Mixed Data Integration & Research Tools

Learning Objectives

By the end of this session, you will be able to:

• Construct a joint display that arrays quantitative and qualitative findings side by side, systematically compares them, and articulates meta-inferences — what the combination reveals that neither dataset alone could show
• Merge mixed-methods findings at the interpretation level by identifying convergence, complementarity, and divergence, and by generating defensible explanations for each pattern
• Configure and use a complete research tools ecosystem — statistical software (SPSS/R/Python), qualitative analysis software (NVivo/ATLAS.ti), reference management (Zotero/Mendeley), academic writing tools (LaTeX/Overleaf/Word), version control (GitHub), and open science infrastructure (OSF)
• Select the appropriate tool for each research task based on your methodology, data type, and analytical requirements — not based on familiarity or habit
• Establish a reproducible research workflow that connects data, analysis, writing, and version control, ensuring that your capstone outputs are traceable from raw data to final conclusions

Session Planner

Suggested breakdown of the 4-hour contact session.

Time	Segment	Activity	Mode
0:00–0:08	Opening	Transition to Sem VIII; overview of Weeks 16–30; the integration-and-writing phase: "You've collected data and begun analysis. Now: how do you bring it all together into a coherent dissertation?"	Whole class
0:08–0:30	Lecture 1	Mixed-methods integration — the joint display as an analytical and presentation tool; merging, connecting, and embedding findings; generating meta-inferences; handling convergence, complementarity, and divergence systematically	Lecture
0:30–0:55	Lecture 2	The research tools ecosystem — a complete walkthrough: statistical tools (SPSS, R, Python), qualitative tools (NVivo, ATLAS.ti, Taguette), reference management (Zotero, Mendeley), writing tools (LaTeX, Overleaf, Word), version control (GitHub), project management (OSF, Notion, Trello). Choosing your stack.	Lecture
0:55–1:10	Activity	Joint display construction: given sample QUAN and QUAL findings, construct a joint display and write one meta-inference. Exchange with a partner — does the meta-inference go beyond restating each finding separately?	Pairs
1:10–1:25	Discussion	Share joint displays; discuss how meta-inferences differ from separate conclusions; address common errors (listing findings side by side without integration, overclaiming convergence)	Whole class
1:25–1:40	Break	—	—
1:40–2:00	Tools Demo	Rapid walkthrough of each major tool category — what it does, how to get started, where to find documentation. Students follow along, installing and configuring tools for their own workflow.	Demo
2:00–3:40	Lab Work	Part A: Construct joint display from your own QUAN and QUAL findings; Part B: Configure your research tools stack — install and test all tools; Part C: Set up a version-controlled writing project on Overleaf/GitHub	Individual
3:40–3:55	Peer Review	Exchange joint displays; peer checks: is the meta-inference genuinely integrative? Are convergent and divergent findings handled?	Pairs
3:55–4:00	Exit Ticket	Submit joint display draft; confirm tools stack is configured	Individual

1. Mixed Data Integration — Moving Beyond Separate Chapters

Integration is what distinguishes mixed-methods research from two separate studies published under one title. The most common failure mode in mixed-methods capstones is the "parallel chapters" problem: Chapter 4 presents the quantitative results, Chapter 5 presents the qualitative findings, and the discussion chapter briefly references both without systematically bringing them together. Integration must be designed, executed, and demonstrated — not hoped for.

1.1 The Joint Display — Making Integration Visible

A joint display is a table or figure that arrays quantitative and qualitative findings side by side, organised by theme, construct, or research question. It is simultaneously an analytical tool (forcing you to systematically compare findings rather than selectively notice convergence) and a presentation tool (showing the reader — and the evaluation committee — exactly how integration was achieved).

1.2 Three Integration Patterns — Convergence, Complementarity, Divergence

Pattern	What It Looks Like	How to Handle It	Example Meta-Inference
Convergence	QUAN and QUAL findings point in the same direction — the survey results and the interview themes tell the same story	Report convergence explicitly. The meta-inference: increased confidence in the finding because two independent methods with different strengths (breadth and depth) converge. Don't just state "both datasets agreed" — articulate what the convergence MEANS.	"The convergence of survey and interview data on the centrality of trust — with the survey establishing its predictive strength (breadth) and the interviews revealing its experiential basis (depth) — provides strong evidence that trust is not merely one predictor among many but the foundational condition on which other adoption factors depend."
Complementarity	QUAN and QUAL findings address different facets of the same phenomenon — the survey reveals the pattern; the interviews reveal the mechanism	Report complementarity explicitly. The meta-inference: what we now understand about the phenomenon that we would not understand from either dataset alone. This is the richest and most common pattern in well-designed mixed-methods studies.	"The survey established that autonomy is the strongest predictor of gig worker satisfaction (β = 0.38). The interviews revealed WHY: autonomy enables workers to schedule work around childcare, avoid peak-hour traffic, and maintain dignity by choosing which tasks to accept. Together, these findings suggest that autonomy's predictive power is not about task control per se but about the life-management flexibility it enables — a mechanism invisible to the survey alone."
Divergence	QUAN and QUAL findings point in different directions or appear to contradict each other	Do NOT ignore, minimise, or choose one dataset over the other. Analyse the divergence systematically. Work through possible explanations: different questions, different samples, social desirability, methods effects, genuine complexity. The meta-inference: what the divergence reveals about the phenomenon's complexity.	"While the survey indicated that organisational culture was not a significant predictor of innovation (β = 0.09, p = .18), interviews revealed that culture was the most frequently and passionately discussed factor. This divergence may reflect a measurement artefact — the survey measured culture as 'clan, adhocracy, market, hierarchy' (CVF), while participants described culture in terms of 'psychological safety to fail' and 'permission to challenge senior leaders' — constructs the CVF does not capture. This suggests that the non-significance of culture in the survey reflects construct underrepresentation in the measurement instrument, not culture's irrelevance."

2. The Research Tools Ecosystem — Building Your Workflow

The capstone research workflow spans multiple tool categories: data collection, statistical analysis, qualitative analysis, reference management, academic writing, version control, and project management. A well-chosen, well-configured tool stack reduces friction, prevents data loss, and ensures that your research process is reproducible. A poorly chosen stack — or worse, no deliberate stack at all — produces version chaos, lost files, inconsistent citations, and analysis you cannot reconstruct.

2.1 Statistical Analysis Tools

Tool	Type	Best For	Learning Curve	Key Strengths	Key Limitations
SPSS	Point-and-click + syntax	BBA students; standard social science analyses (t-tests, ANOVA, regression, factor analysis); quick descriptive statistics and charts	Low — menu-driven interface; syntax optional. Easiest to learn from zero.	Low barrier to entry; widely taught in BBA programmes; extensive documentation; APA-style output options; handles labelled data well	Proprietary (requires licence); limited advanced/modern methods; poor reproducibility (point-and-click leaves no audit trail unless you use syntax); clunky graphics
R	Programming language	Advanced statistics; data visualisation (ggplot2); reproducible research (RMarkdown); cutting-edge methods; free and open source	Moderate to high — requires learning R syntax and the tidyverse ecosystem. Steeper initial investment, higher long-term capability.	Free and open source; enormous package ecosystem (CRAN: 20,000+ packages); publication-quality graphics; RMarkdown integrates analysis and writing; strong community	Steeper learning curve; less intuitive for students without programming experience; multiple ways to do the same thing can be confusing; error messages can be opaque
Python (pandas, scipy, statsmodels, scikit-learn)	Programming language	BCA students; ML evaluation; data manipulation at scale; integrating analysis with model development; reproducible workflows (Jupyter notebooks)	Moderate — more intuitive syntax than R for students with programming background; pandas and scikit-learn have consistent APIs	Free and open source; dominant in ML/AI; Jupyter notebooks enable literate programming; same language for data prep, analysis, and ML; excellent for reproducibility	Statistical capabilities less comprehensive than R for specialised tests; visualisation (matplotlib/seaborn) less elegant than ggplot2 by default; package fragmentation

2.2 Qualitative Analysis Tools

Tool	Type	Best For	Learning Curve	Key Strengths	Key Limitations
NVivo	Full-featured CAQDAS	Large qualitative datasets; team-based coding; mixed-methods integration; sophisticated querying (matrix coding, text search, cross-tabulation)	Moderate	Industry standard; rich feature set; strong visualisation (word clouds, cluster analysis, concept maps); can import and code PDFs, audio, video, survey data	Proprietary (expensive licence); Windows/Mac only; overkill for small datasets; can encourage over-reliance on software features at expense of interpretive depth
ATLAS.ti	Full-featured CAQDAS	Grounded theory; visual network views of codes and themes; multi-modal data (text, audio, video, images, geo-data)	Moderate	Excellent network visualisation; strong support for grounded theory procedures; handles diverse data types; cross-platform (Windows, Mac, web)	Proprietary; interface less polished than NVivo; smaller user community; licence cost
Taguette / QualCoder	Free / open-source CAQDAS	Capstone projects with modest budgets; straightforward thematic coding; when you need basic CAQDAS functionality without cost	Low	Free and open source; sufficient for most capstone coding needs; simple, focused interface that doesn't distract from analysis	Limited features compared to NVivo/ATLAS.ti; no team-coding support; less robust for large datasets; limited visualisation; may not be institutionally supported
Manual (Word/Excel/Google Docs + highlighters)	Manual coding	Small datasets (5–10 transcripts); students who prefer tactile engagement with data; when software learning time is better spent on analysis	None	No cost; no software learning curve; direct engagement with data; full control over the coding process	Difficult to manage with 10+ transcripts; no automated retrieval of coded segments; harder to demonstrate systematic analysis; greater risk of inconsistency; difficult to collaborate

2.3 Reference Management

Tool	Key Features	Best For	Cost
Zotero	Browser connector (one-click import from databases); organisation with collections and tags; Word/Google Docs/LibreOffice plugin for cite-while-you-write; 300+ citation styles; group libraries for collaboration; PDF annotation; free and open source	Most capstone students. Free, powerful, and actively developed. The default recommendation unless there is a reason to use something else.	Free (300 MB storage); paid plans for more storage
Mendeley	PDF organisation and annotation; academic social network features; Word/LibreOffice plugin; citation style support; web and desktop	Students working extensively with PDFs who want integrated annotation. Owned by Elsevier — some users object to the publisher affiliation.	Free (2 GB storage); paid plans available

2.4 Academic Writing Tools

Tool	Type	Best For	Key Consideration
Microsoft Word	WYSIWYG word processor	Most BBA capstones; familiar interface; good Zotero/Mendeley integration; track changes for supervisor feedback	Universally available; lowest friction. Large documents (50+ pages with figures, tables, cross-references) can become unstable. Use section breaks, heading styles, and automatic table of contents from the start.
LaTeX / Overleaf	Typesetting system / cloud editor	BCA capstones with equations, algorithms, code listings; students who want professional typography; automatic formatting of citations and cross-references; version history	Higher learning curve (markup language, not WYSIWYG). Overleaf provides a cloud-based, collaborative LaTeX editor with templates — no local installation needed. Produces consistently formatted, professional documents. Particularly valuable for documents with mathematical notation.
Google Docs	Cloud word processor	Collaborative writing with supervisor; real-time commenting; accessible from any device; automatic saving	Good for drafting and collaboration. Zotero integration is available but less seamless than Word. Limited formatting control. Not recommended for the final dissertation due to pagination and formatting issues.

2.5 Version Control & Open Science

Tool	Purpose	Why You Need It
GitHub / GitLab	Version control for code, analysis scripts, and (optionally) writing	Prevents the "final_final_v3_REVISED.docx" problem. Every change is tracked, every version is recoverable. Essential for BCA capstones (code); valuable for BBA capstones (analysis scripts, data processing). Public repositories enable sharing your analysis code — supporting reproducibility and open science.
OSF (Open Science Framework)	Project management and open science infrastructure	Centralises your capstone project: data, materials, analysis scripts, pre-registration, and outputs. Enables you to share your entire research workflow (not just the final paper). Supports pre-registration of hypotheses and analysis plans. Free and designed for academic research.

3. Building Your Capstone Workflow — From Data to Dissertation

3.1 Recommended Tool Stacks by Profile

Profile	Statistical	Qualitative	Reference	Writing	Version Control
BBA — Quantitative (Survey/Regression)	SPSS or R	N/A	Zotero	Word + Zotero plugin	GitHub (for analysis scripts and data processing)
BBA — Qualitative (Interviews/Thematic Analysis)	N/A	Taguette or NVivo	Zotero	Word + Zotero plugin	GitHub (for codebook, thematic map, analysis documentation)
BBA — Mixed Methods	SPSS or R	NVivo or Taguette	Zotero	Word + Zotero plugin	GitHub + OSF (for full project transparency)
BCA — ML/System Evaluation	Python (scikit-learn, pandas, matplotlib)	N/A or Taguette (for error analysis narratives)	Zotero	Overleaf (LaTeX)	GitHub (essential — all code, scripts, and results)
BCA — User/HCI Study	Python or R	Taguette or NVivo	Zotero	Overleaf or Word	GitHub + OSF

3.2 The Reproducible Research Workflow

Think Deeper — Cross Questions

Discuss in pairs before sharing with the class.

Quick Check — Joint Display Diagnosis

Diagnose the problem with each joint display entry.

Knowledge Check — Interactive Quiz

Test your understanding of data integration and research tools.

Lab Activity — Joint Display & Tools Configuration

Part A: Construct Your Joint Display (60 min)

1. Using your own QUAN and QUAL findings (or provided sample data), complete the joint display template (Section 1.1). Create one row per theme/RQ/construct.
2. For each row: populate the QUAN finding, QUAL finding, identify the integration pattern (convergence/complementarity/divergence), and write the meta-inference.
3. Self-check each meta-inference: does it say something that NEITHER the QUAN nor QUAL column already says? If you removed the meta-inference column, would the reader lose insight?

Part B: Configure Your Research Tools Stack (60 min)

1. Reference management: Install Zotero + browser connector + Word/Overleaf plugin. Import your existing references. Create collections for your capstone chapters. Test cite-while-you-write.
2. Statistical/ML tools: Confirm your chosen tool is installed and updated. Open your analysis scripts — can you run them from scratch? If using point-and-click, convert at least one analysis to syntax/script.
3. Version control: Create a GitHub repository for your capstone. Add a README describing the project structure. Commit your analysis scripts and (if applicable) your writing draft.
4. Writing: Set up your writing environment — Word with Zotero plugin, or Overleaf project with appropriate template. Ensure automatic citation and reference list generation works.

Part C: Reproducibility Audit (30 min)

Exchange with a partner. Attempt to reproduce one of their analyses: given their raw data and scripts/syntax, can you regenerate their key result? For qualitative: given their codebook and a transcript excerpt, can you understand their coding logic? Identify and document every step that requires guesswork or additional information.

Exit Ticket

Submit with your joint display and tools confirmation.

Key Takeaways — Week 16