One basic idea in the analysis of spatial data is that values of variables in near-by locations are more similar than values in locations that are further apart, described as spatial dependence or spatial autocorrelation. The latter is not only of statistical relevance, violating the independence assumption in OLS regressions, but also an expression of interesting unobserved inter-dependencies between regions, i.e. spatial spillovers (e.g. how much does the patenting activity in a region affect those in neighbouring regions).


Depending on the nature of data (in STI studies most often area or spatial interaction data), different model classes have come into use. Area data (e.g. regional R&I outputs) is usually analysed by spatial regression models; interaction data (e.g. cross-region R&D collaboration networks) are modelled by spatial interaction models.


For inquiries and support on spatial data analysis and models in RISIS you can contact Martina Neuländtner and Thomas Scherngell at the AIT Austrian Institute of Technology (;