Jeffries, R., Desurvire, H.W.: Usability Testing vs. Heuristic Evaluation: Was There a Contest? ACM SIGCHI Bulletin, 24(4), 39-41 (1992)
This article focuses on why usability testing is still needed. Recently, many people have been trying to say usability testing is not needed anymore due to the fact that heuristic evaluation (inspection method) can find problems quicker and cheaper when compared together. This article’s goal is to provide valid information on why usability testing is still needed.
This paper is a combination of much of the authors previous work. They are trying to show people, who have misunderstood their research, what they are actually trying to say, not what others think they are trying to say. They combine articles they have written as well as collect information from people who have interpreted their research incorrectly.
They found people are interpreting their previous research incorrectly. The authors wonder why people are trying to discount usability testing, and why people think only one evaluation technique is needed. They argue heuristic evaluation, cognitive walk-through, etc. are all good and well, but under no circumstances should they be the only thing.
They then go on to list some results from heuristic evaluation and usability tests, and the results are as expected. The heuristic evaluations found some errors the usability tests missed and vice versa. The other fact they bring up is all the heuristic evaluation were conducted by experts, meaning they knew what to look for. The experts did an expert job (see what I did there) while less capable evaluators missed a lot and did not do favorably. This makes an even stronger case for why usability testing should be used; no experts needed.
I overall liked this article. I know it may not be a full blown research article, but it combines facts from many previously published articles. It seems evident to me usability testing is a must, or at least a try very hard to get done. Lastly, the authors state that more that one validation technique should be used, and I agree with them. Can every error be found by one person going through/using one site/product? I think not.