Skip to main content
Data quality should be actively monitored during data collection. This is recommended because it allows you to do corrections on the data while the enumerators are still in the field. KoboToolbox provides automatic visualisations that can be used to track the responses that come in and act quickly when an outlier is detected. You can do this tracking using the KoboToolbox feature of automatic charts through the “Data” menu, then choosing “Reports”. Here you will see visualisations of each variable collected. The survey is very extensive so tracking every question will not be feasible. We recommend paying attention to the following variables, as they are most important to be of good quality for the LIP calculation:
  • Quantity produced
  • Quantity sold
  • Measurement units
  • Price
  • Questions that will lead to labour cost
    • Number of labourers
    • Number of days for which labour is hired
    • Salary per day
  • Land size
  • Costs: general farm costs, input costs, other costs
  • Family size
What to look out for?
Anything that seems very unusual for the local context. Talk to your local partners to determine what is possible and try to establish reasonable ranges for the key variables mentioned above. Some questions to answer together may include:
  • What is the largest realistic quantity of crop that farmers can produce?
  • Are there some local measurement units that are likely to be used by farmers?
  • What is the maximum and minimum realistic price?
  • What is the most a temporary labourer might get paid per day of labour?
Once you know what is realistic, you can start to flag values outside of these ranges. Remember, collected data sometimes does not confirm what we believe, so it is important to be cautious when removing or replacing values. Correcting the Data
Supervisors can directly contact enumerators to resolve any issues or to request clarification on unusual data entries. Ideally, the data gets corrected directly in KoboToolbox or the data collection app that is being used. Remember that we also collect phone numbers through the survey, so in some cases it is also possible to correct values over the phone. When you are conducting a very large data collection, it might be worthwhile to connect your KoboToolbox database to a dashboard with your key variables of interest. An example of such a dashboard developed in Google Looker Studio can be found here.