One of the final conference streams on Wednesday afternoon was a masterclass titled ‘Human factors engineering: the path to usable software design’, presented by Prof. Peter Elkin, of the Mayo Clinic, Rochester, Minnesota, USA. Peter set his presentation, and the resulting discussion, within the premise that software usability is vitally intertwined with good interface design, supported by usability testing with real end users, and that most health IT failures result from lack of respect for, or attention to, human-computer interaction, rather than necessarily technical flaws.

Peter covered some of the basic theory of human-computer interaction, beginnning with the three elements – humans, activity, context – within Bailey’s Human Performance Model. He discussed some of the reasons that software and other IT products are often hard to use, including their being designed with an emphasis on the machine or system, rather than the end-users, the target audience often being a moving target, lack of full design specification, and development teams not being well integrated. Too often, he said, engineers design machines to talk to machines, but they are not the end-users who must wrestle with the product. He stressed the need for user-centred design, including an early focus on the needs of users and the teasks they will perform, and empiric measurement of usage, often through usability testing laboratories (such as he runs).

He moved on to discuss some of the principles of human factors analysis, including contextual inquiry (understanding the end-user needs), competitive usability evaluations, and low-fidelity prototypes for ealry testing. He suggested that ‘the user is always right’ – although this may not always be the case – and that in terms of interface design, ‘less is more’, as each widget in an interface (eg screen) places an additional burden on the user. He suggested that usability of a system often depends on ‘minor’ interface details, and that without proper usability testing, these may not be uncovered, and so the reasons for resistance to use, for example, may not be found and corrected.

Peter discussed 5 usability attributes:

  • learnability;
  • efficiency;
  • memorability;
  • error prevention; and
  • satisfaction.

He devoted the last part of the masterclass to description of the theory and some examples of usability testing. He defined usability testing as replicable studies of products and processes in a controlled environment, and described 4 types:

  • exploratory;
  • assessment testing – often expanding on exploratory test findings;
  • validation testing – often occurring late in the development cycle; and
  • comparative testing, often of two products, processes or designs.

He described the limitations of usability testing, including it always being an artificial situation, difficulties in determining whether the test users are typical of the target population, and that testing may not always be the best technique. He recommended adherence to the relevant ISO stantards, including ISO 13407, the human-centred design development cycle.

This was a very useful and informative session that it would have benefitted many of a more technical or policy orientation to attend.

Technorati Tags: , ,

Listen to this article Listen to this article