Building a Codebook
A codebook is the backbone of your qualitative analysis — a structured collection of codes, each representing a concept, theme, or category you're identifying in your data. This page builds a real codebook for the Remote Team Trust Study project two ways: by hand, nesting a code three levels deep, and by letting AI draft the rest from the project's framework.
Watch it happen: manual codes, imports, and AI generation
Every screen below is real. Click through it yourself, or let it play.
What is a codebook?
In qualitative research, a code is a label applied to a segment of text that identifies something meaningful — a theme, concept, pattern, or idea. A codebook is the organised list of all your codes, usually arranged in a hierarchy with parent codes and subcodes, like the Ability branch built above.
Hierarchical codes let you group related concepts and drill down into specifics during analysis. Nesting is purely organisational — it doesn't change how the AI codes your data, only how you browse and report on it afterward.
Creating codes manually
Each code needs a name and, ideally, a definition. Only the name is required to create a code, but a code with no definition is invisible to the AI (more on that below) — so filling in the definition, as the walkthrough does for each of the three demo codes, is what actually makes a manually-created code useful.
Click New code
Type a name and press Enter, or click Add.
Click Edit under Code Definition
Write a clear, specific rule for when this code applies, then click Save definition.
Select the code, then click + Add under Sub-codes
This nests a new child code underneath. Select that child and repeat to go a level deeper — parent, child, grandchild, and beyond.
Importing from Excel or CSV
If you already have a codebook in a spreadsheet — or you'd rather build a large one in Excel than click through the UI dozens of times — click Import Codebook and choose a .csv, .xlsx, or .xls file. The importer auto-detects columns for ID, Code Name, and Description.
| ID (hierarchy) | Code name | Description |
|---|---|---|
1 | Academic Belonging | Sense of belonging in classroom settings… |
1.1 | Peer Connection | Social relationships with classmates… |
1.1.1 | Dorm Life | Informal bonds in student housing… |
2 | Financial Aid | Grants, loans, and tuition barriers… |
The dotted ID column is what builds the hierarchy automatically: 1.1 becomes a child of 1, and 1.1.1 becomes its grandchild — no manual nesting required. Click the ⓘ format-help icon any time for this same reference, plus a one-click Download Sample CSV with working examples.
Generating codes with AI
Click Generate with AI, describe your research goals and themes in plain language, and Paideias drafts a hierarchical codebook — parent categories with color-coded sub-codes, each with a real definition — grounded in your project's description and conceptual framework.
Why definitions matter
This is critical: the AI uses your code definitions verbatim when deciding whether a passage matches a code. It doesn't infer meaning from the code name alone — it reads the full definition. A code with an empty definition is skipped entirely; it's invisible to AI coding until you write one in.
- A vague definition like "related to trust" produces vague, inaccurate coding.
- A precise definition like "Specific instances where someone did or did not do what they said they would do, by when they said they would do it" (the real definition behind Reliability & Follow-Through above) produces focused, accurate coding.
Editing and deleting codes
- Edit a code — click Edit under its definition, or click the pencil icon next to its name in the tree.
- Delete a code — removes the code and all its associated coding data, including any nested children. This cannot be undone.
- Rearrange codes — drag and drop in Code Manager to reorganise the hierarchy; see Code Manager.
Changes are immediate. Existing coded passages aren't affected by renaming or redefining a code — but the next AI coding run uses the updated definition.