Paco—Applying Computational Methods to Scale Qualitative Methods

Abstract

For several years we have been building and using an open mobile research platform, called Paco, that enables the scaling of qualitative research through quantitative, computational techniques. The platform provides a mechanism to design and deliver remote research instruments to mobile devices in the field and it provides mechanisms to abstract and develop new research tools.The most immediate way the platform scales qualitative research is by enabling researchers to visually design, deploy, and manage research instruments comprised of surveys, triggers and sensor logging without needing to program or build a new mobile app. The combination of sensors, surveys and triggers supports idiographic, phenomenological, qualitative inquiry as well as contextual data collection in participants’ natural setting.Stepping back, the platform is an experiment in scaling the generation of new instruments and the generation of new knowledge in the science itself. Under the covers, Paco is implemented as an open construction kit of research components for all to use and modify as they like. Its design enables computational thinking [Wing] about both qualitative and quantitative behavioral research. It makes it possible to generate an infinite combination of research instruments from basic building blocks. Specifically, it borrows concepts and practices from programming language design, software architecture, and the software community.By writing down the elements and methods for research instruments in a precisely specified, machine‐executable language, they become more clear. This makes them better understood. This scales the generation of knowledge in behavioral science.Automated tools cannot replace the researcher. Ethnography has a very deep rich practice of immersive field work and analysis. The researcher is the ultimate instrument for understanding what is significant both individually and culturally within any study. Paco merely offers tools to support and advance the practice by scaling methods, automating parts that are amenable, and, by facilitating precise characterization of the data and data collection protocols.There are many challenges to how well computational methods can model and support behavioral research, particularly the qualitative methods used in ethnography. We finish with a discussion of some of the theoretical and practical challenges and how our method meets, and doesn't meet, those challenges.