Two approaches to research on UX models and measures are discussed on basis of experiences from the field of usability research and an ongoing case of user involvement in software development (SD) by way of social media. It is suggested that simple measures and ad-hoc models, rather than complex models and measures, may be beneficial to the relevance of UX research for SD practice.
The complexity of usability models and measures is
beneficial for the rigor of usability research. However,
complexity may be detrimental to the relevance for
usability models in SD. For SD practitioners, complex
generic models may be impractical. The generic model may
include several components that seem irrelevant to the
development project at hand. Also, seemingly important
aspects may not be handled by the model. One may
speculate that the prevalence of homegrown usability
measures at the expense of standardized measures [
Sauro and Kindlund [
The position of Sauro and Kindlund is controversial [
By ad-hoc UX models I mean models developed in response to a given concept, prototype or running system. Instead of utilizing a general UX model as basis for a generic UX measure, relevant UX components may be established on basis of users’ responses.
As an exemplification of simple measures and ad-hoc models, I will briefly present an ongoing development case addressing new functionality for mobile phone e-mail clients. Design typically involves the “simultaneous investigation of multiple alternatives by the same designer or team” [12, p. 1243]. In the present case 22 ideas were generated and then refined as six early concepts; all across a working period of 60 hours. Following this, the concepts were made available for user feedback.
User feedback was collected through a social software application for sharing audio-visual content, modified for the purposes of design feedback. The participants were 212 regular users of e-mail-clients on mobile phones (use several times a week or more). They were presented for six concepts in sequence. For each concept they were asked to make a rating (1-5 stars) and one or more comments. The comments were made in response to open questions: How would you use the suggested function? How may the suggested function be improved? As part of a research design not to be detailed here, half of the participants were allowed to see the other participants’ comments prior to making their own and half of them were not. All participants were allowed to comment on each other’s comments. None were allowed to see the others’ ratings prior to making their own.
The user feedback provided the following key information: • •
Differentiation between the concepts: Three of the concepts were rated low, three rated high. The ratings corresponded closely to the frequency of positive vs. negative comments for each concept.
Establishment of concept specific issues that may serve as basis for an ad-hoc UX model: For one of the concepts, detailing functionality for ‘Postponed sending of messages’, relevant issues were mainly targeting utility. For another, ‘Reading aid for long e-mails on small screens’, relevant issues were lack of comfort and utility. For a third, ‘Context-dependent e-mail receipt’, relevant issues were privacy and utility.
The case thus illustrates that simple measures and ad-hoc models may serve as basis for choosing which design alternative to pursue, and enable the establishment of adhoc models that may be used to control the development process. In the present case, a UX model for the first concept would need to include only utility in addition to usability, whereas the UX models for the second and third concepts would also need to include comfort and privacy/trust respectively.
CONCLUSION By this paper, I hope to contribute to a discussion on how research on UX models and measurements should be approached in order to obtain relevance. I hold that complex models and methods will indeed be beneficial for the establishment of UX theory. However, research relevance may require a different approach. Possibly, such an approach may be simple measures and ad-hoc models. ACKNOWLEDGMENTS This paper was written as part of the research project RECORD (www.recordproject.org), supported by the VERDIKT programme of the Norwegian Research Council.