KoboToolbox Data-Collection Readiness Checklist

You are not ready to collect data yet. Go back to the questionnaire logic, translation, consent and data-protection plan, and a proper pilot before treating this as a go/no-go gate. Deploying now would put both your data and your respondents at risk.

13 critical item(s) still unchecked

Questionnaire & XLSForm design

• Every question maps to a specific indicator, research question, or decision — questions that no one will analyse have been cut.
• Each question uses the right field type (select_one, select_multiple, integer, decimal, date, geopoint, text), and numeric fields have sensible min/max constraints.
• All skip logic and relevant conditions have been tested so every respondent sees only the questions that apply to them.critical
• Genuinely mandatory questions are set as required, and questions where 'no answer' or 'prefer not to say' is valid are not forced.
• Validation constraints and constraint messages catch impossible values (negative ages, dates in the future, household size of zero) at the point of entry.
• Answer option lists are complete, mutually exclusive where needed, and include 'Other (specify)', 'Don't know', and 'Refused' codes where appropriate.
• Hidden metadata is captured — enumerator ID, start/end time, GPS where permitted, and a unique respondent or household ID for tracking and de-duplication.

Translation & language

• The form is fully translated into Burmese and any local languages your respondents speak (e.g. Sgaw Karen, Mon, Rakhine, Shan, Chin, Kachin, Rohingya) — not left for enumerators to translate on the spot.
• Key questions have been back-translated or reviewed by a second speaker to confirm the local-language wording carries the same meaning as the English/Burmese source.
• Multiple languages are set up as label columns in the XLSForm so enumerators can switch language inside Kobo Collect rather than juggling separate forms.
• Sensitive or technical terms (consent, disability, gender-based violence, income) have been checked with local staff to ensure they are accurate, respectful, and not stigmatising in each language.
• Burmese and local-script text renders and wraps correctly on the actual devices, including Unicode (not Zawgyi) encoding throughout.

Consent & data protection

• An informed-consent statement is built into the form — purpose, who is collecting, how data will be used and stored, voluntariness, and the right to stop at any time — and the survey cannot proceed without a recorded consent response.critical
• There is a clear, separate protocol for consent from children, and for assent plus guardian consent or extra safeguards for vulnerable respondents.critical
• Data minimisation has been applied — you collect only what you need, and you have justified collecting any directly identifying information (names, phone numbers, exact GPS, photos).critical
• Sensitive personal data (ethnicity, religion, health, protection or GBV disclosures, political opinion) is collected only where essential, with a referral pathway and do-no-harm plan in place.critical
• The project is set to encrypt submissions where supported, server access is restricted to named staff, devices are PIN/password-locked, and a clear data-retention and deletion plan exists.critical
• A plan is agreed for anonymising or pseudonymising data before analysis or sharing, and for who may access raw identifiable data.critical
• Any required ethics, IRB, donor, or local-authority data-collection approvals and permissions are obtained and documented before fieldwork starts.

Enumerator training & preparation

• Enumerators are recruited with the right language skills, gender balance, and community acceptance for the areas and respondents involved.
• Every enumerator has hands-on practice with Kobo Collect on the actual device — installing the form, navigating, saving drafts, and submitting.
• The team has been trained question by question so everyone reads and probes each question the same way and understands the intended meaning.
• Enumerators are trained to deliver consent properly, handle sensitive topics, respect refusals, and follow safeguarding and referral procedures.critical
• Each enumerator has completed practice interviews or role-plays and been observed before being cleared for real data collection.
• Field protocols are clear — respondent selection, call-backs, refusals, safety and security check-ins, and who to contact when something goes wrong.

Devices & offline setup

• Kobo Collect (or the ODK app) is installed on every device and configured to connect to the correct server and project.
• The final approved form version is downloaded onto every device, and old or test versions have been removed to prevent mixed-version data.critical
• Offline data entry has been tested end to end — enumerators can complete and save interviews with no internet and sync later without losing records.critical
• Devices have enough free storage, working batteries, and power banks or charging arranged for full field days with limited electricity.
• A realistic sync plan exists for low-connectivity townships — where, when, and how submissions will be uploaded, and who confirms they arrived.
• Spare devices, SIMs/data, chargers, and paper backup forms are prepared in case of device failure or app crashes in the field.

Sampling & deployment

• The sampling approach (random, cluster, purposive, quota) is documented with a defined sampling frame and target sample size per area or group.
• Quotas and respondent selection rules are clear so each enumerator knows exactly who to interview and how many, avoiding gaps and double-counting.
• The sample design lets you disaggregate results as required (sex, age, disability, location, displacement status) and meets donor or analysis needs.
• Access, permissions, and security for each location have been checked against the current situation, with go/no-go and contingency plans for areas that become unreachable.
• The form is deployed (not in draft), submissions are confirmed to land in the right project, and the collection start date and schedule are set.

Data quality & backup

• The whole instrument has been piloted with real respondents in field conditions, and the form was revised based on what the pilot revealed.critical
• All pilot and test submissions have been deleted from the live project so they cannot contaminate the real dataset.critical
• A daily data-quality routine is set — reviewing incoming submissions for missing fields, outliers, duplicates, and suspiciously fast or identical interviews.
• There is a clear channel to give enumerators fast feedback and corrections during collection, not only after the round ends.
• Data is backed up regularly — submissions exported or downloaded and stored securely in at least two places — so a lost or wiped server cannot destroy the dataset.critical
• Roles are assigned for data management, cleaning, storage, and final deletion, with one named person accountable for the dataset's safety.