I remotely oversee a group of enumerators collecting data in the field (usually places with no internet until the evenings) using Kobo Collect loaded on tablets.
We had build the forms last August and they successfully did a pilot using these forms in September. So the database already has over a hundred entries.
Due to changes in our organization and the sampling strategy, I needed to change some of the choices for some questions in the forms (not the questions - no questions were changed, deleted, or added). I made the changes and then redeployed the forms. Now the enumerator team will restart their data collection in a few weeks.
Questions:
Is it correct that there will be no problems adding to the same database given that in this second round of data collection, a few of the questions have different choices? For these questions, I know that the data collected before would not be directly comparable to that collected now, but I just want to ensure that the database won’t get corrupted and the columns will still line up.
Do the enumerators need to re-download the forms onto their tablets or will the update to the form be done automatically when they connect to the internet? I don’t know enough about how Kobo Collect functions to answer this.
Suppose you have a question and choices as follows in Round 1 survey Q1. What is your favorite fruit? and have choices as 1. Apple, 2. Orange, 3. Banana. You now wish to update some choices for the same question in Round 2. If you update the choices as 1. Apple, 2. Orange, 3. Banana, 4. Cherry, 5. Coconut. It shouldn’t be a problem (as you are only adding the choices with the same code and label). Or you could also update your choices safely as 1. Apple, 3. Banana, 4. Cherry, 5. Coconut (if you have to update the choices by adding new choices and removing unwanted choices that existed previously). Here you will not see choice 2. Orange. But if you update the choices as 1. Cherry, 2. Coconut, 3. Apple, 4. Banana then that could be an issue later on as you are mixing up the choices label. You will need to careful with choices name (value label) that has already been used for any variable.
The same applies with the variable name. If you have already used it once, means it’s not safe to re-use the same variable name with for that project. You will need to use a different variable name. E.g. you have Q1. What is your favorite fruit? with the variable name Q1. If you wish to remove this variable and have another question you should not use Q1 for the new question you are planning. Take another variable name like Q10 or Q100 so something like that which is unique for this project.
It’s good to re-download the forms into the device manually as sometimes you may think it was automatically updated but in fact you could still be using the old version. It is also good to keep a version name or number in your survey project to track the form version.
Besides, please try this out in a dummy project before you implement them in your real project. This could affect all your data. If you are still not confident with this, maybe it’s still safe to have 2 different projects as baseline and endline. With this you should have all your data safe (if and incase you are not confident to make changes to your project).
Thank you, this information is exactly what I was looking for. For 1., I deleted some choices and added some new ones, but I did not change labels of existing choices, so I hope it will be okay.
As you suggested, we will try this out with the new forms to ensure it is uploading to the database properly, before doing it in the field.
