In data fusion we consider microdata consisting of records that are composed of information from different sources. Such composite records may consist of several combinations of sources (see the module “Micro-Fusion – Data Fusion at Micro Level”). Records may be a combination of values obtained from a register with values obtained from a survey for the same units (obtained by record linkage). Records may also combine information from several surveys with non-overlapping units, in which case a unit from one source is matched with a similar (but not identical) unit from another source. In addition, records with values obtained from different sources can also arise as a consequence of item non-response and subsequent imputation in which case the two sources are the directly observed values versus the values generated by the imputation method.
In all these cases the composition of a record by combining information obtained from different sources may give rise to consistency problems because the information is conflicting in the sense that edit rules that involve variables obtained from the different sources will often be violated.
The purpose of reconciling conflicting microdata is to solve the consistency problems by making slight changes or adjustments to some of the variables involved. Apart from the choice of variables to be adjusted, an adjustment method should also be specified since there are a number of methods to handle the adjustment problem. In this module three different approaches to the reconciliation problem will be described and the properties of the solutions will be discussed.
To read the entire document, please access the pdf file (link under "Related Documents" on the right-hand-side of this page).
Your feedback is appreciated. Please send your remarks, suggestions for improvement, etc. to email@example.com.