation. In one experiment, he asked participants to rate the "Beauty" and "Goodness" of skins without any experience of using them. In a second study, he asked participants with experience using the skins to rate "Beauty" and "Goodness."

These studies are important because they explore how the relationships between beauty, usability and overall perception change after consumers actually use the product. In Hassenzahl terms, they seek to tease apart "Beauty" and "Goodness."

Hassenzahl found that when participants only looked at the skins:

• Beauty was influenced only by Hedonic-Identification ratings
• Goodness was influenced by Hedonic Attributes and (anticipated) Pragmatic ratings

That is, when participants only looked at skins, ratings of goodness enfolded both Hedonic and Pragmatic perceptions.

In contrast, when participants used the skin:

• Beauty was influenced only by Hedonic-Identification ratings – remaining constant
• Goodness was more influenced by (actual) Pragmatic ratings reflecting experience with getting the tasks done

Hassenzahl concluded that ratings of beauty are hedonically driven. Perceived beauty is an independent pragmatic attribute, and largely doesn't change over time. To wit: Just because I still can't answer my Bang & Olfsen phone without thinking, doesn't mean it's less beautiful – even after 4 years. Completely unusable things can still be beautiful.

However, Hassenzahl's studies separate beauty from goodness. In contrast to beauty, goodness is influenced by hedonic qualities only initially. Over time the influence of the perceived hedonic attributes on goodness may fade in importance relative to the quality of the consumer experience – particularly if the interface is hard to use. That is, a beautiful interface that seemed like a good idea at the time becomes less appealing if it's not usable.

Beauty. What is it good for?

Hassenzahl's studies suggest that the emotional aspects of the design are important in attracting customers in the first place. Hedonic properties around beauty clearly influence first impressions. However, when getting stuff done matters, perceived usability – judged through usage over time – is what matters most.

References

Benedek, J., Miner, T. (2002). Measuring Usability: New Methods for Evaluating Desirability in a Usability Lab Setting. Presented at UPA conference, July 2002.

Csikszentmihalyi, M. (1991). Flow: The Psychology of Optimal Experience. Harper Perennial.

Desmet, P. (in press). Measuring Emotions. Development and application of an instrument to measure emotional responses to products. Published on personal Web site.

Hassenzahl, M. (2000). Hedonic and Ergonomic Quality Aspects Determine a Software’s Appeal. Presented at ACM CHI conference, April 2000.

Hassenzahl, M. (2004). The Interplay of Beauty, Goodness and Usability in Interactive Products. Human-Computer Interaction, Vol. 19, pp. 319-349.

Marcus, A. (2003). The Emotion Commotion. ACM Interactions, November-December, pp. 29-34.

Norman, D. A. (2004). Emotional Design: Why We Love (or Hate) Everyday Things. New York: Basic Books.

Tractinsky, N. (1997). Aesthetics and Apparent Usability: Empirically Assessing Cultural and Methodological Issues. Presented at ACM CHI, March 1997.

It is nice to be able to measure facets of emotion and we have done that for ages. But we are weak in DESIGNING for emotion. It is good to make the connection that the brand/application requires "Exciting" and the design elicits "Excitement." In particular, we are happy to reject designs that are failing in this regard (we've done that a lot). But, what we really need are models that tell us how to design for "Excitement."

What does this MEAN for interface design? There are a few models that help (c.f., Flow: The Psychology of Optimal Experience by Mihaly Csikszentmihalyi). But our field is just beginning to create the predictive and prescriptive models we really need. We need fundamental models (like the concept of "flow") for how to design for emotion. Then we need to understand how these differ in mechanism between groups of different age, experience, role, and culture.