Computing and algorithms can take different roles in the research process. For example, there is a clear difference between using computational resources in data collection and pre-processing and conducting the analysis with traditional methods and using computing to engage in exploration of the phenomena - be that simulation models, network analysis, constructive research or algorithmic data analysis. There is a range of opportunities is relevant when thinking what computational methods can do for a research project. Beyond the opportunities, it is important to remember that there are also different roles computational methods can have in the analysis, the two extremes being using only computational methods and using computational methods together with other more traditional methods.
In solo computational social sciences, the research process is based primary on computation. The research questions are answered only through computational methods. For example, using machine learning (Levy and Franklin, 2014), network analysis or simulations (Epstein, 2002) alone are good examples of such work. Necessarily validations and checks can be conducted using other methods, but they do not constitute an answer to the research question. Sometimes scholars can use different methods together to answer the research question (e.g., Maier et al., 2018). This is an efficient way of working with computational methods if the research question allows such study.
Another direction is to use computational methods together with other methods, in a mixed method setting. For example, computational methods can be used as triangulation tool to integrate human-understanding and intuition into the process. Some examples of this approach include “big-data-augmented ethnography”, Laaksonen et al. (2017) propose that ethnographic observations may guide in computational data analysis. They suggest that analysis of ethnographic observations and fieldnotes help researchers to grasp a theoretically meaningful articulation of the research problem as well. Thus, by gaining familiarity with the data researchers and thinking through it researchers are able to describe an aspect critical in the data as well. Similarly, “computational grounded theory” (Nelson, 2017) suggests how researchers gain insights through “human-centred computational exploratory analysis.” These insights are used to determine and operationalize patterns from the data, which can be elaborated by identifying similar patterns using computational mechanisms. Both of these processes propose using researchers and their knowledge to support in developing theoretically meaningful conceptualizations. Therefore, we see hybrid approaches were humans and computational analysis are deeply engaged in the sense-making process as one emergent approach for doing computational social science work.