A Bayesian Probabilistic Framework for Identification of Individuals with Dyslexia.

Purpose: Bayesian-based models for diagnosis are common in medicine but have not been incorporated into identification models for dyslexia. The purpose of the present study was to evaluate Bayesian identification models that included a broader set of predictors and that capitalized on recent developments in modeling the prevalence of dyslexia. Method: Model-based meta-analysis was used to create a composite correlation matrix that included common predictors of dyslexia such as decoding, phonological awareness, oral language, but also included response to intervention (RTI) and family risk for dyslexia. Bayesian logistic regression models were used to predict poor reading comprehension, unexpectedly poor reading comprehension, poor decoding, and unexpectedly poor decoding, all at two levels of severity. Results: Most predictors made independent and substantial contributions to prediction, supporting models of dyslexia that rely on multiple rather than single indicators. RTI was the strongest predictor of poor reading comprehension and unexpectedly poor reading comprehension. Phonological awareness was the strongest predictor of poor decoding and unexpectedly poor decoding, followed closely by family risk. Conclusion: Bayesian-based models are a promising tool for implementing multiple-indicator models of identification. Ideas for improving prediction and implications for theory and practice are discussed.

Combining Old and New for Better Understanding and Predicting Dyslexia.

Despite decades of research, it has been difficult to achieve consensus on a definition of common learning disabilities such as dyslexia. This lack of consensus represents a fundamental problem for the field. Our approach to addressing this issue is to use model-based meta-analyses and Bayesian models with informative priors to combine the results of a large number of studies for the purpose of yielding a more stable and well-supported conceptualization of reading disability. A prerequisite to implementing these models is establishing informative priors for dyslexia. We illustrate a new approach for doing so based on the known distribution of the difference between correlated variables, and use this distribution to determine the proportion of poor readers whose poor reading is unexpected (i.e., likely to be due to dyslexia) as opposed to expected.

A model-based meta-analytic examination of specific reading comprehension deficit: how prevalent is it and does the simple view of reading account for it?

Many individuals with poor reading comprehension have levels of reading comprehension that are consistent with deficits in their ability to decode the words on the page. However, there are individuals who are poor at reading comprehension despite being adequate at decoding. This phenomenon is referred to as specific reading comprehension deficit (SRCD). The two purposes of this study were to use a new approach to estimate the prevalence of SRCD and to examine the extent to which SRCD can be explained by the simple view of reading. We used model-based meta-analysis of correlation matrices from standardized tests to create composite correlation matrices for the constructs of reading comprehension, decoding, and listening comprehension. Using simulated datasets generated from the composite correlation matrices, we used residuals from regressing reading comprehension on decoding to create a continuous index of SRCD. The prevalence of SRCD is best represented not as a single number but as a continuous distribution in which prevalence varies as a function of the magnitude of the severity of the deficit in reading comprehension relative to the level of decoding. Examining the joint distribution of the residuals with reading comprehension makes clear that the phenomenon of reading comprehension that is poor relative to decoding occurs throughout the distribution of reading comprehension skill. Although the simple view of reading predictors of listening comprehension and decoding makes significant contributions to predicting reading comprehension, nearly half of the variance is unaccounted for.

The Prevalence of Dyslexia: A New Approach to Its Estimation.

How prevalent is dyslexia? A definitive answer to this question has been elusive because of the continuous distribution of reading performance and predictors of dyslexia and because of the heterogeneous nature of samples of poor readers. Samples of poor readers are a mixture of individuals whose reading is consistent with or expected based on their performance in other academic areas and in language, and individuals with dyslexia whose reading is not consistent with or expected based on their other performances. In the present article, we replicate and extend a new approach for determining the prevalence of dyslexia. Using model-based meta-analysis and simulation, three main results were found. First, the prevalence of dyslexia is better represented as a distribution that varies as a function of severity as opposed to any single-point estimate. Second, samples of poor readers will contain more expected poor readers than unexpected or dyslexic readers. Third, individuals with dyslexia can be found across the reading spectrum as opposed to only at the lower tail of reading performance. These results have implications for screening and identification, and for recruiting participants for scientific studies of dyslexia.