Sequencing interviews and surveys with triangulation and sample-size decisions

[FrankLL]

I'm designing the research methodology for my master's thesis in sociology, which will use a mixed-methods approach to study community responses to a new urban development project. I've settled on conducting semi-structured interviews followed by a survey, but I'm uncertain about the optimal sequence and how to effectively integrate the qualitative and quantitative data during analysis without one component dominating the other. For researchers experienced with mixed-methods, what practical steps did you take to ensure your methodology was cohesive? How did you decide on your sample size for each phase, and what specific techniques or software did you find most useful for triangulating your findings from different data types to build a stronger, more nuanced argument?

[LoganL]

Explanatory sequential design here: conduct 12–20 interviews to surface core constructs, then design survey items to quantify those constructs. If your community is fairly diverse, aim 20–30 interviews. For the survey, 200–400 responses usually gives reasonable precision depending on population size; do a quick power calc if you can, focusing on your expected effects.

Keep cohesion by building a data integration plan up front: a one-page research matrix linking interview themes to survey items, response scales, and planned analyses. Run a small pilot integration with a subset to spot mismatches; iterate before full data collection.

Tools: code interviews in NVivo or MAXQDA; analyze surveys in R (tidyverse, survey; use lavaan for SEM if you’re ambitious); use RMarkdown or Jupyter notebooks to keep a transparent chain of analysis; create joint displays (tables/figures that align qualitative themes with quantitative results). Think about a simple dashboard or appendix to present integrated findings clearly.

Common pitfalls: letting one method drive the narrative; sampling bias; not planning validation of constructs; ignoring nonresponse; plan for missing data with imputation or weighting; predefine what counts as triangulation and document criteria; don't let 'nice quotes' overshadow the numbers.

Two practical tips: (a) map your must-test questions to themes in multiple ways so you can compare qualitative interpretive frames with quantitative measures; (b) schedule a mid-study design review with teammates to reframe questions or adjust sampling if gaps emerge; this helps maintain balance.

Ethics and dissemination: consider sharing results with participants; ensure confidentiality; document consent for data reuse; maintain a data management plan and pre-register core hypotheses or analysis steps when feasible.