Testing with Five Users

My interest in this paper was a 7, I had seen the name of this paper in Fast and Cheap User Testing.

The three main ideas of this paper were the importance of efficiency in the iterative design, that since users are different one user is not enough to test usability, and there are times when five users may not be enough. In order to have human and computer interaction there must be some human users to help detect problems with the usability of a new design. As a class learning about this interaction we have agreed that the use of users in developing a project is a necessity. This paper displays graphically that more insight is gained by five users because there are just enough people to still find new problems with the usability and not have too many repetitions of the exact same results. The use of a small test group also makes it easier to have cheaper models to use over and over again. This makes sure that resources are not wasted, thus improving efficiency. However, if there are very distinct groups of users, like the example of parents and children given in the paper, then a small group of about three users from each category is necessary for testing.

My question: What is the process to select a group of users that can give the best results?

My interest in this paper was, again, a 6 (I had to beat Nicole.) Like “Fast and Cheap User Testing” it like a few things to be wondered. Perhaps, it would have been useful to see there previous research, but I am skeptical of their 15-user curve and even more skeptical of their 5-user testing. Although they do mention it in their last paragraph, I think they do not account for diversity as much as they should. After discussing Microsoft Word in class and the variety of users it might have, I think that testing with only 5 users would not allow for thorough testing.

The three main ideas of this paper were 1) Iterative design is better than a single test with more users 2)5 users is the ideal test condition because it allows for variety but is inexpensive 2) You may need to test with more users if their a multiple group of distinct users. (“distinct” bothered me here, but they never really define what that means)

This paper relates to our class in the same way as “Fast and Cheap User Testing”: Iterative Design. However, this paper goes into greater detail as to why you should only use 5 users. It also seems to contradict class, in that it lumps users into one category making the assumption that 5 users will cover all types of people.

What would this author define as “distinct” user groups?