Regression imputation for space-time datasets with missing values

Data consisting in repeated observation on a series of fixed units are very common in different context like biological, environmental and social sciences, and different terminology is often used to indicate this kind of data: panel data, longitudinal data, time series-cross section data (TSCS), spatio-temporal data. Missing information are inevitable in longitudinal studies, and can produce biased estimates and loss of powers. The aim of this paper is to propose a new regression (single) imputation method that, considering the particular structure and characteristics of the data set, creates a “complete” data set that can be analyzed by any researcher on different occasions and using different techniques. Simulated incomplete data from a PM10 dataset recorded in Palermo in 2003 have been generated, in order to evaluate the performance of the imputation method by using suitable performance indicators.