Application of Statistical Disclosure Control methods to protect the confidentiality of the 2020 agricultural census microdata

In this paper, we describe an attempt made to develop an efficient disclosure control algorithm for microdata in a statistical portal used for releasing detailed statistical information at various levels of spatial aggregation. The proposed algorithm is based on perturbative methods, such as microaggregation with Gower's distance for categorical variables and the addition of correlated noise for continuous variables, but it also offers several alternative options in this regard. Moreover, the algorithm can be used to assess the loss of information by measuring distribution disturbances (based on a complex distance that accounts for all measurement scales) and the impact of the Statistical Disclosure Control (SDC) on the strength of correlations between variables (for continuous variables). Through the application of the tools offered by the sdcMicroR package, the algorithm was tested using microdata about agricultural farms and farm animals collected in the 2020 Polish Agricultural Census. We present the results of the tests and discuss the main problems and challenges connected with the use of such tools.