[Executive function in early childhood and Generalizability Theory: Typical babies and preterm babies].

In the areas of Social, Health, and Behavioral Sciences, observational designs are used as one of the best ways to study human behavior in natural and quasi-natural settings. Observing and recording a sample of individual patterns of behavior is a feature common to all of them, but a broad sampling error can be implicit, especially at early ages. Sampling errors arise also as a result of the observation of only a fraction of each possible case, and this is part of the problem involved in the estimation of generalizations of data recorded by systematic observation. The sampling process usually takes place at different stages (observers, sessions, tasks, age, etc...) and requires estimating variance components that can be combined with each other to produce one or more estimates of board coefficients. The Generalizability Theory (GT) proposes that there are many (infinite) sources of variation (facets) in any measurement. Our main objective is to identify the variance components that contribute to estimation errors and implement strategies to reduce the influence of these error sources on the measurement. Executive functioning in typical and premature babies is studied at 15 and 24 months. The results show differences between groups, sources of variation, and generalizability.