The expected value test: a new statistical warrant for theoretical saturation

The basic goal of quantitative ethnography (QE) is to use statistics to warrant theoretical saturation. In QE studies that use epistemic network analysis (ENA), statistical comparisons between two or more samples are often used as warrants. However, no standard quantitative techniques have been developed to provide warrants absent of differences between samples—e.g., for the data overall or for particular individuals in the data. In this paper, I introduce the expected value test (EVT), a technique for finding statistical warrants for single samples in the context of ENA-based QE analyses. Building on the concepts of agreement due to chance and randomization, the EVT generates a distribution of networks whose edge weights approximate the expected value of the edge weights due to chance. Using these distributions affords tests of statistical significance been observed networks and chance-based networks. These tests provide warrants that connections observed qualitatively and measured by ENA are theoretically saturated.