|
Abstract:
|
In a survey context, we often need to know and quantify the main sources of an observed change between two estimations across time. This allows us to explain changes or to point out outlying units with large impact on the change. Depending on the estimator, the type of data, or the number of units, this can be complex. Using implicit derivatives, we will describe a method to find the sources of change, usable in many contexts. Knowing the estimator formula, we need to consider the variables as functions of time, calculate the derivative of the estimator, and approximate the differentials using the differences of the variables. This will help quantify the contribution to the change by variables and units. The advantage of this simple method is to identify easily and precisely which micro-changes have a real impact among all the data.
|
