POLS1009 Chap.13 Text Analysis in Political Research
Text Analysis in Political Research
This chapter follows the course's treatment of text analysis, which much qualitative research uses to extract meaning from documents, speeches, transcripts and field notes. It starts by defining a text broadly and by asking who produced it, for whom, for what purpose and with what consequences for the evidence it can offer.
Analysis aims to summarise, organise and interpret, usually by treating statements as information to be coded and occasionally by treating individual words as data. The central contrast is direction: inductive work builds categories from raw material, with grounded theory as its purest form, while deductive work applies a category scheme taken from theory and then counts or extracts coded units.
The chapter sets out the content-analysis procedure step by step, from choosing a unit of observation to analysing code frequencies, and uses the Manifesto Project to show what a large coding scheme and handbook look like. It lists seven approaches, including thematic and narrative analysis, contrasts discourse with content and framing analysis, and ends by weighing the strengths and blind spots of automated methods.
What this chapter covers
- 01
What counts as a text and the six questions about its producer
- 02
Statements as information versus words as data
- 03
Inductive, bottom-up coding and the grounded theory approach
- 04
Deductive, top-down coding with a category scheme
- 05
Units of observation, codes and rules for ambiguity
- 06
The Manifesto Project as a large coding example
- 07
Framing, discourse, thematic and narrative analysis
- 08
Automated text analysis: methods, strengths and limits
Worked example · free
Coding residents' comments inductively and deductively, from raw text to a category scheme
- 1Inductive start. Read the raw text with no fixed list and tag each statement with a code for what it raises, such as construction noise, journey times or shop trade.
- 1Building categories. Merge similar codes into higher-level categories, for example disruption during building and long-run benefits, and stop when no broader grouping fits.
- 1Deductive scheme. Start instead from a theory that people weigh local costs against shared benefits, and define codes for local cost, shared benefit and neutral, with rules for comments that mention both.
- 1Coding units. Fix the unit of observation, say each sentence, give every unit one code, and count how often each code occurs.
- 1What each yields. The inductive pass reveals concerns the planners had not anticipated; the deductive pass tests whether costs or benefits dominate and allows comparison with other projects.
Key terms
- Grounded theory
- A purely inductive way of analysing text that assigns codes to individual statements and merges them into ever broader categories before interpreting the result, although the research question still guides what is noticed.
- Framing analysis
- A method that identifies how a source presents and makes sense of an issue, for instance as an economic or a legal problem, and that is carried out with the same coding steps as content analysis.
- Discourse analysis
- An approach that studies the way language itself is used to construct and understand political issues, and that tends to work inductively.
- Category scheme
- The set of codes a deductive study applies to a text, each with a written definition and rules for handling ambiguous or mixed statements.
- Automated text analysis
- Computer-based analysis that treats words as data, counting them, their proximity or dictionary matches, or scoring documents with an algorithm trained on hand-coded examples.
Text Analysis in Political Research FAQ
What is the difference between inductive and deductive text analysis?
They run in opposite directions. Inductive analysis starts from the raw material and lets codes and categories emerge, so it is suited to building theory. Deductive analysis brings a category scheme derived from theory to the material, codes it and then counts or extracts the coded passages, so it is suited to testing a theory.
Is grounded theory really free of theory?
Not entirely. Grounded theory is described as purely inductive because codes emerge from the text, but the course stresses that your research question and theoretical approach still shape what you look for while reading. Acknowledge that influence in your method section rather than claiming a blank slate.
What is a unit of observation in content analysis?
It is the piece of text you code one at a time: a whole document, a chapter, a paragraph, a sentence or a defined argument. The right choice depends on the genre of document, the aim of your study and the language involved. Choosing it before coding keeps your counts comparable across documents.
How does framing analysis differ from content analysis?
Content analysis records what is said, such as positions and opinions. Framing analysis asks how the issue is conceptualised by the source, for example whether a policy problem is presented in economic, legal or humanitarian terms. The two follow the same procedure of defining units, applying codes and analysing results.
What are the weaknesses of automated text analysis?
Its assumptions are strong, since counting words treats meaning as if it lived in vocabulary alone. It also struggles with hidden meaning such as humour and metaphor, and working across languages remains a difficulty even as machine translation improves. Its strengths are speed on very large collections and perfectly consistent coding.
Exam move
Practise text analysis on short pieces of material you can finish in a single sitting. Take a single press release, ask the six producer questions about it, and write down what it can and cannot be evidence of. Then code the same release twice: once inductively, letting codes emerge and merging them, and once with a small category scheme you write in advance, including a rule for statements that fit two codes.
Compare what each pass revealed and which surprises only the inductive pass caught. Finally, make a short one-line card for each of the seven approaches the lecture lists, and test yourself on which one a described study is using, and why.
Working through Text Analysis in Political Research in POLS1009? Sia is AskSia’s AI Arts and Humanities tutor — ask any POLS1009 Text Analysis in Political Research question and get a clear, step-by-step explanation grounded in how POLS1009 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.