Möbus, Claus (1981) Zur Beschreibung und Analyse kurz- und langfristiger Testintelligenzveränderungen mit zeitdiskreten und zeitkontinuierlichen dynamischen Modellen. In: Bericht über den 32. Kongreß der Deutschen Gesellschaft für Psychologie in Zürich 1980. Verlag für Psychologie, Hogrefe, Göttingen, pp. 185-188. ISBN 978-3-8017-0176-5

Preview

- Published Version Volltext (1169Kb)

Abstract

Description and analysis of short- and longtimed changes in intelligence test scores with time-discrete and time-continous dynamic models: Lonqitudinal studies or repeated measurements seldom include process models of the empirical phenomenon. Evaluation of change or intervention effects is nearly always done by static statistical methods (e.g. ANOVA). Static models, however, are not very adequate for the planning of optimal intervention schedules. Especially if the statistical results are nonsignificant we should explain this finding in terms of dynamic processes: (a)was the intervention not effective at all? (b) was the intervention effective but too short? (c) was the change in the dependent variable only transitory, so that the intervals between observations were too long? It is the aim of this paper to demonstrate how some of these questions could be answered. We analyze a short-termed intelliqence test coaching and crossvalidate the results with a nonexperimental longitudinal study. It can be shown that a simple crossvalidation of the experimental results fails. It is possible, however, to crossvalidate the results of a simulation study which is based on the experimental findings. This confirms the hypothesis, that the experimental study was too short for stabilizing the effects on the dependent variables. It suggests itself, that many unsuccesful cross-validation studies could be explained in a similar fashion. The simulation study was based on a time-discrete model (difference equation system),whereas the study of intervention effects and the development of hypotheses at any time necessitates the development of time-continous models. This is especially true, when the intervals between measurements are very long. These models are based on differential equation systems with constant coefficients. It is shown, how to identify and estimate the parameters with panel or two-wave data. Identification is done via the LAPLACE-transformation of the system and the comparison of the input-output equation of the model with its state-variable represention.

Item Type:	Book Section
Uncontrolled Keywords:	short- and longtimed changes, time-discrete and time-continous dynamic models, lonqitudinal studies, process models, evaluation of change or intervention effects, difference equation system, differential equation system, identification, LAPLACE-transformation, state-variable represention
Subjects:	Generalities, computers, information > Computer science, internet Philosophy and psychology > Psychology
Divisions:	School of Computing Science, Business Administration, Economics and Law > Department of Computing Science
Date Deposited:	23 Oct 2015 09:19
Last Modified:	23 Oct 2015 09:19
URI:	https://oops.uni-oldenburg.de/id/eprint/2529
URN:	urn:nbn:de:gbv:715-oops-26109
DOI:	10.13140/RG.2.1.2691.9127
Nutzungslizenz:

Actions (login required)

View Item