Page 471 -

15.3 Care and handling of research participants 463

Special populations may require creative incentives and accommodations. If you
are working with children, you might give them small toys as gifts for participating
(cash compensation for accompanying parents is probably always welcome). Elderly
people or others without easy access to transportation may be interested in partici-
pating but may be unable to make the trip to your lab or office. You might consider
trying to conduct your study in participants' homes, community centers, or other
locations that would be easy for interested participants to travel to.
Some studies may have additional requirements that require screening of interested
participants to determine whether or not they meet important criteria. For example,
tools designed for novices should probably not be evaluated by people who work pro-
fessionally with similar interfaces. Initial questions and interviews with potential sub-
jects can be important tools for ensuring that an individual is appropriate for your study.
Specific questions about education, age, experience, and other important attributes can
be asked to verify that there is indeed a good match. If you take this approach, you
might also consider asking whether they are willing to be contacted in the future for
subsequent studies. People who agree to future contact can form the basis for a home-
grown database of study participants. Maintaining such a database may involve a fair
amount of work, but it can be potentially very useful if you plan to run many studies.
Your database of potential subjects can be an important safety net in the event of
difficulties along the way. You may start out with 15 (or 20, 30, or 60) participants
with confirmed appointments, only to find that several cancel at the last minute or
simply fail to show up. Other problems associated with participant characteristics
may force you to dig deeper for a wider range of ages, skills, or backgrounds. If the
participants in your study of a general-purpose tool for managing personal photos
are all men between 35 and 40 years old (or women over 60), you might have a hard
time arguing that your results are indeed generalizable. It's easy to argue that better
planning and participant screening might help with this problem, but such details
are often not obvious from the beginning. If you're faced with this dilemma, your
best option might be to dig deeper into your list, inviting more participants to form a
larger (and hopefully more representative) study.
Experiments that involve multiple experimental conditions may require reanomiz-
ing participants into roughly equal-sized groups. If you are comparing performance
across user attributes—such as age or gender—your groups must differ in the relevant
attributes, while remaining as comparable as possible for other characteristics. If your
potential pool of participants is large, you need to select participants in a manner that
minimizes any potential bias in selection: selecting the first names from a list that is
sorted by gender may get you a group of subjects that is entirely male or female. See
Chapter 4 for more discussion of these and related issues in population sampling.

15.3 CARE AND HANDLING OF RESEARCH PARTICIPANTS

Studies with human participants put researchers in a privileged position. As “scien-
tific experts”, researchers have expertise, experience, and contextual knowledge that