The Human Element

By Elizabeth Ballard, Senior Analyst

Two things stood out to me as a measurement professional while reading a recent study on the radicalization of young men and the different factors that lead some to violence while others remain non-violent. The first was how the researchers utilized “grounded theory methodology,” a research process that involves the generation of theory from data (and not vice versa). The second was the importance of human interaction between the researchers and their subjects.

image by mazeo

The researchers first established a system of categories for the life histories, politics, activities, and religious beliefs of all the subjects. The researchers then interviewed each participant, assigning the various categories according to how the subjects responded to questions and engaged in conversation. This allowed the researchers to recognize patterns and develop theories about how environment, education, religious affiliation, community involvement, and family life influenced the radicalization of the men. Most importantly, the structured human-centric research was effective in gauging nuance, sarcasm, personality, and other unique traits and characteristics, which provided for a deeper understanding of the causes and motivations for the behavior of the subjects. The research led the academic team to gain critical insights into why some radical youth turn to violence and others are satisfied with non-violent protest.

This process mirrors what we do at CARMA in media measurement. We are given a massive load of information from across the media landscape, and all of this information must be analyzed, categorized, and transformed into data that can provide insight into trends, correlations, and opportunities for our clients.  Like the researchers in the radicalization and violence study, we must do this without being prescriptive; otherwise, we will be blind to patterns and outcomes that we have not anticipated. We should not set out looking for any specific answer, but instead allow general themes and associations to emerge from the data in order to reach and provide meaningful conclusions. 

Both the above study and our approach put a premium on human involvement, as human insight is critical to identifying aspects of behavior unrecognizable to a machine. While many are satisfied with automated analysis solutions, it is important to recognize that computers cannot distinguish sarcasm or metaphor, for example, and cannot split unforeseen hairs as the researchers do between violent and non-violent radicals. I doubt the study would have been as illuminating if the human component were absent from the process, as the subjects would have been statistics on a page without the nuanced perspective that comes from human engagement. The ability to identify these things is of great significance, particularly with the growing influence and dominance of social media where unfettered opinions run rampant.