What to Consider When Designing a Biometric Study

In collaboration with Hannah Postings.

Biometrics can be a valuable addition to most research protocols, providing support for effects observed in both performance-based and self-reported data. Such metrics are unique because they provide insight into the autonomic biological processes of a user, often reflecting an implicit change to their cognitive state. Although this insight is often valuable, planning for any physiologically-based research protocol should include careful consideration of both the research plan and interpretation of data. The question is: What to consider when designing a biometric study?

As a researcher it is important to consider different factors when designing a biometric study:

Remember that biometric data is high-level

Biometric data is based on physiological processes that have been scientifically linked to certain psychological constructs, most commonly arousal, stress, or anxiety. Although manipulations in a study are often the cause of biometric differences, this is not always the case. The reality is that a wide variety of user individual differences can underlie effects observed in biometric data. These can arise from some fairly common sources, such as:

Chronic diseases: conditions, such as asthma, generalized anxiety disorder, depression, heart disease, diabetes, etc. will augment almost any biometric measures across all conditions, increasing the chance of conclusion errors
Prescription drug usage: use of medications such as such as alpha-2-agonists, amphetamines, or benzodiazepines can heavily influence some common biometrics, such as heart rate variability or skin conductance
General fitness level: variations in physical health can result in more efficient cardiovascular activity as well as greater metabolic rates or vice versa. In turn, there can cause abnormal heart rate, temperature, cortisol, and GSR data. These variations can undermine significant effects if outliers are not controlled
Dietary habits: users that have eaten within an hour or two of testing can demonstrate heightened levels of metabolic activity, spiking cortisol levels and body heat. Similarly, drinking coffee within a few hours of testing can cause errors in heart rate and GSR measurement

It is important to identify what level of control your needs may dictate

If you want your data to be more reliable, adequate screening measures must be taken but this can lengthen the amount of time it takes to complete the study
If less control is necessary, then screen requirements can be relaxed and total study time will be lessened but the data should be taken with a grain of salt

Time-of-day effect and scheduling

Some commonly used biometric measures, such as heart rate and cortisol, have demonstrated circadian variation, i.e., differences in measurement dependent on time of day
Cortisol tends to spike early in the day and drop significantly during evening hours, making enzyme-based stress measurement difficult in early morning sessions or after working hours
Heart rate variations follows a predictable pattern as well, peaking around in the late morning and evening hours while dropping in the early afternoon
To ensure that the biometric data obtained it of good quality, efforts should be made to accommodate testing around time frames that work best for the data you wish to capture

Cortisol measures should be obtained between meal times and later in the morning or afternoon, avoiding the midday hours when possible
Heart rate measurement should be taken in the early morning, early afternoon, or evening hours when possible

Take more conservative estimates of total time with studies including biometric data: allow some extra time so that participants can be scheduled during times of low circadian influence

If testing multiple treatments, leave some “cool down” time

If you’re testing multiple options that have the potential to be irritating, frustrating, or stressful, consider allowing additional experimental time to have participants “cool down”
Biometric data is implicit, which means that participants can’t fake it but it also means that they can’t control it
If the first condition(s) are stress-inducing, it is very possible that physiological data will remain heightened in the subsequent treatments
Blocking off just a little bit of extra time for participants to reset can improve the data obtained significantly

Biometrics is a tactful approach to get a deeper insight into the user’s cognitive mindset. It opens up a new era of usability testing. It is important to think through the study design, and what factors need to be considered to conduct a smooth study. We hope you find these three points helpful while designing a Biometric study.