Selecting Test Participants
Ideally, each test participant should belong to or be able to represent a portion of the audience of people who will eventually work with the product or system. It is usually helpful to create some sort of User Matrices (though likely not nearly as complex as this one for simple user studies) of potential user types to ensure you get good representation of the types of users you want to test among your test participants. It may also be helpful to give potential participants a screening survey, or even collect information on potential subjects including with their basic characteristics in a database, in order to ensure that you have a good set of representative users for your testing session.
You will likely have a good sense of the different types of users your product or service targets; if not you should make an educated guess. User groups can range from a small group of highly specialized people (say an accounting web application for a specific company) all the way to the general public for a government information site. The Gradebook on Moodle for example might have a user audience mainly of instructors and teaching assistants. uPortal would likely have a much larger group of students, faculty and staff as its user base.
How many users to test? Neilson suggests that testing 5 users, with a minimum of 3 from each user group, within an iterative design process should be sufficient to find most usability issues.
Recruiting Test Participants
Groups performing user testing will often have limited budgets and time, so you many need to be creative in your approach to recruiting users. If you aren't able to hire an agency that specializes in it to recruit your users, you may find it difficult to entice participants using emails and/or posting fliers alone.
One creative user testing method which can make recruiting participants easier is 'hallway' usability testing. For instance, in the university environment, if students are your users of interest you can often set up your user study in a cafeteria, library, commons room or hallway and select people as they walk by. Faculty and staff users may be more difficult to find but again, you might think of posting information about your user testing sessions in areas where faculty and staff meet, work, or eat.
Even if you recruit users as they walk by in a busy setting, it is best to conduct the test in a quiet area free of distractions. Try to find a conference room nearby, perform tests one at a time, and assiduously follow a protocol. Unless you would like to see how participants interact with each other, you will usually test people individually.