A collective textbook-guide by Colaboratoria
Working with documents and qualitative data analysis in international relations research
- Ilona StadnikAuthor
Qualitative Data Analysis (QDA) is a fundamental methodological tool for scholars in international relations, enabling a nuanced understanding of complex social, political, and cultural phenomena. This chapter outlines the core principles of QDA, including methods of data collection and organization, analytical approaches, and the integration of digital tools and artificial intelligence into the research process.
/01
Working with Documents
In international relations (IR) research, qualitative methodologies are frequently employed. When engaging with a research topic, IR scholars typically conduct extensive literature reviews in academic databases, gather official statements from foreign ministries, analyze policy documents—such as national foreign policy strategies or United Nations resolutions—review think tank publications, and perform media content analyses. Together, these materials constitute a corpus of sources processed primarily as qualitative data, encompassing texts, audio, and video.
A primary challenge at the outset of such research is ensuring an adequate volume of data. The required amount depends on the research objective and underlying hypotheses; however, in qualitative research, a broader dataset generally enhances analytical depth.
The second challenge arises in extracting meaningful insights from this unstructured mass of documents and information. This is where data preprocessing and systematic structuring become essential—tasks for which QDA is specifically designed.
QDA — the systematic examination of non-numerical data—such as texts, audio recordings, video, and images—with the aim of identifying patterns, themes, and conceptual frameworks. Its primary purpose is to generate a comprehensive understanding of the phenomenon under study, taking into account its social, cultural, and historical contexts.
Key Features of QDA:
-
QDA addresses research questions focused on "why" and "how", rather than merely "what" or "how much", which are typically the domain of quantitative methods. - QDA begins with data collection, followed by the inductive development of theories, concepts, or themes grounded in the empirical material.
- QDA seeks to provide a holistic interpretation of social phenomena by considering the broader contextual factors that shape actors' experiences and behaviors.
- QDA is a flexible and iterative process, allowing researchers to refine their research questions, sampling strategies, or data collection methods as new insights emerge.
/02
What Data Are Needed for QDA?
Originating in sociology, QDA traditionally relies on data such as interviews, focus groups, and ethnographic observations. While these remain relevant in certain IR studies, the discipline predominantly utilizes documentary sources and articles— both primary and secondary. Data may appear in various forms: written texts, audiovisual materials, images, or transcripts.
-
- important
The entire collected corpus — including multi-page reports, dozens or hundreds of media or social media posts, and numerous official documents — constitutes raw data. Without systematic processing, only superficial conclusions can be drawn from such material.
- important
The entire collected corpus — including multi-page reports, dozens or hundreds of media or social media posts, and numerous official documents — constitutes raw data. Without systematic processing, only superficial conclusions can be drawn from such material.
/03
Organizing Data for Analysis
Regardless of the specific research approach, the central objective is to systematize the data corpus to enable deeper analysis.
Two principal strategies support this process:
- Reducing Data VolumeThis involves condensing information through summarization or paraphrasing. Common practices include identifying key ideas, salient quotations, and relevant excerpts.
- CodingData are classified and labeled according to predefined or emergent criteria. In QDA, a "code" is a concise descriptor assigned to a segment of text, enabling researchers to identify patterns, track thematic occurrences, and interpret underlying meanings.
Approaches to Data Coding
- Inductive Coding
The most foundational approach, inductive coding aims to generate knowledge directly from the data without imposing pre-existing theoretical frameworks. Insights emerge organically from the material.
- Deductive Coding
Researchers apply a predetermined set of codes derived from prior theory or research to newly collected data.
- Thematic Coding
This advanced stage involves grouping coded segments into broader themes or subthemes, revealing relationships and conceptual structures within the data.
Inductive and deductive coding are often used in combination. For instance, while analyzing digital diplomacy, a researcher might apply deductive codes related to diplomats' use of social media while simultaneously identifying inductive codes capturing unexpected manifestations of digital engagement.
It is important to recognize that most qualitative research eventually incorporates quantitative elements. Visualization techniques—such as frequency counts of thematic mentions, network graphs of actor interactions, or chronological event timelines—help reveal patterns not immediately apparent in raw textual data.
Today, these processes can be significantly enhanced through digital tools, which facilitate both coding and visualization. The following section discusses available software before demonstrating how coding and structuring look in practice.
/04
Software for QDA
Several specialized software programs have been developed to support qualitative data coding and visualization. Among the most widely used are MAXQDA и NVivo. These platforms allow researchers to import diverse data types—texts, videos, images—organize them into categories, and apply codes by highlighting relevant text segments.
MAXQDA software interface
NVivo software interface
Both MAXQDA and NVivo offer built-in visualization tools, ranging from basic word clouds to sophisticated network diagrams and correlation tables.
Common Types of Visualizations in QDA
Despite their analytical power, these tools have a significant limitation: cost. Licensed versions are often prohibitively expensive, making them accessible primarily to institutional users. Although some researchers resort to unauthorized versions, this practice is ethically and legally problematic and not endorsed here.
An alternative is the use of free or open-source software with more limited functionality. For example, QDA Miner Lite provides sufficient capabilities for basic coding and analysis, particularly in early-stage research.
QDA Miner Lite software interface
- Tip
When specialized software is unavailable, researchers can employ simpler organizational methods:- structuring documents into clearly labeled folders;
- using color-coded highlighting in text files (accompanied by a legend in a separate file);
- leveraging annotation and comment features in DOC and PDF formats.
Tip
When specialized software is unavailable, researchers can employ simpler organizational methods:
- structuring documents into clearly labeled folders;
- using color-coded highlighting in text files (accompanied by a legend in a separate file);
- leveraging annotation and comment features in DOC and PDF formats.
Cloud-based solutions such as Atlas.ti offer subscription-based access, eliminating the need for a full license purchase. However, access to such platforms may be restricted in certain regions—for instance, due to geopolitical or financial constraints —and researchers should consider these limitations when planning their workflow.
The following example demonstrates the use of Atlas. ti in analyzing a corpus of U.S. cybersecurity strategies. The dataset has been preprocessed to extract key phrases related to cyber threats.
/05
Using AI for QDA
Sample Prompt for Manual Text Coding
- /promptTask: Conduct qualitative coding of the provided text according to the following instructions.
1. Контекст исследования
Цель анализа: [Впиши цель, например: "выявить барьеры в использовании ИИ в дипломатической работе"]
Тип данных: [интервью/ответы на открытые вопросы / фокус-группы / соцсети / доклады]
Единица анализа: [отдельные предложения / абзацы / полные реплики]
1. Research Context
Purpose of analysis: [ Insert e.g., "to identify barriers to the use of AI in diplomatic work"]
Type of data: [ Insert e.g., interviews, open-ended survey responses, focus groups, social media, reports]
Unit of analysis: [ Insert e.g., individual sentences, paragraphs, full statements]
2. Coding Scheme
Use the following categories and subcodes (modify or expand as needed):
Theme 1 (Technical Issues):- Subcode 1.1: "Lack of equipment"
- Subcode 1.2: "Low internet speed"
- Subcode 2.1: "Lack of management support"
- Subcode 2.2: "Lack of time"
- Subcode 3.1: "Fear of change"
- Subcode 3.2: "Enthusiasm"
3. Инструкции по кодированию
Внимательно прочитай текст, разбивая его на смысловые единицы. Для каждой единицы:- Выдели ключевую идею.
- Присвой подходящий код/подкод (или создай новый, если нужно).
- Объясни выбор кода (1−2 предложения).
3. Coding Instructions
Read the text carefully, dividing it into meaningful units.
For each unit:- Identify the central idea.
- Assign one or more appropriate codes/subcodes (create new ones if necessary).
- Justify the coding choice in 1−2 sentences.
Avoid interpreting implicit meanings unless explicitly supported by the text.
4. Output Format
Present coding results in a table:Text (fragment)
Code
Subcode
Comment
"У нас нет денег на закупку серверов…"
Технические проблемы
Нехватка оборудования
Direct mention of lack of resources
"The director says it’s not a priority…"
Organizational barriers
Lack of leadership support
An indication of the administration’s position
[Optional: highlight quotations, count code frequencies, etc.]
5. Text for Analysis
[Insert the text to be coded here]
How to use the prompt?
- Fill in the research context and coding scheme before submitting the prompt.
- If coding categories are not predefined, request:
[Analyze the text and propose a preliminary coding scheme consisting of 3−5 themes with illustrative examples]
- For complex or contextually rich data, specify:
[Consider not only explicit statements but also metaphors and contextual contradictions]
Potential challenges
AI-assisted coding may produce irrelevant or inaccurate results. A common issue is code noise —spurious or redundant codes generated by the algorithm.
This is what code noise looks like in Atlas.ti when using AI-based coding
Additionally, the segmentation of text into analytical units may lack transparency, and contextual nuances may be lost. Therefore, AI should be used cautiously as a starting point for inductive coding, followed by manual validation on a representative sample. Alternatively, it can accelerate deductive coding when the researcher has a clear, predefined coding framework.
QDA is a powerful method for international relations researchers to uncover the underlying causes of events and the behaviour of actors. Despite the development of digital technologies, a researcher's critical thinking plays a key role: even the most advanced algorithms will not replace human interpretation.
Practicum
- What are the main goals of Qualitative Data Analysis (QDA) in international relations research?
- What types of data are most commonly used in QDA for analyzing international relations?
- What is the difference between inductive and deductive coding?
- Which digital tools can assist in qualitative data analysis, and what are their advantages and limitations?
- How can artificial intelligence be used in data coding, and what risks are associated with its use?
- Matthew B. Miles, A. Michael Huberman, Johnny Saldana. Qualitative Data Analysis: A Methods Sourcebook. SAGE Publications, 2013. www.metodos.work/wp-content/uploads/2024/01/Qualitative-Data-Analysis.pdf
- Johnny Saldaña. The Coding Manual for Qualitative Researchers. SAGE Publications, 2009. www.daneshnamehicsa.ir/userfiles/files/1/انیدلاس.pdf
- Practices of Qualitative Data Analysis in the Social Sciences: A Textbook, edited by E. V. Polukhina. National Research University Higher School of Economics. Moscow: HSE Publishing House, 2023.
- Atlas.ti Research Guides: atlasti.com/guides?_gl=1*5guyzs*_up*MQ.*_ga*MTE2Mzc5MzU4Mi4xNzQ3Njc4MDE0*_ga_K459D5HY8F*czE3NDc2NzgwMTIkbzEkZzEkdDE3NDc2NzgwMjIkajAkbDAkaDAkZGVkeUM3c1FKd20tYjFhalQxS3ZnTHlRcjVNbHZHNFd3OHc.
