Qualitative, Content and Case Analysis

Introduction

Mass media management research examines organisations, texts, technologies, workers and audiences. These objects cannot always be understood through numerical measurement alone. A broadcaster’s editorial culture, a platform’s representation of a crisis, or an audience’s interpretation of a campaign may involve meanings, identities, power relations and institutional routines. Qualitative analysis, content analysis and case analysis provide systematic ways of studying these dimensions.

This chapter explains how to move from collected evidence to credible interpretation. It focuses on coding, thematic analysis, qualitative content analysis, discourse analysis and comparative case analysis. It also introduces triangulation and reflexivity, which help researchers demonstrate that findings are well supported rather than merely subjective impressions.

The central principle is that interpretation must be systematic, transparent and connected to the research question. Qualitative research is not an absence of method. It requires explicit decisions about what counts as evidence, how evidence is classified, how patterns are identified and how alternative interpretations are considered.

Learning objectives

By the end of this chapter, students should be able to:

1. Distinguish qualitative, content, discourse and case analysis.
2. Design and apply a defensible coding framework.
3. Conduct thematic analysis from raw material to developed themes.
4. Analyse media texts for manifest content, latent meaning and discourse.
5. Compare media organisations or cases using structured procedures.
6. Use triangulation and reflexivity to strengthen credibility.
7. Address ethics, confidentiality, representation and the responsible use of software and AI.
8. Present interpretations with appropriate evidence and limitations.

1. The analytical task in media research

Analysis is the process of transforming research material into an argued account of a phenomenon. Data may include interview transcripts, field notes, organisational documents, social-media posts, news reports, programme schedules, audience comments, images and audiovisual material.

Three distinctions are useful:

• Data are the collected materials.
• Codes are concise labels attached to relevant segments.
• Themes or categories are broader patterns that organise related codes and answer the research question.

Analysis is different from summary. A summary states what appears in the material; analysis explains how elements relate, what they mean, whose interests they serve and why the pattern matters.

A researcher should begin with the research question, unit of analysis and analytical tradition. The unit might be an interview statement, news story, headline, image, programme, organisation or audience interaction. Confusion about the unit of analysis produces inconsistent coding. For example, coding individual sentences is different from coding complete news stories.

Epistemological position

A positivist or post-positivist study may treat content categories as observable features and emphasise consistency and reliability. An interpretivist study views meaning as situated and co-produced by researchers and participants. A critical study asks how media discourse reproduces power, ideology, inequality or institutional interests. These positions are not merely philosophical labels; they affect sampling, coding, the role of theory and the interpretation of findings.

A project can combine approaches, but the researcher must explain how. For example, a study may count the frequency of crisis frames and then interpret how those frames construct responsibility. The count does not replace interpretation; it supports it.

2. Coding: the foundation of analysis

Coding involves identifying a segment of material and assigning a meaningful label. A code should be relevant to the research question, sufficiently precise and applied consistently. Coding can be deductive, using concepts derived from theory or previous research; inductive, allowing concepts to emerge from the material; or abductive, moving iteratively between evidence and theory.

Common coding levels include:

Building a codebook

A codebook improves transparency. It normally contains the code name, definition, inclusion criteria, exclusion criteria and an example. Consider the following code:

A codebook should be piloted on a small sample. During the pilot, overlapping codes are merged, vague definitions are revised and missing categories are added. The final codebook and its revision history may be included in an appendix.

Coding procedure

A practical procedure is:

1. Read or view all material once to gain familiarity.
2. Record initial observations and possible analytic questions.
3. Select a manageable pilot sample.
4. Apply preliminary codes line by line or segment by segment.
5. Review code overlap and revise definitions.
6. Code the complete dataset.
7. Write analytic memos explaining important decisions and emerging relationships.
8. Group codes into categories or themes.
9. Search for negative cases and contradictory evidence.
10. Produce an interpretation linked to the research question and literature.

An analytic memo is more than a note saying “interesting.” It might state: “Managers describe audience data as empowering, while producers describe it as restrictive. This may indicate that data has different meanings according to organisational position.”

Reliability and consistency

In interpretive research, consistency does not mean that all researchers must produce identical interpretations. It means that analytical decisions are explainable and evidence-based. Where several coders are used, they should jointly code a pilot sample, discuss disagreements and refine the codebook. A disagreement is often analytically useful because it reveals ambiguity in a category.

If a project reports intercoder agreement, the statistic must suit the design and be interpreted cautiously. Simple percentage agreement is calculated as:

This measure is easy to understand but does not account for agreement occurring by chance. It should not be presented as proof of validity. In small interpretive projects, documented discussion, codebook refinement and audit trails may be more appropriate than a numerical reliability claim.

3. Thematic analysis

Thematic analysis identifies patterned meaning across a dataset. A theme is not merely a topic. “Advertising” is a topic; “advertising revenue creates hidden editorial constraints” is a potential theme because it makes an interpretive claim.

A widely used process contains six connected phases:

1. Familiarisation: reading, viewing or listening repeatedly and recording initial thoughts.
2. Generating codes: marking relevant features systematically.
3. Constructing candidate themes: grouping related codes into broader patterns.
4. Reviewing themes: testing whether they fit both coded extracts and the complete dataset.
5. Defining and naming themes: specifying the central organising idea and its boundaries.
6. Writing the report: selecting evidence and explaining its significance.

These phases are recursive rather than strictly linear. A theme may need to be divided, combined or abandoned.

A strong theme has internal coherence, external distinction and explanatory value. It should not be supported only by the most articulate participant or by a single dramatic example. Researchers should examine prevalence, intensity, variation and importance. A less frequent theme may be theoretically significant, particularly when it reveals a marginalised or contradictory perspective.

Worked example: audience research

Suppose a researcher interviews twelve podcast listeners about a public-service podcast. Initial codes include “easy access,” “expert credibility,” “host intimacy,” “commercial interruption,” “lack of local relevance” and “listening while commuting.” These may be organised as follows:

• Convenient public service: easy access and listening during daily routines.
• Relational trust: host intimacy and expert credibility.
• Conditional loyalty: appreciation of the programme combined with concern about commercial interruption.
• Uneven inclusion: strong national identity but limited local relevance.

The final report should not simply list these themes. It should explain the relationship among them: convenience and trust may support loyalty, but loyalty is conditional where commercialisation and weak local representation undermine the programme’s public-service identity.

4. Qualitative content analysis

Content analysis is a systematic examination of communication. It can be quantitative, qualitative or mixed. Quantitative content analysis counts predefined features; qualitative content analysis classifies manifest content and interprets latent meaning in context.

• Manifest content is directly observable, such as the presence of a named source, image or topic.
• Latent content concerns implied meanings, assumptions and cultural significance.

A content-analysis design should specify the population, sample, unit of analysis, categories, coding rules and treatment of ambiguous material. A sample might consist of the front-page stories of three news websites during a defined week. The unit could be each article, while variables include topic, source type, headline frame and prominence.

A category system should be:

• Exhaustive enough to cover relevant material.
• Mutually exclusive where a single category is required.
• Conceptually relevant to the research question.
• Operationally clear.

Some variables legitimately permit multiple codes. An article may contain both an economic and a political frame. The researcher must state whether categories are single-response or multiple-response.

Qualitative content analysis commonly proceeds by condensing material, assigning categories, comparing cases and interpreting relationships. Context must be retained. A headline that appears alarmist in isolation may have a different function when read with the article, publication style and surrounding coverage.

5. Discourse analysis

Discourse analysis studies language as social action. It asks how texts construct reality, identities, problems and possible solutions. A discourse is not simply a set of words; it is a patterned way of representing and organising knowledge and social relations.

Questions may include:

• How is a social problem named?
• Who is represented as responsible, vulnerable or authoritative?
• Which voices are absent or marginalised?
• What assumptions are treated as common sense?
• What actions become legitimate through the representation?
• How do institutional genres and professional conventions shape meaning?

For example, a media organisation may describe platform workers as “flexible entrepreneurs” in one document and “precarious labour” in another. These terms imply different understandings of responsibility, regulation and worker identity.

A discourse analysis can examine vocabulary, metaphors, pronouns, modality, presuppositions, narrative structure, visual composition and intertextual references. It should connect close textual analysis to the social and institutional context. Researchers must avoid claiming that a single word proves an entire ideology; interpretations require patterns and contextual evidence.

6. Case analysis and comparative case analysis

A case is a bounded system studied in depth. The boundary may be an organisation, campaign, programme, platform partnership, newsroom, crisis or period. Case study is especially useful when the research asks how or why a contemporary process occurs in its real context.

A case design should explain:

• Why the case is significant.
• What its boundaries are.
• Which sources will be used.
• How evidence will be analysed.
• What can and cannot be generalised.

Cases may be intrinsic, where the case itself is the main interest; instrumental, where it illuminates a wider issue; or collective, where several cases are studied together.

Comparative case analysis

Comparison can be most similar systems: cases share many characteristics but have different outcomes. It can also be most different systems: cases differ in several respects but display a similar outcome. In either design, the researcher must avoid attributing an outcome to one factor merely because it appears alongside it.

A useful comparison matrix might include:

The researcher then conducts within-case analysis before cross-case analysis. Within-case analysis preserves context; cross-case analysis identifies similarities, differences and explanatory conditions. The final argument should distinguish factual difference from analytical significance.

7. Triangulation and credibility

Triangulation involves examining a research question through more than one perspective, source, method, researcher or theory. It is not a mechanical guarantee of truth. Different sources may disagree because they reveal different positions or experiences.

Types include:

• Data triangulation: comparing interviews, documents and observations.
• Method triangulation: combining content analysis with interviews or observation.
• Investigator triangulation: involving more than one analyst.
• Theory triangulation: interpreting evidence through competing conceptual lenses.
• Time triangulation: examining material across different periods.

Suppose managers claim that a newsroom has adopted audience-centred decision-making. Meeting documents may support this claim, while interviews with reporters may reveal that audience metrics are used mainly to justify cost-cutting. The disagreement should be reported and analysed, not hidden. It may indicate that the same policy has different consequences across organisational roles.

Credibility is also strengthened through prolonged engagement, thick description, participant clarification where appropriate, peer debriefing, negative-case analysis and an audit trail. Transferability is supported by detailed contextual description, allowing readers to judge whether findings apply elsewhere. Dependability requires a clear account of procedures. Confirmability requires evidence that interpretations arise from the material rather than undisclosed personal preferences.

8. Reflexivity

Reflexivity is the continuing examination of how the researcher’s identity, assumptions, relationships and methodological choices shape the research. It is not a ceremonial paragraph stating that the researcher is unbiased; it recognises that complete neutrality is rarely possible.

A reflexive account may consider:

• The researcher’s professional or political relationship to the organisation.
• Differences of class, gender, language, ethnicity or generation between researcher and participants.
• How access was negotiated and who was excluded.
• How interview questions may have shaped responses.
• Why some interpretations seemed initially persuasive.
• How contradictory evidence changed the analysis.

A reflexive journal can record decisions, emotional reactions, access problems, coding changes and alternative explanations. Reflexivity improves rather than weakens credibility because it makes the production of knowledge visible.

9. Ethics, governance and responsible technology

Media research frequently involves public speech, vulnerable participants, confidential organisational information and potentially identifiable online users. Ethical review should address informed consent, privacy, data security, withdrawal rights, risk of harm and lawful data handling.

Publicly available does not automatically mean ethically unrestricted. A social-media post may be technically public but produced by a person who does not expect academic quotation. Researchers should consider whether direct quotation could enable reverse identification, particularly for minors, activists or victims of abuse. Paraphrasing, anonymisation and removal of unnecessary identifiers may be appropriate.

Organisational research can expose whistle-blowers or employees. Researchers should separate identifying information from transcripts, restrict access and report findings in ways that do not create avoidable employment or reputational harm. Ethical obligations continue after publication.

Qualitative-analysis software can assist with storage, retrieval, coding and memoing, but it does not interpret context automatically. Automated sentiment or topic models may reproduce linguistic and cultural bias. Generative AI can help produce draft summaries or suggest possible code names, but researchers must not upload confidential transcripts without permission, treat generated output as evidence or conceal its use where disclosure is required. Human review, source checking and secure data governance remain essential.

10. Applied case: comparing audience participation in two newsrooms

This is an illustrative case. It demonstrates an analytical process rather than reporting an actual organisation.

Background

A researcher investigates why two online newsrooms introduced audience participation schemes. Newsroom North, an established regional publisher, invited readers to submit questions for a weekly civic programme. Newsroom Stream, a small digital start-up, used a membership forum to suggest and discuss stories.

Problem

Both organisations described their initiatives as “audience-led,” but it was unclear whether participation influenced editorial decisions or merely increased engagement metrics.

Approach

The researcher bounded the case to the first six months of each initiative. Evidence included twelve staff interviews, 240 audience contributions, editorial guidelines, meeting documents and published stories. The analysis used:

1. Deductive codes for influence, inclusion, moderation and accountability.
2. Inductive codes for unexpected experiences.
3. Thematic analysis of interviews and contributions.
4. Qualitative content analysis of published stories.
5. Within-case and cross-case comparison.
6. Triangulation across staff accounts, documents and outputs.

Analysis

In Newsroom North, staff described participation as a source of questions and local leads. Documents showed that editors retained final authority, and only a small proportion of contributions appeared in published work. Audience participation therefore operated mainly as consultative input. In Newsroom Stream, members discussed story ideas directly with reporters, and editorial meetings recorded reasons for accepting or rejecting suggestions. This created greater visible influence but also produced conflict when highly active members dominated discussion.

Thematic analysis identified three themes: participation as trust-building, participation as unpaid labour and participation as contested editorial power. Content analysis showed that both newsrooms increased coverage of service information, but only Newsroom Stream regularly published explanations of editorial decisions. The cases therefore differed not simply in technology but in governance and accountability practices.

Outcome

The researcher concluded that “audience-led” was an ambiguous organisational label. Participation had stronger editorial consequences where procedures made influence visible, but openness also required moderation, representation safeguards and explanations of decisions.

Learning points

• A case must have clear boundaries.
• Organisational claims should be tested against documents and outputs.
• Participation should be analysed as a power relationship, not only as a volume of comments.
• Comparison is strongest when similarities and differences are both explained.
• Negative evidence—such as rejected contributions—may be analytically important.

11. Writing the findings and interpretation

A strong findings chapter normally moves between claim, evidence and interpretation. A useful paragraph structure is:

1. State the analytical claim.
2. Present carefully selected evidence.
3. Interpret the evidence using concepts or theory.
4. Consider variation or a counterexample.
5. Explain the significance for the research question.

Quotations should be purposeful, not excessive. Introduce each quotation, identify its source using the approved system and explain what it demonstrates. Do not allow a quotation to stand in place of analysis.

Researchers should distinguish participants’ claims from the researcher’s conclusions. “The editor stated that metrics support autonomy” is a report of an account. “The account frames metrics as neutral, thereby obscuring managerial power” is an interpretation requiring textual and contextual support.

Limitations should address sampling, access, language, time period, researcher position and the boundaries of inference. A qualitative case does not usually support statistical generalisation to all media organisations. It may nevertheless support theoretical or analytical generalisation by showing how a concept operates under particular conditions.

Best-practice checklist

Before submitting an analysis, ask:

• Is the research question visible in every major analytical decision?
• Is the unit of analysis clear?
• Are codes defined and illustrated?
• Have themes been distinguished from topics?
• Is contradictory evidence discussed?
• Are claims supported by more than one convenient quotation?
• Has context been preserved?
• Are case boundaries and comparison criteria explicit?
• Have triangulation and reflexivity been documented?
• Are ethical and data-security decisions explained?
• Are software or AI tools treated as aids rather than substitutes for judgement?

Chapter Summary

Qualitative, content and case analysis provide complementary ways to study media organisations, audiences and texts. Coding converts raw material into manageable analytical units; thematic analysis develops patterned interpretations; content analysis systematically examines manifest and latent features; discourse analysis investigates how language constructs realities and power relations. Case analysis preserves organisational and historical context, while comparative case analysis identifies meaningful similarities and differences.

Credible interpretation depends on transparent procedures, a carefully designed codebook, iterative memoing, attention to negative cases and a clear relationship between evidence and claims. Triangulation tests interpretations across sources, methods, researchers, theories or time. Reflexivity recognises the researcher’s role in producing knowledge. Ethical research protects participants, handles organisational information responsibly and uses software and AI with human oversight. The ultimate aim is not to eliminate interpretation but to make interpretation reasoned, inspectable and useful for understanding media management.

Key Terms

Analytic memo: A written record of an emerging interpretation, question or methodological decision.

Case study: An in-depth investigation of a bounded contemporary system in context.

Code: A concise label assigned to a relevant segment of research material.

Codebook: A structured record of code definitions, boundaries and examples.

Content analysis: Systematic examination and classification of communication material.

Data triangulation: Comparison of evidence from different sources or groups.

Deductive coding: Coding guided by concepts established before examining the data.

Discourse: A patterned way of representing knowledge, identity and social relations.

Intercoder agreement: The degree to which different coders assign the same codes.

Interpretivism: An approach that treats meaning as socially situated and constructed.

Latent content: Implied or underlying meaning in communication.

Manifest content: Directly observable content in communication.

Negative case: Evidence that challenges or does not fit an emerging interpretation.

Reflexivity: Examination of how the researcher shapes the research process and findings.

Thematic analysis: A method for identifying and interpreting patterned meaning.

Theme: A coherent, significant pattern that makes an analytical claim.

Transferability: The extent to which readers can judge relevance to another context.

Unit of analysis: The entity or segment that is coded or compared.

Review Questions

Short Answer Questions

1. What is the difference between a code and a theme?
2. Distinguish manifest content from latent content.
3. Why is a codebook useful in qualitative research?
4. What is reflexivity?
5. Name two types of triangulation.

Long Answer Questions

1. Explain the stages of thematic analysis and discuss why the process is recursive.
2. Compare qualitative content analysis with discourse analysis.
3. Explain how a researcher can establish credibility in an interpretive media study.
4. Discuss the importance of clear boundaries in case-study research.
5. Examine the ethical risks of quoting social-media content in a dissertation.

Analytical / Application Questions

1. Design a coding framework for studying how a news organisation represents artificial intelligence. Include at least five codes and their inclusion criteria.
2. A manager claims that audience metrics increase editorial freedom, while journalists claim that metrics restrict professional judgement. Explain how you would investigate this disagreement using triangulation.
3. Compare two media organisations that adopted different audience-participation systems. Propose dimensions for comparison and explain how you would avoid confusing correlation with explanation.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes a theme from a topic?

A. A theme is always a numerical frequency

B. A theme makes an interpretive claim about a patterned meaning

C. A theme is a single quotation selected by the researcher

D. A theme is identical to the unit of analysis

Correct Answer: B. A theme makes an interpretive claim about a patterned meaning

Question 2

What is the primary purpose of a codebook?

A. To replace the research question

B. To guarantee that participants agree with the findings

C. To define coding categories and promote consistent application

D. To calculate the sample size

Correct Answer: C. To define coding categories and promote consistent application

Question 3

Which is an example of manifest content?

A. The assumed ideology of a headline

B. The emotional effect of a broadcast on viewers

C. The number of named sources in an article

D. The hidden power relation in an image

Correct Answer: C. The number of named sources in an article

Question 4

What does reflexivity require the researcher to examine?

A. Only the response rate of a survey

B. How the researcher’s position and assumptions shape the study

C. Whether all findings can be generalised statistically

D. Whether software can code without human oversight

Correct Answer: B. How the researcher’s position and assumptions shape the study

Question 5

Which procedure is most appropriate before applying a new codebook to a complete dataset?

A. Delete all ambiguous material

B. Pilot the codebook on a small sample

C. Publish the findings immediately

D. Allow each coder to define codes independently

Correct Answer: B. Pilot the codebook on a small sample

Question 6

What is data triangulation?

A. Using only the largest available dataset

B. Comparing evidence from different sources or groups

C. Converting every qualitative response into a percentage

D. Removing contradictory evidence

Correct Answer: B. Comparing evidence from different sources or groups

Question 7

In comparative case analysis, why is within-case analysis normally conducted before cross-case analysis?

A. To preserve the context of each case before comparison

B. To ensure that cases become statistically identical

C. To avoid collecting documentary evidence

D. To eliminate the need for theoretical concepts

Correct Answer: A. To preserve the context of each case before comparison

Question 8

Which statement about publicly available social-media posts is most accurate?

A. They can always be quoted without ethical review

B. They are never suitable for academic research

C. Their public status does not remove privacy and harm considerations

D. They automatically represent the views of an entire audience

Correct Answer: C. Their public status does not remove privacy and harm considerations

Question 9

Why should a researcher examine negative cases?

A. They increase the number of quotations automatically

B. They test and refine the strength and boundaries of an interpretation

C. They prove that the original theme is incorrect

D. They remove the need for a literature review

Correct Answer: B. They test and refine the strength and boundaries of an interpretation

Question 10

Which statement best describes the role of qualitative-analysis software?

A. It independently determines the correct interpretation

B. It replaces ethical review and data governance

C. It assists organisation and retrieval but does not replace researcher judgement

D. It guarantees intercoder validity

Correct Answer: C. It assists organisation and retrieval but does not replace researcher judgement