Project: LDbase Integrated Dataset


This project page describes the creation of a latent reading score created by combining data across 12 datasets available on LDBase. This variable is the product of an Integrated Data Analysis (IDA) which was modeled using a multiple-group measurement model described below. This analysis leveraged assessments across multiple projects to create scores that are placed on the same scale across all of the projects, which allows researchers to make direct comparisons across these projects on reading skill.
This score was estimated using multiple-group measurement invariance modeling that employed a missing variable strategy described by Widaman (2013) when manifest variables are completely missing. This involves creating random normal variables for the variables that are completely missing from the dataset. The second feature that differentiates this strategy from a standard multiple-group measurement invariance model is the use of the good-enough principle (Serlin & Lapsey (1985). The good-enough principle is designed to combat the problem of very large samples indicating statistical significance in the presence of very small parameter differences. Using this strategy, the researcher defines the range of differences between two models that they consider good-enough. Functionally this means that the obtained chi-squared difference between the models is based upon a prespecified noncentral chi-squared distribution. This noncentral distribution is characterized by the difference in degrees of freedom of the two models and a noncentrality paramenter. The noncentrality parameter is the product of the sample size prespecified by the researcher. MacCallum et al (2006) proposed to base this difference on values of the RMSEA index of fit. To date there is no agreed upon recommendation for this parameter, however, MacCallum et al (2006) recommended the use of .5 to .6 to represent small differences between models. This entire procedure is laid out in van Dijk, Schatschneider, Al Otaiba, and Hart (2022).

The following projects contributed data to the construction of a latent reading score.
Project Kids. This project contains data from 8 studies of investigating interventions upon early reading development

Florida Longitudinal Study. This was a longitudinal individual differences study of reading and reading related skills of students from first grade to fourth grade

Promoting Adolescents’ Comprehension of Text (PACT). This was a study of students from seventh through twelfth grade who had been identified as having reading difficulties

Colorado Twin Project. This was a longitudinal individual differences study of twins from 8 to 18 years of age. Reading and reading-related skills were assessed yearly.

Western Reserve Reading and Math Project (WRRMP). This was a longitudinal individual differences study of twins from 8 to 18 years of age. Reading and reading-related skills were assessed yearly.

MacCallum, R. C., Browne, M. W., & Cai, L. (2006). Testing differences between nested covariance structure models: Power analysis and null hypotheses. Psychological methods, 11(1), 19.
Serlin, R. C., Lapsley, D. K., Keren, G., & Lewis, C. (1993). Rational appraisal of psychological research and the good-enough principle. A handbook for data analysis in the behavioral sciences: Methodological issues, 199-228.
van Dijk, W., Schatschneider, C., Al Otaiba, S., & Hart, S. A. (2022). Assessing measurement invariance across multiple groups: When is fit good enough?. Educational and Psychological Measurement, 82(3), 482-505.
Widaman, K. F., Grimm, K. J., Early, D. R., Robins, R. W., & Conger, R. D. (2013). Investigating factorial invariance of latent variables across populations when manifest variables are missing completely. Structural equation modeling: a multidisciplinary journal, 20(3), 384-408.

Project Active From
2022 to present
Project Descriptor(s)

Most Recent Datasets in Project

There are no datasets for this project.

Most Recent Documents in Project

There are no documents for this project.

Most Recent Code in Project

There are no code for this project.