Caveat on use of the Internet of Things in behavioral analytics

Big Data Evangelist, IBM

Participatory observation is a valuable data-gathering method in any research discipline that involves studying human behavior. That includes any business function that focuses on customer engagement—especially marketing, customer service and sales.

Participatory observation is the heart of ethnography, research designed to explore cultural phenomena. Ethnography was pioneered by anthropologists, but has been applied to other social sciences as well as to marketing and customer research in business. Ethnographic research is a valuable tool in the market researcher's methodological kit. It can act as a counterweight and a reality check to customer surveys, focus groups, social sentiment analysis and other established research techniques. Correlated with these and other customer data feeds, ethnographic findings can deeply enrich and contextualize your analysis of customer behaviors, experiences and desires.

But ethnography, like photography, depends on the point of view of the person capturing that snapshot. Ethnographic research can mean either of two things:

  1. That researcher is participating in the very activity in which they are observing the subjects.
  2. The subjects provide reports to the researcher on what exactly they are doing and what their perspective is on the activity while they're engaged in it.

How trustworthy is subject-reported data? At first glance, you might think that information issuing directly from your subject is the rock-bottom truth, but that would be naive. When your ethnographic customer research depends on asking subjects to self-report on their normal behaviors, you've already biased your findings. That's because even if people were totally honest, introspective and detailed in all their personal affairs, they unconsciously edit what they choose to note. If nothing else, people don't usually jot down the data while they're in the midst of whatever behavior you're studying. Doing so would interrupt the normal flow of that behavior, thereby essentially distorting the behavior that you're trying to observe in its natural state.

When people do get around to logging whatever it was they did, they succumb to the usual failings of reconstructed memory. They may unconsciously leave out details, change the sequence, falsely attribute to the present occasion some details that occurred in a similar occasion in the recent past and so forth. Subjectivity creeps into and distorts the observations at every step.

Wouldn't it be great if you could instrument the subjects of participatory research to automatically, continuously and objectively log their behaviors? Well, you can, thanks to the Internet of Things (IoT). As discussed in this recent article by Jake Sorofman, the IoT, via wearables and other portable endpoints, opens the way for people to participate 24x7 in self-reported ethnographic studies.

Though he has no case studies to feature, Sorofman nicely summarizes the feasibility and promise of this approach: "Consider what happens when you bring this now somewhat ancient idea of participatory observation to the digital domain. Here, sensors, quantified self and the Internet of Things become a source for new insights revealed through close observation of reflexive, utterly human patterns of behavior. As the digital and analog universe becomes wholly instrumented for measurement, this approach to digital ethnography will go mainstream and it will shine light on the truth like never before."

"The truth"? I'm not as sanguine as Sorofman that tapping into wearable-sourced behavioral-data feeds will provide 100 percent unbiased findings. Remember the "Hawthorne effect"? Noting that it's also called the "observer effect," Wikipedia describes it thusly: "A phenomenon whereby workers improve or modify an aspect of their behavior in response to the fact of change in their environment, rather than in response to the nature of the change itself."

Let's frame that in the present context. When you're studying behavior and foster awareness in the subjects that they are being observed—by others, via wearables, by the subjects themselves in cooperation with researchers or by other means—you have created the conditions for a Hawthorne effect. In concrete terms, people who participate in any quantified-self/wearable-sourced study are likely to change their behavior just because they're being observed. They may not realize they've changed their behavior. They may think their daily movements, dietary practices, sex life and other behaviors remain unaffected, but they're probably fooling themselves. To the extent that they're also aware of the goals from the study, the subjects will probably act in such a way as to confirm the researchers' hypotheses.

Put bluntly, you can't observe people's behavior objectively unless you're actually spying on them. And that would be unethical. Perhaps illegal.

Read more on the Internet of Things and check out my weekly posts on the Hub.