Sessie
onderzoeksgroep
Longitudinale Modellen (LOMO)
14.10-14.35
Siem-Jan
Koopman, André Lucas, Marius Ooms, Kees van Montfort en Victor van der Geest
Estimating systematic continuous time trends in
recidivism using a non-Gaussian panel data model*
14.35-15.00
Elmar Schlueter, Oliver Christ, Peter Schmidt
en Ulrich Wagner
Putting the Autoregressive Latent
Trajectory model to practice: A test of the longitudinal relations between
anti-immigrant prejudice and discriminatory behaviorial intentions in Germany
15.35-15.15
Pauze
15.15-15.40
Eva Jaspers, Marcel
Lubbers en Duane Alwin
Attitudes around homosexuals:
contact effects from a life-course perspective
15.50-16.05
Jeroen Vermunt en Gregory Palardy
Multilevel Growth Mixture Models for
Classifying Group-level Observations
*Paper wordt gepresenteerd door Marius Ooms
Estimating systematic
continuous time trends in recidivism using a non-Gaussian panel data model
Siem-Jan Koopman
Department of Econometrics, Vrije Universiteit
André Lucas
Department of Finance, Vrije Universiteit
Marius Ooms
Department of Econometrics, Vrije Universiteit
Kees van
Montfort
Department of Finance, Vrije Universiteit
Victor van der
Geest
Netherlands
Institute for the Study of Crime and Law Enforcement (NCSR)
We model crime
careers of juveniles from a Dutch Judicial Juvenile Institution. The data are
decomposed into a sytematic and a an individual-specific component, of which
the systematic component reflects the general time-varying conditions including
the ciminological climate. Within a model-based analysis, we treat (1) shared
effects of each group with the same systematic conditions, (2) strongly
non-Gaussian features of the individual time series, (3) unobserved common
systematic conditions, (4) changing recidivism probabilities in continuous
time, (5) missing observations. We adopt a non-Gaussian multivariate state
space model that deals with all of these issues simultaneously. The parameters
of the model are estimated by Monte Carlo maximum likelihood methods. This
paper illustrates the methods empirically. We find interesting time-variation
in the recidivism behavior of the juveniles during a perios of 13 yeard, while
taking account of significant heterogeneity determined by personality characteristics
and initial crime records.
Putting the
Autoregressive Latent Trajectory Model to Practice:
A Test of the
longitudinal relations between anti-Immigrant prejudice and discriminatory
behaviorial intentions in Germany
Elmar Schlueter
DFG-Research Training School ‘Group-focused enmity’,University of
Marburg
Oliver Christ
Department of Psychology, University of Marburg
Institute for Interdisciplinary Research on Conflict and Violence,
University of Bielefeld
Peter Schmidt
Department of Social Sciences,University of Giessen
Ulrich Wagner
Department of Psychology, University of Marburg
The aim of this
presentation is to demonstrate the application of the autoregressive latent
trajectory model (ALT, Bollen & Curran 2004) to a large-scale panel data
set. The ALT has been proposed as a combination of the autoregressive- and the
latent trajectory model as two established methods for the analysis of panel
data. However, substantive applications of the ALT are still largely missing.
To illustrate the potentialities of the ALT, we use this approach to examine
the longitudinal relations between ethnic prejudice and discriminatory
behavioral intentions using three-wave panel data from the German general
population.
Attitudes around
homosexuals: contact effects from a life-course perspective
Eva
Jaspers
Marcel
Lubbers
Sociologisch
Instituut, Radboud University Nijmegen
Duane Alwin
Pennsylvania State University
The Netherlands
is known for its unusually tolerant climate toward gay men and lesbians. This
paper addresses the dramatic shift in the attitudes toward homosexuals in the
Netherlands over the second half of the twentieth century. It draws on
sociological theories on structural positions and attitudes toward homosexuals,
as well as on psychological experiments on contact with homosexuals. There is
ample evidence that contact with homosexuals has a positive influence on the
attitude toward homosexuals. However, cross-sectional studies on the impact of
contact can not disentangle the causality of the positive association between
contact and attitudes. Experimental designs tend to use highly selective groups
and measure attitudes over a very short period of time. This paper is
innovative, as it draws on retrospective data on attitudes toward homosexuals.
Dutch respondents are asked for their opinions on gay men and lesbians at age
18, age 30, age 50 –all of these when applicable- and their present attitude.
They have also provided information on gay friends or relatives in their
network, and the ages at which these ties came into being. As we are well aware
of the controversy regarding the retrospective measurement of attitudes, we
perform various checks to find out to what extent the data are corrupted by the
research design. We use multi level growth curve models to (1) address how
attitudes towards homosexuals develop over the life-span and how structural
positions influence these attitudes; and (2) test hypotheses about experiences
with gay friends and relatives, and their timing in the development of
attitudes toward gay men and lesbians. We also evaluate the use of
retrospective data on attitudes for research on the life-span development of
attitudes.
Multilevel Growth
Mixture Models for Classifying Group-level Observations
Jeroen K.Vermunt
Department of Methodology and
Statistics,Tilburg University
Gregory J. Palardy
University of Georgia
This paper introduces a new multilevel
growth mixture model (MGMM) for which
latent classes can be extracted from both the within- and between-levels of
analysis, whereas prior presentations of the MGMM had limited classifications
to within-level observations. Nine variations of the general model are
describes that differ in terms of categorical and continuous latent variable
specification at the within- and between-level of analysis. We provide a
detailed example of an application of this model for studying school effects
using data from the Early Childhood Longitudinal Study (ECLS). A
multilevel piecewise growth model that partitions out summer learning from
school-year learning is employed. A conditional MGMM that controls for
difference in student background and school social composition is used to
classify schools into effectiveness categories based on their mean student
learning trajectories. The multinomial effectiveness categories are
regressed on a set of school practice variables to
investigate predictors of latent class membership of schools. We discuss
various issues related to model selection and model specification.