Imagine a US agency asking the same purchase-barrier question in Mexico, Colombia, Chile and Argentina. Each local team returns plausible themes. When the files are combined, one team has coded limited spending power as a price complaint, while another has kept the two separate. The labels look comparable; the decisions behind them differ.
A multi-country codebook needs shared definitions, examples that explain the boundaries, and a record of consequential changes. This guide focuses on governing that system across Spanish-language studies in Latin America. All survey answers, code entries and study scenarios below are synthetic illustrations, not respondent data or measured findings.
For the wider language context, start with our guide to coding Spanish responses in Latin America and Spain. Spain is an additional comparison outside Latin America.
Start with the question and the reporting decision
Record the exact question, its routing and the population answering it. A response to “Why did you decide against buying?” cannot automatically be interpreted like a response to “How much would you spend?” Keep each market's questionnaire wording beside the shared analytical objective.
Decide what the client needs to distinguish. Will the report separate perceived price, available budget, value for money and payment timing? If those distinctions affect the recommendation, define them before coding. Also record which differences in question wording, eligibility or fieldwork limit the comparisons. A common codebook cannot resolve every difference in study design.
The Cross-Cultural Survey Guidelines on data harmonization recommend planning comparability and preserving documentation that traces transformations back to the original data. The workflow below applies that principle to open-ended coding.
Give each concept a stable identity
Use an ID that survives an editorial change to its display label. For example, 101 can identify “Price perceived as high” and 102 “Insufficient available budget.” English and Spanish labels can sit beside the same ID. Correcting a translation should not silently create a different reporting variable.
A working entry needs a definition, inclusion and exclusion rules, examples, permitted co-codes, an owner and a version. Our survey codebook entry example covers the basic fields. For multi-country work, add the shared reporting category and any market-specific notes needed to apply it.
Keep retired IDs in the historical register and never reuse them for a new meaning. When a definition changes enough to alter assignments, record it as a measurement change. Consider a new ID or an explicit versioned mapping so a familiar label does not conceal a broken trend.
Write the boundary, then test it
For the synthetic question “Why did you decide against buying?”, define code 101 to include an explicit judgment that the offer costs too much. Exclude statements only about available funds, neutral amounts and material descriptions.
- “El precio me parece demasiado alto” (“The price seems too high to me”) supports 101.
- “Este mes no me alcanza, aunque el precio está bien” (“I cannot afford it this month, although the price is fine”) supports 102 and excludes 101.
- “Está caro y este mes no me alcanza” (“It is expensive, and I cannot afford it this month”) supports both when multiple codes are allowed.
If the study requires one main reason, write a selection or manual-review rule. Do not let each coder invent a priority from the order of the words. For more boundary cases, see coding money talk in Spanish surveys, which separates high price, poor value, budget constraints, spending priorities, amounts and materials.
Keep local evidence beside the common definition
Maintain an example bank with the original Spanish, the question, confirmed study market and a short explanation of the assignment. Keep any English translation in a separate field. Ask reviewers familiar with the relevant varieties and subject matter to check difficult cases.
In a synthetic Mexican study, “Ahora no tengo lana para eso” (“I do not have the money for that right now”) can illustrate a budget constraint. The Diccionario del español de México entry for lana documents both money and wool senses. It does not establish that the money use is exclusive to Mexico or identify a speaker's nationality.
Pair that example with “La lana es suave” (“The wool is soft”) in a sweater question. The second response concerns a material. Together, the examples teach a contextual boundary. Country metadata helps organize review; it does not determine the code.
Use a local subcode only when it supports an analytical distinction. Define how it rolls into the shared category and which responses are eligible for it. If the subcode is not collected or supported in another market, distinguish that missing coverage from a genuine zero.
Make change control an analytical step
Keep a dated change register that everyone uses. For each change, record:
- The old rule, new rule and reason
- The approver and release date
- Affected questions, markets, waves and code IDs
- Whether earlier assignments require review
- The decision on historical comparability
Separate editorial corrections from changes to meaning. Fixing a typo differs from splitting a broad “Cost” code into price and budget. Before adopting a split, inspect affected responses from each relevant market and decide whether the original text supports the new distinction.
If historical responses are available, a documented recoding exercise may support the split. If only the old combined counts remain, retain the broader comparable measure and state where the finer series begins. Do not manufacture historical detail that was never captured.
Preserve the original response, prior assignment and revised assignment. A reconciliation file should explain why a reported count changed. Set one release owner and a clear cutoff for the codebook used in each delivery.
Put the approved rules into the workflow
SurveyCoder's coding guide documents imports with code numbers, definitions and examples, question-level Coding Guidance, and a strict-codebook setting that prevents proposals for new codes. Use those controls to express the approved frame. Keep a separate release record: the guide states that Versions retains generated codebooks and does not preserve manual edits.
Before the next delivery, have two reviewers independently apply the same frozen rules to a mixed-market set. Resolve boundary disagreements, update the development version, and repeat with fresh responses. Check overlooked themes and residual categories as well as assigned codes. Our regional Spanish evaluation protocol explains how to reserve a separate test set and assess the resulting workflow.
The handoff should include the released codebook, original question wording, local notes, change log and unresolved cases. Another analyst should be able to explain both a coding decision and any break in comparability.
Follow the step-by-step coding guide to prepare your next multi-country pilot, or start a SurveyCoder account. Use only study data you are authorized to upload, then review the assignments before exporting the accepted file.