Skip to content
Back to search results

NVivo and ATLAS.ti for Qualitative Analysis

Introduced by
QSR International (NVivo) and Scientific Software-Citational (ATLAS.ti)
Year
1999
Type
Tool
Family
analysis-tool
Font size

What does it do?

NVivo and ATLAS.ti are Computer-Assisted Qualitative Data Analysis Software (CAQDAS) programs that facilitate coding, organizing, and analyzing qualitative data—including text (transcripts, documents), images, video, and audio.

NVivo, developed by QSR International, is widely used in academic research and supports data organization, coding, memo-writing, retrieval, and analysis visualizations. ATLAS.ti, developed by Scientific Software-Citational, emphasizes hermeneutic interpretation and network visualization. Both tools were introduced in the late 1990s and have become standard across disciplines. CAQDAS is not analysis itself—the researcher must make analytical decisions—but rather augments human analysis by managing large data volumes, organizing codes systematically, tracking analysis decisions, and generating visualizations. These tools improve transparency and rigor in qualitative research.

The intuition behind the method

Qualitative analysis is cognitively demanding: keeping track of hundreds of coded segments, maintaining consistency in coding across time and team members, remembering which codes appear in which documents, and managing memo writing while coding. CAQDAS automates the mechanical work, freeing researchers to focus on interpretation. For small projects (single analyst, 5–10 interviews), manual coding with highlighters and index cards may be sufficient. For larger projects—20+ interviews, multiple analysts, multiple document types—CAQDAS provides efficiency and consistency. Both NVivo and ATLAS.ti allow researchers to code data, link codes together, organize codes into hierarchies or networks, write memos attached to codes, search and retrieve coded segments, and visualize coding patterns. The software does not decide which codes to create, how to interpret data, or what themes matter; these remain the researcher's intellectual work. CAQDAS enhances this work by providing reliable organization and powerful retrieval.

History

Early CAQDAS programs (Ethnograph, HyperRESEARCH) emerged in the 1980s as personal computers became powerful enough to manage text. NVivo (QSR International, originally built as successor to N4, first released 1999) became dominant in academic research, particularly in Australia and UK. ATLAS.ti (Scientific Software-Citational, founded 1991) emphasized hermeneutic interpretation and network mapping, gaining traction in European qualitative research. Both companies have evolved their software substantially: modern versions support video/audio analysis, mixed-methods integration, and cloud collaboration. Recent critiques have highlighted the risk of CAQDAS producing mechanical coding without genuine interpretation (Lewins & Silver 2007). Contemporary best practices emphasize that CAQDAS is a tool to support, not replace, human interpretive work.

When to use it

CAQDAS is valuable for research with moderate-to-large data sets (15+ interviews, 5+ focus groups, 100+ documents). For smaller projects, manual analysis may be equally rigorous and faster to learn. CAQDAS is essential when multiple analysts are coding (ensures consistency); when tracking is complex (many variables, subgroups, time points); or when transparency and audit trails are priorities (software documents all coding decisions). CAQDAS is less beneficial for purely narrative or interpretative analysis that requires sustained, close reading and flow; some researchers find that software creates distance from text. Mixed-methods research can use CAQDAS for qualitative data while using statistical software for quantitative analysis, enabling integration. Training in the software is required; novice users often find steep learning curves, so budget time for learning.

Limitations and considerations

  • Learning curve and cost: software requires training and subscription or purchase; investment of time and money.
  • Risk of mechanistic coding: software can create illusion of analytical rigor without genuine interpretation; garbage in, garbage out.
  • Creates distance from text: some researchers find that coding in software distances them from the richness and nuance of original text.
  • Not intuitive: software workflows differ from researcher thinking; some find it constraining rather than enabling.
  • Software does not analyze: the software stores and retrieves codes but does not generate interpretations; analysis is still human intellectual work.
  • Version compatibility and data migration: software updates may create compatibility issues; data can be difficult to export and reuse.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Manages large data volumes and complex coding schemes efficiently; reduces clerical burden of manual organization.
  • Ensures consistency: codes are applied uniformly; retrieval is comprehensive (no missed segments).
  • Supports transparency: all coding decisions are traceable; software keeps audit trail of changes.
  • Enables systematic analysis: code-frequency reports, matrices, and network visualizations reveal patterns humans might miss.
  • Facilitates collaboration: multiple analysts can code the same data and software calculates inter-coder agreement.
  • Integrates data: combines text, images, video, and audio in single project, enabling multimedia analysis.

Applications

  • Academic qualitative research: students and faculty use CAQDAS for theses, dissertations, and publications across disciplines.
  • Health and social care research: analyzing patient interviews, healthcare provider perspectives, or implementation of interventions.
  • Organizational and business research: coding interview data about organizational culture, employee experiences, or change management.
  • Evaluation and program evaluation: systematically coding stakeholder interviews, program documents, and observational data.
  • Mixed-methods research: integrating qualitative coded data with quantitative analysis to explore mechanisms or validate findings.

How it works

How it works

1

Import data into the software: transcripts (text files), documents (PDF, Word), video, audio. Create a project file to organize all data and analytical work. Begin coding: select text segments, assign codes (open coding: codes emerge from data; or use a predetermined codebook). Create code books: document each code's definition, inclusion/exclusion criteria, and exemplar quotes. Link codes: show relationships between codes (hierarchical, network-based). Write memos: attach analytical notes, reflections, and emerging interpretations to codes, segments, or the project. Use the software's retrieval functions to search: 'Show all segments coded with X' or 'Show all segments with codes X and Y together' (code intersection). Generate code co-occurrence matrices or visualizations showing how codes relate. Use software's auto-coding features cautiously: software can flag segments matching keywords, but human review is essential. Create query tools (e.g., 'Codes appearing in interviews with female participants') to explore patterns.

2

Use CAQDAS to support systematic analysis. Maintain a coding log documenting when codes were created, revised, or merged. Track code frequency: the software counts how many segments are coded with each code. Create hierarchies: organize codes into categories (parent codes) and sub-codes, showing conceptual organization. Build code matrices: compare coding patterns across participants or documents (e.g., age groups). Visualize networks: some software creates node-link diagrams showing code relationships. Use the software's output (code frequency reports, code lists, matrices) to identify patterns: Which codes appear most frequently? Which co-occur? Perform quality checks: is coding consistent across documents and time? If multiple coders, check inter-coder agreement using software functions. Generate summaries: the software allows export of coded segments, matrices, and memos to support writing.

Common pitfalls

  • Over-reliance on software: assuming that systematic coding through software automatically produces rigorous analysis without reflexivity and interpretation.
  • Creating too many codes: software allows proliferation of codes without discipline; many researchers create 100+ codes, then struggle to manage or interpret them.
  • Coding without a codebook: failing to document code definitions, leading to inconsistent application and coding drift (code meanings change over time).
  • Neglecting to read full documents: staying in code segments without reading full interviews or documents; context is lost.
  • Treating code frequency as equivalent to importance: codes appearing most frequently are not necessarily the most important or meaningful.
  • Inadequate inter-coder training: when multiple analysts code, skipping detailed codebook development and training leads to disagreement and inconsistency.

Frequently asked questions

Should I use NVivo or ATLAS.ti?

Both are robust and produce equally rigorous analyses. NVivo is more widely used (easier to find tutorials, support), has stronger visualization features, and integrates with quantitative software. ATLAS.ti emphasizes network mapping and hermeneutic philosophy; some researchers find its interface more intuitive. Try both (many offer free trials) or ask colleagues which they prefer. Your institution may provide licenses to one or both, influencing your choice.

Can I do qualitative analysis without CAQDAS?

Yes. Manual coding (highlighting, index cards, physical organization) is perfectly legitimate and produces equally rigorous analysis. CAQDAS is efficient for large projects and supports transparency, but is not required. Choose based on project size, team size, and personal preference. Some researchers combine approaches: manual initial coding, then software organization for later phases.

How many codes should I create?

Codebooks vary: some researchers use 20–30 codes (parsimonious, easy to manage); others use 100+ codes (detailed). There is no 'correct' number. Aim for a balance: codes should be specific enough to be meaningful and consistent (not ambiguous), but not so numerous that you struggle to remember or manage them. Start with 15–30 codes; refine as you analyze more data.

How do I know if my coding is reliable?

For single analyst: check consistency over time by recoding a subset of data (5–10% of documents) after weeks or months and comparing coding. For multiple analysts: calculate inter-coder agreement using software functions or Cohen's Kappa. Agreement of 80%+ on a sample of data suggests reliable coding. Disagreement is useful: discuss differences, refine codebook definitions, and resolve inconsistencies.

Can CAQDAS software do my analysis for me?

No. Software organizes data, facilitates coding, and generates output (matrices, visualizations, retrievals), but does not interpret. Analysis—deciding what codes mean, identifying themes, drawing conclusions—is intellectual work you must do. CAQDAS is a tool, not a substitute for human interpretation and reflexivity.

Related methods

Sources

  1. Lewins, A., & Silver, C. (2007). Using Software in Qualitative Research: A Step-by-Step Guide. SAGE Publications.

    ISBN: 978-1412903653
    Copied.
  2. Friese, S. (2019). Qualitative Data Analysis with ATLAS.ti (3rd ed.). SAGE Publications.

    ISBN: 978-1526424944
    Copied.
  3. Bazeley, P., & Jackson, K. (2013). Qualitative Data Analysis with NVivo (2nd ed.). SAGE Publications.

    ISBN: 978-1446257098
    Copied.
  4. Hoover, S. M., Polasek, D. W., & Shuart-Faris, N. (2012). Qualitative software and rigor in systematic reviews. Journal of Electronic Resources in Medical Libraries, 9(3), 194-204.

    Open publication reference
    Copied.