Testing informative hypotheses in SEM increases power: An illustration contrasting classical hypothesis testing with a parametric bootstrap approach

In the present paper, the application of a parametric bootstrap procedure, as described by van de Schoot, Hoijtink, and Deković (2010), will be applied to demonstrate that a direct test of an informative hypothesis offers more informative results compared to testing traditional null hypotheses against catch-all rivals. Also, more power can be gained when informative hypotheses are tested directly. In this paper we will (a) compare the results of traditional analyses with the results of this novel methodology; (b) introduce applied researchers to the parametric bootstrap procedure for the evaluation of informative hypotheses; and (c) provide the results of a simulation study to demonstrate power gains when using inequality constraints. We argue that researchers should directly evaluate inequality-constrained hypotheses if there is a strong theory about the ordering of relevant parameters. In this way, researchers can make use of all knowledge available from previous investigations, while also learning more from their data compared to traditional null-hypothesis testing.

[1]  M. Lee,et al.  Model selection for the rate problem: A comparison of significance testing, Bayesian, and minimum description length statistical inference , 2006 .

[2]  M. Kenward,et al.  An Introduction to the Bootstrap , 2007 .

[3]  Herbert Hoijtink,et al.  Testing Inequality Constrained Hypotheses in SEM Models , 2010 .

[4]  Jan-Willem Romeijn,et al.  Evaluating expectations about negative emotional states of aggressive boys using Bayesian model selection. , 2011, Developmental psychology.

[5]  C. Spiel,et al.  The goal to be accepted by friends as underlying function of overt aggressive behaviour in immigrant adolescents. , 2012, Scandinavian journal of psychology.

[6]  D. Andrews.  Inconsistency of the Bootstrap when a Parameter is on the Boundary of the Parameter Space , 2000 .

[7]  Jeroen K. Vermunt,et al.  The order-restricted association model: Two estimation algorithms and issues in testing , 2004 .

[8]  Herbert Hoijtink,et al.  Equality and inequality constrained multivariate linear models: objective model selection using constrained posterior priors , 2010 .

[9]  Jacob Cohen.  The earth is round (p < .05) , 1994 .

[10]  H. Hoijtink,et al.  Bayesian Evaluation of Informative Hypotheses. , 2008 .

[11]  E. Wagenmakers,et al.  Bayesian hypothesis testing for psychologists: A tutorial on the Savage–Dickey method , 2010, Cognitive Psychology.

[12]  Rens van de Schoot,et al.  Bayesian model selection of informative hypotheses for repeated measurements , 2009 .

[13]  Paula J. Fite,et al.  Replication and extension of Little et al.'s (2003) forms and functions of aggression measure , 2008 .

[14]  Todd D. Little,et al.  Disentangling the “whys” from the “whats” of aggressive behaviour , 2003 .

[15]  M. Lee,et al.  Bayesian statistical inference in psychology: comment on Trafimow (2003). , 2005, Psychological review.

[16]  Jeroen K. Vermunt,et al.  Testing log-linear models with inequality constraints : A comparison of asymptotic, bootstrap, and posterior predictive P values , 2005 .

[17]  Conor Dolan,et al.  On the likelihood ratio test in structural equation modeling when parameters are subject to boundary constraints. , 2006, Psychological methods.

[18]  R. Tsonaka,et al.  Parameter constraints in generalized linear latent variable models , 2007, Comput. Stat. Data Anal..

[19]  Herbert Hoijtink,et al.  Inequality constrained analysis of variance: a Bayesian approach. , 2005, Psychological methods.

[20]  Dagmar Strohmeier,et al.  Bullying and Victimization Among Native and Immigrant Adolescents in Norway , 2009 .

[21]  Ronald Schoenberg,et al.  Constrained Maximum Likelihood , 1997 .

[22]  Y. Ritov,et al.  Analysis of contingency tables by correspondence models subject to order constraints , 1993 .