Question coding

Teníamos la encuesta con códigos establecidos por pregunta, ubicada dentro del formulario en el campo de “Nombre De La Columna De Datos”, la encuesta se implementa y empieza su llenado. Tras 12 dias de estar implementada fue necesario realizar una nueva codificación. Se realzia y se implementa. Se pierden las respuestas en la base de datos asociadas a las preguntas re codificadas.
Ya habíamos realizado pruebas con encuesta de prueba haciendo codificaciones y nuevas codificaciones y todo estaba ok.
En ningun caaso, al momento de hacer la implementacion con nuevos codigos, aparecen anuncios de riesgo de perdida de datos o borrado de datos.
¿Cuentan en koboToolbox con sistema para savaguardar estos datos que no se ven al poner códigos (historial)?
¿Existe alguna forma de recuperar la información?

Welcome to the community, @monitoreodigitalCNV! Could you also share with us the coding in the first version and the coding you updated in the second version? This should help the community to understand your issue and help you out.

Thank you for responding to our request. I am not a native English speaker (I speak Spanish). I will specify the report and attach the excel with the codes, accompanied by a photo to show where each code is assigned:

We have been users of Kobotoolbox for years. We have implemented at least 6 surveys and everything has always worked correctly. However, we have had a major problem with a survey we are currently implementing: we have lost data.

We had a new survey with codes added to each question, added into the form in the “Data Column Name” field. We deployed the survey end of September and had respondents filling it. After 12 days of the survey being deployed and respondents filling in data, we needed to adjust some of the coding on some of the questions. We re-coded some of the questions and deployed the survey again. But when we reviewed the responses the next day we discovered that the questions with new codes did not appear in the database and we lost at least 22 surveys with their data. This surprised us because we had already tested this with a test survey, that included coding and re-coding then implementing it, and everything worked correctly. Also because we never received or saw a warning about the risk of data loss or erasure at the moment of the deployment of the new codes.

Please could you help us with answering the following questions:

  1. Does KoboToolbox have a system to save this data that is not visible when coding (data memory)?
  2. Is there some way to recover the information?
  3. We would also like to understand more in depth how the answer is related to a question; on which fields are answers joined to a question (question, code, ID, version, survey, etc)? A reference to clear documentation is also fine here, so we can understand why data was lost, and understand what modifications can and cannot be done at the several stages of a survey.

Additionally, we have noticed some errors in this survey that have never happened before: the data isn’t fully being shown (in yes/no answers, for some respondents we only see “y” instead of the full “yes” or “n” instead of the full “no”) and the languages of the survey questions is being mixed up (the survey is deployed in Spanish, but some questions appear in English). This isn’t a major problem but we thought we would mention it in case it is related, as KoboToolbox has worked so perfectly for us until now.

Thanks again,

Maria Paula

Kobo coding CNV (Mining).xlsx (31.2 KB)

Dear Kal, is there an answer to our request?