Page 21 -

6 CHAPTER 1 Introduction to HCI research

2000, versus 2017 might simply not be relevant. What would you compare? However,
a trend analysis over time might be useful, because there are some audiences for HCI
research, for whom trend analyses, over time, are considered a primary approach for
data collection (such as the CSCW researchers described in Section 1.6 and the policy-
makers described in Section 1.7). Furthermore, there are areas of HCI research where
longitudinal data would be both appropriate and very relevant. For instance, Kraut has
examined, over a 15-year period, how internet usage impacts psychological well-being,
and how the types of communication, and the trends, have changed over time (Kraut
and Burke, 2015). There are other similar longitudinal studies that are also very useful,
for instance, documenting that 65% of American adults use social networking tools in
2015, up from 7% in 2005 (Perrin, 2015), or documenting internet usage trends over a
15 year period (Perrin and Duggan, 2015). One could easily imagine other longitudinal
studies that would be useful, such as how much “screentime” someone spends each
day, over a 20 year period. The lack of longitudinal research studies in HCI, is a real
shortcoming, and in some cases, limits the value that communities outside of computer
science, place on our research.
Another reason why HCI research is complex is that, for much of the research,
not just any human being is appropriate for taking part as a participant. For instance,
a practice in many areas of research, is simply to recruit college students to partici-
pate in the research. This would certainly be appropriate if the focus of the research
is on college students. Or this potentially could be appropriate if the focus of the
research is on something like motor performance (in which the main factors are
age and physiological factors). However, for much of HCI research, there is a focus
on the users, tasks, and environments, which means that not only must the users be
representative in terms of age, educational experience, and technical experience, but
also in terms of the task domain (it is often said that you must “know thy user”). For
instance, that means that to study interfaces designed for lawyers, you must actually
have practicing lawyers taking part in the research. It will take time to recruit them,
and they will need to be paid appropriately for their participation in a research study.
Perhaps it is possible, although not ideal, to substitute law students in limited phases
of the research, but you would still need to have actual practicing lawyers, with the
right task domain knowledge, taking part in the research at the most critical phases.
Recruitment of participants is much more complex than just “find some people,” and
it can be actually quite complex and take a fair amount of time. For someone coming
from a background of, say, sociology, the number of participants involved in HCI
studies can seem small, and the focus may be different (strict random sampling in
sociology, versus representativeness in HCI). But our goals are also different: in HCI,
we are primarily trying to study interfaces, and how people interact with interfaces,
we are not primarily studying people, so we don’t always necessarily have to claim
representativeness.
Despite historic roots in the early 1980s, only in the last 10–15 years or so
have individuals been able to graduate from universities with a degree that is titled
“Human-Computer Interaction” (and the number of people with such a degree is still
incredibly small). Many people in the field of HCI may have degrees in computer