Making sense of data using automated content analysis: an illustration using archival data from newspaper articles

Abstract

With the increasing amount of data being generated, marketers and marketing researchers face the challenge of effectively analyzing and interpreting insights from the data. The volume of data poses challenges for humans; however, using automated content analysis techniques frees the researcher to focus on the distilled data. Even though multiple forms of text analysis techniques have been discussed in prior marketing literature, few articles simplify the techniques enough to allow for easy adoption by readers. This article discusses three text analysis techniques and then applies these techniques to a dataset of 1287 newspaper articles following the major demonetization announcement in India. It provides an interesting insight into the life of the Indian citizen faced with a government-mandated drive that demonetized 86% of the currency, endangering everyday retail transactions in a cash-dominated economy. Interesting insights emerging from simple techniques such as comparative word frequencies and sentiment analysis are presented which highlight the coping techniques used by the people to continue retail transactions. The initial desperation led to attempts to use the demonetized currency notes by splurging on gold, liquor, and fuel. Once the awareness about the absence of valid currency seeped in, people focused on more thought-out attempts to sustain normal retail transactions. Further, topic modeling was applied to discover the underlying topics in the data corpus, which further revealed the repertoire of coping strategies used by the people. A topic that stood out in the analysis was related to retail-focused mobile payment services, which subsequently found large-scale acceptance in the economy. The article drives home the point that while automated content analysis may provide a quick and simplified view of the data, the role of the researcher in qualitatively interpreting the data is not trivial.