Multiple Case-Based Thematic Analysis — Cross-Case Thematic Analysis

Standard thematic analysis is typically applied to a single data set, whereas MCBTA applies thematic analysis separately within each case first and then compares the resulting thematic maps across cases. The addition of the cross-case comparison stage is what makes it a multiple case-based design. MCBTA also inherits case study methodology's emphasis on bounded cases and replication logic, which is absent in single-data-set thematic analysis.

A minimum of three cases is recommended to support meaningful cross-case comparison. Yin's guidance suggests that two to three cases provide literal replication, while four to six cases can sustain theoretical replication. Very large case sets (more than ten) become difficult to manage without reducing the depth of within-case analysis. The right number depends on the complexity of the phenomenon and the available resources.

Not necessarily identical, but the data should be sufficiently comparable. Using the same interview guide, the same document types, and the same observation protocol across cases ensures that differences in findings reflect real contextual variation rather than differences in data quality or type. Pilot the protocol in one case before rolling it out across all sites.

Yes. MCBTA is compatible with both inductive and deductive coding. A deductive approach begins with a pre-specified framework (for example, derived from theory or a prior review) and applies it consistently across all cases before comparing case-level findings. An inductive approach builds codes from each case's data. Many MCBTA studies use an abductive, iterative approach that combines both.

Treat divergent cases as theoretically important rather than as errors. Inspect what is different about that case — context, participant characteristics, intervention fidelity — and use the divergence to refine or qualify the overarching themes. Negative case analysis strengthens the credibility and transferability of the findings.