|
|
|
|
|
|
|
|
New Statistical Methods for the Study of Change Over Time: Implications for Research in Child Language Disorders |
|
|
|
|
|
|
|
|
|
|
|
|
|
David Kaplan - University of Wisconsin-Madison
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
SRCLD Year: |
2009 |
Presentation Type: |
Invited Speaker |
Presentation Time: |
(na) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This tutorial lecture presents an overview of quantitative methodologies for the study of stage-sequential development based on extensions of Markov chain modeling. Four methods are presented that exemplify the flexibility of this approach. The first method is the manifest Markov model which provides estimates of transitions over time in categorical responses that are assumed to be measured with perfect reliability. The second method is the latent Markov model which directly accounts for measurement error and provides transition probabilities at the latent level. The third method is latent transition analysis, which addresses the use of multiple indicators of a latent categorical variable and is arguably better suited for the study of stage-sequential developmental processes than the manifest or latent Markov model. The final method is the mixed Markov model. This latter model addresses unobserved heterogeneity in the Markov chains. A special case of the mixture latent Markov model, the so-called “mover-stayer" model, is presented in this lecture. Issues of model specification, estimation, and testing are briefly discussed. These four methods are applied to an example of stage sequential development in reading competency in the early school years utilizing data from the Early Childhood Longitudinal Study – Kindergarten Cohort. The tutorial closes with a brief discussion of Bayesian approaches when sample sizes are small.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|