The effect of microaggregation on regression results: an application to Spanish innovation data

Microaggregation is a technique for masking confidential data by aggregation. The aim of this paper is to analyze the extent to which microaggregated data can be used for rigorous empirical research. In doing this, I adopt an empirical perspective. I use data from the Technological Innovation Panel (PITEC) and compare regression results using both original and anonymized data. PITEC is a new firm-level panel data base for innovative activities of Spanish firms based on CIS data. I find that the microaggregation procedure used has a slight effect on the coefficient estimates and their estimated standard errors, especially when estimating linear models.