[Various methods of dynamic functional analysis in human sciences and economics].
Including the temporal and developmental dimension into the measurement of human conduct is a fundamental concern for those who do research in natural surroundings. Observing an individual day after day may possibly give a more complete vision of how behavior works than measuring a group of individuals at a single time and analyzing the differences found among them. Unfortunately most of the tools allowing analyzing individual time series call for large numbers of repeated observations. Thus, practicable longitudinal research designs often do not involve either enough repeated measurements for traditional time series analyses nor either replicate enough individuals for traditional, large-sample analyses. Dynamic factor analysis is a rationale and procedure for both pooling relatively short time series information across limited numbers of participants and analyzing the pooled information for its dynamic, process-relevant elements. It is a merging of two important analytical tools - multivariate time series and the common factor model, from which it distinguishes itself mainly by the fact that in dynamic factor analysis, the values of the common factors can influence the values of the observed variables both concurrently and in delayed fashion. Dynamic factor analysis is actually a method which allows detecting structures in the time series as well as the relations between the series and the explanatory variables. We illustrate the different models used in psychology and social sciences, as well as in econometry and economics.