Page 397 -

13.5 Data collection, analysis, and interpretation 387

13.5.1 DATA COLLECTION
Physiological data collection presents some challenges that are not generally encoun-
tered in more traditional HCI research. To make use of the data sources that literally
measure the body, researchers must be in direct physical contact with their subjects.
For galvanic skin response or blood-volume measurements, this may be as simple as
placing an electrode on a finger tip. Surface electrodes (for EKG or EMG) and chest-
mounted sensors (for respiration measurements) are substantially more complicated.
These electrodes must be attached carefully in the appropriate position to ensure
high-quality recording of the desired data.
Measurements based on body-mounted sensors involving pressure (Brady
et al., 2005) or skeletal positioning (Dunne et al., 2006b) present a different set of
challenges. As these approaches are relatively new and the technology is rapidly
evolving, off-the-shelf tools with clear guidance may be few and far between. You
may need to familiarize yourself with the pros and cons of a variety of sensors
before conducting this sort of work. Before using any of these tools for measuring
physiological data, you should make sure that you have appropriate training in their
use. Partnering with an experienced health professional is an attractive means of
ensuring correct use of sensors and other—probably expensive—equipment.
Although electrodes and sensors are not physically invasive, they may cause some
discomfort and unease for some participants in your study. You may want to take ex-
tra care to be sensitive to participant's concerns, particularly involving the placement
and attachment of electrodes. Some researchers suggest that electrodes should be
attached only by someone of the same gender as the participant, in order to reduce
anxiety and embarrassment (Stern et al., 2001). As some participants may become
uncomfortable, your informed consent forms (Chapter 15) should be particularly
explicit regarding potential risks. Take extra care to observe the participants' moods:
when faced with a particularly distressed subject, you may wish to remind them
that they can withdraw if they are uncomfortable. In addition to being considerate,
this approach may save you from difficulties in data interpretation: if a participant's
anxiety levels are high due to concern about the experiment, it may be difficult or
impossible to identify anxiety responses caused by your stimuli.
These logistical challenges are even greater for more invasive techniques that
require the involvement of a trained expert. Although surface electrodes are widely
used in EMG measurements, needles placed in muscles are a possible alternative
for many applications (Raez et al., 2006). Although the needles are safe, they must
be used correctly, making them a strictly “don't try this at home” proposition. HCI
researchers have shied away from this approach (Mandryk and Inkpen, 2004); unless
your team has an experienced EMG professional, you would be well-advised to do
so as well.
Even if you are not using needles or electrodes, more prosaic restrictions might
apply. Eye-tracking devices might require that users be seated within an optimal dis-
tance range from the monitor, wired sensors might have limited ranges, and external
distractions must be controlled to minimize confounding stimuli that might distract
users and add unwanted cognitive load.