Fieldwork quality and data preparation
Fieldwork procedures should make contact, consent, question delivery, probing and recording consistent without pressuring respondents. Train interviewers with realistic examples and a written manual. Verification contacts can identify fabricated or misunderstood records, but should respect the agreed privacy and contact arrangements. Sensitive experiences may require an individual interview rather than a group setting in which participants suppress disagreement.
Create a codebook before producing the final analysis file. Record variable names, units, response codes, skip rules and different reasons for missing values. A valid zero is not interchangeable with refusal, uncertainty or a question that does not apply. Range checks detect impossible values; consistency checks identify incompatible answers across fields. Corrections should be traceable to evidence, with the original record retained under the project’s data-management arrangements.
Distinguish unit nonresponse from missing items within a completed questionnaire. Weighting and imputation address different problems and depend on assumptions; neither guarantees removal of bias. Report the response process, the edits and the limitations rather than treating a clean spreadsheet as proof of valid research. A large operation can contain nonsampling errors, but their direction and magnitude cannot be deduced from sample size alone.