Content Analysis in Qualitative Research: Methods and Applications

Every qualitative researcher eventually faces the same problem: a mountain of text. Interview transcripts, textbooks, policy documents, social media posts, open-ended survey responses – the raw material piles up fast, and reading it once tells you very little. Content analysis is the method that turns this pile into something you can actually work with: a system of categories that lets you classify, compare, and draw conclusions from text in a way that other researchers can check and repeat.

What is content analysis? 🔗

The classic definition comes from Bernard Berelson, who described content analysis as a technique for the objective, systematic description of the content of communication. Later scholars such as Ole Holsti and Klaus Krippendorff dropped the requirement that the description be quantitative, and shifted the emphasis toward making valid, replicable inferences from text to the context in which it was produced. That shift matters a great deal for qualitative researchers, because it opened the door to analysing meaning, tone, and intent – not just counting how often a word appears.

Classifying and comparing textual data 🔗

At its core, content analysis is about classification and comparison. A researcher takes a body of text, breaks it into meaningful units – a sentence, a paragraph, a theme – and sorts these units into categories built either from theory or from the data itself. Once the material is organised this way, comparison becomes possible: across time periods, across authors, across regions, or across different versions of the same document. A textbook published in 2010 can be compared with its 2022 edition. Election speeches from two rival parties can be placed side by side. News coverage of the same event in two newspapers can be checked for bias. The method gives researchers a disciplined way to move from “I noticed this” to “here is the pattern, and here is the evidence for it.”

Approaches to content analysis: characteristics, causes, and audience 🔗

Berelson originally set out several purposes that content analysis can serve, and these still shape how researchers frame their questions today. Broadly, a study can focus on describing what is actually present in the content, inferring something about why it was produced, or inferring how it might affect an audience. These three angles – characteristics, causes, and audience – are worth separating out, because they lead to very different research designs.

Looking at characteristics 🔗

This is the descriptive approach: what does the text actually say, and how is it said? A researcher studying school textbooks might count how many times male and female characters appear, what occupations are assigned to each, or how frequently a particular concept is illustrated. A well-known example from the Indian context is the recurring analysis of gender representation in NCERT textbooks, where researchers have tracked the proportion of gendered pronouns and roles across editions to see whether representation is improving over time. This kind of study stays close to the surface of the text – it describes patterns rather than explaining them.

Looking at causes 🔗

The inferential approach asks a harder question: why does the content look the way it does? Here, the researcher treats the text as evidence about the people or institutions that produced it – their values, priorities, or constraints. NCERT’s own internal review of its textbooks, for instance, found that earlier attempts to address gender equity often leaned on quantitative fixes, such as simply adding more images of girls, rather than rethinking how equality was framed conceptually. Making a claim like this requires more than counting; it requires interpreting choices against the context in which the material was written.

Looking at audience effects 🔗

The third approach asks what the content might do to the people who consume it. This is the most speculative of the three, since content analysis alone cannot prove an effect – it can only build a plausible case, usually by combining textual evidence with other data such as reader surveys or classroom observation. A study might argue that repeatedly depicting women in domestic roles across a curriculum could shape students’ expectations about gender roles, but confirming that effect would need additional research beyond the text itself.

Procedures in qualitative content analysis: Mayring’s step-by-step process 🔗

While Berelson’s approaches tell you what question to ask, German psychologist Philipp Mayring is best known for answering how to actually carry out a qualitative content analysis in a rigorous, repeatable way. Mayring’s method treats content analysis as a mixed-methods approach, integrating qualitative interpretation with quantitative checks on reliability, and it insists that every piece of text be understood within the communicative context in which it was created, rather than in isolation.

The step-by-step model 🔗

Mayring lays out the research process as a sequence of clear stages, which researchers can adapt but rarely skip:

Forming the research question: Turning a broad interest into a specific, answerable question.
Establishing the theoretical background: Grounding the question in existing literature and concepts.
Designing the study: Deciding whether the research is exploratory, descriptive, relational, or causal in nature.
Gathering material and sampling: Collecting the texts to be analysed and deciding how much of them to use.
Choosing and constructing the method: Selecting which analytical technique – summarising, explicative, or structuring – fits the question.
Conducting the study and presenting results: Applying the coding scheme and reporting what the categories reveal about the research question.
Discussing quality criteria: Checking reliability, especially agreement between independent coders.
Reflecting on research implications: Considering what the findings mean theoretically and practically.

Two features of this model stand out. First, categories are not fixed once and forgotten – they are piloted, tested against a portion of the material, and revised if they don’t fit. Second, Mayring places real weight on intercoder reliability: if two independent researchers apply the same coding scheme to the same text and consistently disagree, the categories need refining before the analysis can be trusted.

Techniques of content analysis: summarising, explicative, and structuring 🔗

Mayring describes several specific techniques for working with text, but three form the backbone of most qualitative content analysis projects.

Summarising content analysis 🔗

This technique reduces the material to its essential content while preserving the core meaning. The process typically moves through paraphrasing the text, generalising or abstracting the paraphrases to a higher level, and then reducing similar statements into a smaller, more manageable set of categories. It is largely an inductive process, building categories up from the data rather than applying them from a pre-existing framework. Summarising works well when a researcher has a large volume of interview transcripts and needs to identify the recurring themes without losing the substance of what respondents actually said.

Explicative content analysis 🔗

Sometimes a passage in the text is ambiguous, or a term is used without explanation, and the researcher needs outside material to make sense of it. Explicative analysis addresses this by bringing in additional context – narrow context analysis draws only on material immediately surrounding the unclear passage, while broad context analysis pulls in information from beyond the text itself, such as background knowledge about the author or the situation. The goal is to produce an explanatory paraphrase that clarifies the passage in light of its full context. This technique is particularly useful in historical or policy document analysis, where a single line can carry meaning that isn’t obvious without background research.

Structuring content analysis 🔗

Mayring treats structuring as the most central of the three techniques, because its purpose is to filter a specific structure out of the material using categories defined in advance. This is a deductive process: the researcher fixes a category system, writes clear definitions and coding rules for each category, and works through the text marking and extracting relevant passages. Structuring can organise material by content, by form, or by degree – sorting statements, say, into strongly positive, mildly positive, neutral, or negative attitudes toward a policy. Because the categories are set before coding begins, structuring analysis is often easier to replicate across coders and studies than the more inductive summarising approach, though Mayring is careful to note that categories should still be revised if the pilot coding shows they don’t capture the material well.

In practice, researchers rarely use just one technique. A study might start with summarising to get a feel for the material, use explicative analysis to clarify a handful of confusing statements, and finish with structuring to test a theory-driven set of categories against the whole dataset. The choice depends entirely on the research question – which is exactly why Mayring places “forming the research question” as the very first step in his model.

What do you think? If you were analysing WhatsApp forwards for misinformation, would you reach for summarising to condense the claims first, or structuring to sort them straight into categories of risk? And if you were studying your own college textbooks, would you rather count what’s visibly there, or dig into why it was written that way?

References

1. https://sk.sagepub.com/ency/edvol/socialscience/chpt/content-analysis
2. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8720803/
3. https://www.cgdev.org/blog/analysing-gender-bias-school-textbooks-india
4. https://ncert.nic.in/dgs/pdf/overallreportDGS_24_8_17.pdf
5. https://sustainabilitymethods.org/index.php/Qualitative_Content_Analysis
6. https://link.springer.com/chapter/10.1007/978-94-017-9181-6_13
7. https://www.qualitative-research.net/index.php/fqs/article/download/3343/4557?inline=1