Using Bayesian statistical modelling as a bridge between quantitative and qualitative analyses: illustrated via analysis of an online teaching tool

Abstract

Bayesian methods provide a more general approach to statistical analysis that mathematically includes Null Hypothesis Significance Testing (NHST) and classical statistical modelling as special cases. This expanded, Bayesian, approach provides several benefits, which we illustrate using a case study about decision-making by teachers. We focus on a relatively unexplored topic: the way in which a Bayesian approach provides a 'bridge' between qual/quant methods. We highlight five bridges, illustrated using the case study: (1) visualization of the conceptual framework, (2) generalization via randomization and alternatives, (3) stories for interpretation, (4) computation that is flexible, and (5) continual learning, through priors. This work illustrates these bridges using a case study on a digital tool that wove together: a behavioural study to investigate decision-making, with an inbuilt perceptual component to probe rationale for specific decisions, and an interview component. A mixed method was therefore a natural choice for integrating learnings across these data sources. This online tool inhabits the digital realm of education which enables 'virtual' assessment, and sits midway between theoretical, written pieces and the fully immersed practicums in the classroom. Adopting a similar tool for training in a virtual classroom could provide greater accessibility and privacy, and enhanced feedback through the in situ qual/quant analytics. Thus the digital learning sphere provides a context for raising awareness of the potential that Bayesian statistical paradigm offers researchers who wish to connect qual/quant methods.

Preservice teacher perceptions of preparedness for teaching: Insights from survey research exploring the links between teacher professional standards and agency

“…There is a small mixed methods component of this study, where we triangulate the hypothesis test findings with those arising from previous analyses of interviews: discourse analysis of the short-answer questions associated with the surveys (see Willis, 2013); and text analysis of interview questions following-up responses to an online exercise (Low-Choy et al, 2017). This follows the approach outlined by Low-Choy et al (2017, Bridge #3) to use 'stories' from qualitative analysis to elucidate findings from quantitative analysis (p. 321).…”

“…This follows the approach outlined by Low‐Choy et al . (2017, Bridge #3) to use ‘stories’ from qualitative analysis to elucidate findings from quantitative analysis (p. 321). In addition to analysing each of the hypotheses separately, regarding each of the 16 items, as described above, an exploratory analysis using a classification tree model may consider whether the two cohorts can be profiled in terms of their responses to the 16 items.…”

“…A useful focus for further work could compare responses to the survey tool with in‐depth interviews or other qualitative analysis which would require a mixed methods design (e.g. Low‐Choy et al ., 2017). In addition, having tested the methodology and concept on two small cohorts, further work would expand the size and scale of the research to consider larger cohorts, and hence increased participant numbers.…”

Exceeding Expectations: Teachers’ Decision Making Regarding Aboriginal and Torres Strait Islander Students

“…Sections with the same label or codes were noted and re-examined to confirm individual relevance to the code applied, and as a group, labeled within that code. Statistical analysis (using ordinal regression) revealed that while teachers focused on the grades represented on the record cards, they were also influenced by ascribed characteristics, particularly, for record cards with lower grades (see Low-Choy, Riley, & Alston-Knox, 2017).…”

Social awareness and engagement in undergraduate music students: Generating a foundation for curriculum decisions

“…It provides a foundation for further qualitative studies that could explore in more detail the themes identified in this survey. In turn, this combination of exploratory qualitative and quantitative studies could provide a basis for a far more comprehensive quantitative study: a survey of student attitudes, potentially ‘mixed-in’ with a parallel qualitative study (for an example of how this could be achieved using a Bayesian statistical approach, see Low-Choy et al, 2017).…”