Characterizing optimal intervention intensity: the relationship between dosage and effect size in interventions for...

Although Warren, Fey and Yoder (2007) have described the key components of "dosage", one needs to go beyond description if one is to understand "optimal" dosage, specifically one needs to relate the characteristics of the intervention to the size of the intervention effect. This study examines the association between dose, intensity, and effect size in 20 randomized controlled studies taken from a few systematic reviews focusing on interventions aiming to ameliorate vocabulary, phonology, and syntax. Reporting of dosage characteristics is an important issue. Our analysis shows that "teaching episodes" and "dose form" are rarely reported in the included studies. The other dosage characteristics are present but not always reported in a transparent fashion. Session length and cumulative intervention intensity is lower for phonology interventions than it is for vocabulary intervention. Dosage, however defined, is not directly associated with outcome, although the level of association varies across the three interventions, for example appearing stronger for vocabulary and phonology than syntax. Taking the three interventions together the dosage components are related to the intervention effects size, but the sample is small and the association is not statistically significant. This study concludes that, while the framework suggested by Baker (2012) and adapted from Warren et al. (2007) is useful but without reference to the effect size of a study, it can only ever tell half the story. One needs to be able to relate dosage to outcome, asking questions about the relationship between the different dosage characteristics and the intervention effect size. Given the available data, it is not, at this stage, possible to make recommendations about optimal dosage.