Qualitative research traditionally has been restricted to small samples due to the time-consuming nature of the analysis. Thanks to innovative software programs, it can now be applied to social media’s large data files using text-based natural language programs that sort words and show networked relationships.
The process is not foolproof. The idiomatic idiosyncrasies of conversation on platforms like Twitter, the variations across demographics and cultures, use of emojis and the tendency of people to respond in semantics reflecting affiliation with the source have made machine-generated analysis of constructs like sentiment difficult to operationalize without a human getting involved. There are tools, however, that enable a qualitative approach at scale while still allowing for manual oversight and manipulation.
Tools such as the software program Leximancer1 and the market analytics tool Quid2 are among those that allow us to process large data files and identify preliminary thematic groups and narrative patterns—and visualize relationships at the macro level—while retaining the ability to dive into any single data point for subjective confirmation of the narrative analysis.
As with most qualitative research, the critical phase—and the really hard work—is translating the output into supportable theoretical hypotheses to derive actionable insights.
- 1. Harwood, I., R.P. Gapp, and H. Stewart. 2015. Cross-check for Completeness: Exploring a Novel Use of Leximancer in a Grounded Theory Study. The Qualitative Report 20, no. 7:1029–1045. ↩
- 2. Quid. n.d. What Does the Current Landscape Look Like for Digital Health? ↩
Copyright © 2019 by the Society of Actuaries, Schaumburg, Illinois.