Seminarium 2 - Marcus Dypbukt Källman

The literature for the second seminar examines the role of evaluation when designing a product. Evaluating your design is very important since it affirms that your design is attractive to your target group and to find problems that lowers the usability or hurts the users experience.  What to evaluate depends on the product and it’s important that the evaluation is implemented so benefits the final product. There are generally 3 broad categories of evaluation,

Controlled settings involving users – users’ activities are controlled in staged environments to test and measure how the product is used and to see if the users behave as imagined when interacting with the product. This allows for outside disturbance to be reduced but can lead to troubles capturing the context of use.

Natural Settings involving users – users’ activities are observed in the “real world” to find how the product is being used. Displays how the users would behave outside the test but are often more expensive and harder to conduct.

Any setting not involving users – the product is examined by experts or predicting and modeling how the interface will be used to identify its most obvious faults. This approach is often cheap and easy but can be unreliable since it can miss unpredictable problems.

All of these have different advantages and shed light on different problem areas and are therefore used in different cases. Often a combination of the three should be used to achieve a higher understanding of what to change with the design.

Evaluations not involving users have 4 different methods, heuristic evaluation, walkthroughs, analytics and predictive models. In heuristic evaluation and walkthroughs experts are used to provide feedback on the product. heuristic evaluation lets the expert evaluate the product by a set of pre-determined usability principles to check if the user-interface accommodates these principles. In walkthroughs a task is preformed and its problem areas assessed. Analytics tracks the user’s behavior to see what they do or want from the design. Predictive models are used to measure the user performance of different aspects of the product.

Since we’ve decided our main product is going to be a smartphone app, which of these type can we use to evaluate our design? and which could we realistically execute with our resources?