Survey programming and data management
We design and program the questionnaire, clean the file, document it, and deliver it in the format your analysts already use.
Most of the pain in a Jamaican survey sits between the Word questionnaire and the file the analyst opens. Balcostics Research does that stretch: design, programming, testing, cleaning, documentation and delivery.
The managing director, an economist, works in SPSS, Stata, R, SurveyCTO and CSPro. Tablet programming, skip logic and questionnaire testing are led by our CAPI and data manager, with the firm since 2015. Names and roles are on the about page. This page stays with the work.
Questionnaire design
We review your draft for length, order, skips and words that will not survive a doorstep or a phone call. Show-cards, roster loops and "other, specify" lines are checked before anyone is trained. Where you want a myVoice online questionnaire instead of a field instrument, the Research Hub path is on how myVoice works, including the questionnaire designer at 30 points a question.
We also draft instruments. On a donor job we still expect your technical lead to sign off the final wording. We do not replace the principal investigator.
Programming and skip logic
The programmed instrument is tested on the devices interviewers will carry. Skips, ranges and roster limits are part of that test, not a cleanup task after fieldwork.
Three programming routes are in regular use.
- The myVoice field collector for offline CAPI: case management, household roster, random respondent selection, GPS and per-question timing. Android only. Study owners pay 10,000 points a month. Interviewers are free.
- CSPro, which we used as prime field contractor on UNICEF MICS 2022.
- SurveyCTO, when the client programme requires it.
Online surveys in the Research Hub have their own skip logic. That product is separate from the field collector. A Hub download and a field file are separate pipelines.
Cleaning
Cleaning follows the instrument. We check IDs, parish and enumeration codes, skip violations, roster consistency, and interviews flagged on GPS or timing. On the field collector a supervisor flag fires when GPS at the interview is more than 500 metres from the address. A flag is a review item. Range checks and "other" text are reviewed against the codebook, not silently overwritten.
Edits are listed. The field report says what changed and why. We would rather hand you a short edit log than a spotless file with no history.
Weighting support
If the contract includes weights, we support frame description, selection documentation and weight computation under the project manager. A consulting statistician has supported frame construction, enumeration-district selection and weights on our work.
Send the sample specification with the questionnaire. We program to it. We do not invent a design after fieldwork to rescue a file.
Codebooks
The codebook names every variable the way the program names it, states the question text, the valid codes, the skip that feeds the item, and any recode applied in cleaning. Analysts should be able to move from a table cell back to the question without calling us. If we built the instrument in CSPro, SurveyCTO or myVoice, the codebook reflects that build, not an earlier Word draft that drifted.
Data entry
Paper surveys, and any paper listing sheets, are keyed. Double-check rules are agreed before entry starts: full double entry, or entry plus verification on a sample, as the contract states. Keyed files join the same cleaning and codebook path as CAPI files.
Secure delivery
You receive the dataset by the method written in the contract. Typical working files are CSV, SPSS and Stata. Access on our side stays with named staff. Devices used in the field are encrypted. Interviewers do not keep a personal copy after the job.
The contract states a disposal date for respondent-level files. Disposal follows the Data Protection (Disposal of Personal Data) Regulations, 2024, as described on the data protection page. We do not paste project microdata into a public AI chat. Open-answer sentiment inside myVoice is a separate product feature. It calls an external model, and the export from that product does not include names or email addresses.
If you only need programming and a clean file, say so. Fieldwork is optional. Many clients hire both, because the people who programmed the skips are the people who train the interviewers. The data collection page covers fieldwork and programming together.
Questions buyers ask
Can you program a questionnaire we have already drafted?
Yes. We review wording, flow and skip logic with you, then program and test it on our devices before interviewer training. If your programme requires CSPro or SurveyCTO, we build it there. Otherwise we build it in the myVoice field collector.
Do you calculate survey weights?
We support weighting and frame work under the project manager's direction. Bring your sampling specification and we will program and clean to it.
What file do we receive?
A cleaned dataset, a codebook that matches the instrument, and a field report. Usual working formats are CSV, SPSS and Stata. Delivery follows the method in the contract. Respondent-level files have a written disposal date.
