Örtük Markov Analizlerinde Model Seçimi Üzerine Bir Monte Carlo Simülasyon Çalışması

Örtük Markov modelleri, niceliksel olarak ölçüm almanın mümkün olmadığı boylamsal psikoloji çalışmalarında,kategorik gözlenen ve örtük değişkenlerin zamana bağlı değişimlerini analiz etmek ve yorumlamak için iyi bir alternatif olarak karşımıza çıkmaktadır. Ancak, son yıllarda giderek artan kullanıma rağmen Örtük Markov modellerindeki model seçim süreci konusunda henüz bir görüş birliğine varılamamıştır. Bu bağlamda, çalışmanın ilk amacı, tekdeğişkenli bir görgül veri seti kullanarak bir uygulama örneği sunmaktır. Bir diğer amacı da Monte Carlo simulasyonyöntemi ile oluşturulmuş veri setinden yararlanarak, madde tepki olasılıklarının gücü, ölçüm alınan zaman sayısı veörneklem büyüklüğüne göre bilgi kriterlerinin model seçimi ve parametre tahmin yanlılıklarına etkisini incelemektir.Sonuçlara göre, madde tepki olasılıklarının hem güçlü hem de zayıf olduğu koşullarda, ölçüm alınan üç farklı zamankoşulunda ya da örneklem büyüklüğü 200’den 2000’e yükseldiğinde BIC ve CAIC bilgi kriterleri kullanılarak yapılan model seçimlerinde %100 doğru karar oranı gözlenmiştir. Bulgular, alanyazın ışığında tartışılacaktır.

A Monte Carlo Simulation Study on Model Selection in Latent Markov Models

Latent Markov models emerge as a good alternative for longitudinal psychological studies where it is not possible to take quantitative measurements to analyze and interpret time-dependent changes of categorical observed and latent variable(s). However, despite its increasing use in recent years, a consensus on the model selection process in the Latent Markov models has not been reached yet. In this context, the first objective of this research was to provide an application example by using an empirical dataset with a single variable. Another aim was to examine the impacts of the strength of item response probabilities, the number of times the measurement being taken and sample size on model selection and parameter estimation bias based on using the dataset generated by Monte Carlo simulation method. As a result, using BIC and CAIC information criteria, 100% correct decision rate was observed regardless to item response probabilities (weak or strong), and number of measurement (2 or 3) when the sample size increased from 200 to 2000. The findings were discussed in the light of the related literature

___

  • Bartolucci, F., Farcomeni, A. ve Pennoni, F. (2013). Latent Markov models for longitudinal data. Boca Raton, Florida: Chapman & Hall/CRC. Bartolucci, F. ve Pandolfi, S. (2016). LMest package. (01.06.2017 tarihinde tarihinde https://cran.r-project.org/web/packages/LMest/index.html adresinden alınmıştır.)
  • Bauer, D. J. ve Curran, P. J. (2003). Distributional assumptions of growth mixture models: Implications for overextraction of latent trajectory classes. Psychological Methods, 8, 338–363.
  • Cain, A. S., Epler, A. J., Steinley, D. ve Sher, K. J. (2010). Stability and change in patterns of concerns related to eating, weight, and shape in young adult women: A latent transition analysis. Journal of Abnormal Psychology, 2, 255–267.
  • Castellini, G., Fioravanti, G., Lo Sauro, C., Rotella, F., Lelli, L., Ventura, L., . . . Ricca, V. (2013). Latent profile and latent transition analyses of eating disorder phenotypes in a clinical sample: A 6-year follow-up study. Psychiatry Research, 92–99.
  • Catts, H. W., Compton, D., Tomblin, J. B. ve Bridges, M. S. (2012). Prevalence and nature of late-emerging poor readers. Journal of Educational Psychology, 104(1), 166–181.
  • Collins, L. M. ve Flaherty, B. P. (2002). Latent class models for longitudinal data. Jacques A. Hagenaars ve Allan L. McCutcheon (Ed.) Applied Latent Class Analysis içinde (287–303). U.K.: Cambridge University Press.
  • Collins, L. M. ve Lanza, S. T. (2010). Latent class and latent transition analysis with applications in the social, behavioral, and health sciences. Hoboken, NJ: Wiley.
  • Collins, L. M. ve Wugalter, S. E. (1992). Latent class models for stage-sequential dynamic latent variables. Multivariate Behavioral Research, 27(1), 131–157.
  • Cosden, M., Larsen, J., Donahue, M. T. ve Nylund-Gibson, K. (2015). Trauma symptoms for men and women in substance abuse treatment: A latent transition analysis. Journal of Substance Abuse Treatment, 50, 18–25.
  • Dias, J. G. (2006). Latent class analysis and model selection. M. R. Kruse, C. Borgelt, A. Nürberger ve W. Gaul (Ed.). From data and information analysis to knowledge engineering içinde (95–102). Berlin: Springer-Verlag.
  • Dural, S., Somer, O., Korkmaz, M., Can, S. ve Ogretmen, T. (2010). Zamana bağlı değişimin incelenmesi: Örtük gelişme modelleri. Türk Psikoloji Dergisi, 25(65), 1–13.
  • Gudicha, D. W., Schmittman, V. D. ve Vermunt, J. K. (2015). Power computation for likelihood ratio tests for the transition parameters in latent Markov models. Structural Equation Modeling: A Multidisciplinary Journal, 1–12.
  • Guo, B., Aveyard, P., Fielding, A. ve Sutton, S. (2009). Using latent class and latent transition analysis to examine the transtheoretical model staging algorithm and sequential stage transition in adolescent smoking. Substance Use & Misuse, 2028–2042.
  • Güngör, D., Korkmaz, M. ve Sazak, H. S. (2015). Örtük sınıf analizi ile yapılan ölçme eş değerliği çalışmalarında model seçimi. Hacettepe Eğitim Fakültesi Dergisi, 30(1), 90–105.
  • La Flair, L. N., Reboussin, B. A., Storr, C. L., Letourneau, E., Green, K. M., Mojtabai, R., . . . Crum, R. M. (2013). Childhood abuse and neglect and transitions in stages of alcohol involvement among women: A latent transition analysis approach. Drug and Alcohol Dependence, 132, 491–498.
  • Lanza, S. T. ve Bray, B. C. (2010). Transitions in drug use among high-risk women: An application of latent class and latent transition analysis. Advances and Applications in Statistical Sciences, 3(2), 203–235.
  • Lanza, S. T. ve Collins, L. M. (2008). A new SAS procedure for latent transition analysis: Transition in dating and sexual behavior. Developmental Psychology, 44, 446–456.
  • Lin, T. H. (2006). A comparison of model selection indices for nested latent class models. Monte Carlo Methods and Applications, 12(3-4), 239–259.
  • Marcoulides, G. A., Gottfried, A. E., Gottfried, A. W. ve Oliver, P. H. (2008). A latent transition analysis of academic intrinsic motivation from childhood through adolescence. Educational Research and Evalution: An International Journal on Theory and Practice, 14(5), 411–427.
  • Muthen, B. ve Muthen, L. (1998-2007). M plus user’s guide fifth edition. Los Angeles: Muthen ve Muthen.
  • Muthen, L. K. ve Muthen, B. O. (2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling: A Multidisciplinary Journal, 9, 599–620.
  • Nylund, K. L., Asparouhov, T. ve Muthen, B. O. (2007). Deciding on the number of classes in latent class analysis in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14(4), 535–569.
  • Pelsmacker, P. D. ve Janssens, W. (2007). The effect of norms, attitudes and habits on speeding behavior: Scale development and model building and estimation. Accident Analysis and Prevention, (39) 6–15.
  • Schmittman, V. D., Dolan, C. V., van der Maas, H. L. J. ve Neale, M. C. (2005). Discrete latent Markov models for normally distributed response data. Multivariate Behavioral Research, 40(4),461–488.
  • Seaton, E. K., Yip, T., Morgan-Lopez, A. ve Sellers, R. M. (2012). Racial discrimination and racial socialization as predictors of African American adolescents’ racial identity development using latent transition analysis. Developmental Psychology, 48(2), 448–458.
  • Vermunt, J. K., Langeheine, R. ve Bockenholt, U. (1999). Discrete-time discrete-state latent Markov models with time-constant and time-varying covariates. Journal of Educational and Behavioral Statistics, 24, 179–207.
  • Vermunt, J. K. ve Magidson, J. (2013). LG-Syntax User’s Guide: Manual for Latent GOLD 5.0 Syntax Module. Belmont, MA: Statistical Innovations Inc.
  • Vermunt, J. K. ve Magidson, J. (2015). Technical Guide for Latent GOLD 5.0: Basic, Advanced, and Syntax. Belmont, MA: Statistical Innovations Inc.
  • Vermunt, J. K. ve Magidson, J. (2016). Technical Guide for Latent GOLD 5.1: Basic, Advanced, and Syntax. Belmont, MA: Statistical Innovations Inc.
  • Vermunt, J. K., Tran, B. ve Magidson, J. (2008). Latent class models in longitudinal research, Handbook of longitudinal research: Design, measurement, and analysis. Menard, S. (Ed.) içinde (s. 373-385). Burlington: Elsevier.
  • Visser, I., Rajimakers, E. J. ve Molenaar, P. C. M. (2002). Fitting hidden Markov models to psychological data. Scientific Programming, 10, 185–199.
  • Vrieze, S. I. (2012). Model selection and psychological theory: A discussion of the differences between the Akaike information criterion and the Bayesian information criterion. Psychological Methods, 17(2), 228–243.
  • Warner, H. W. ve Aberg, L. (2008). Drivers’ beliefs about exceeding the speed limits, Transportation Research, Part F, 376–389.
  • Yang, C. ve Yang, C. (2007). Seperating latent classes by information criteria. Journal of Classification, 24, 183–203.