Proceedings from a Working Meeting on School Readiness Research: Guiding the Synthesis of Early Childhood Research. Endnotes

12/15/2009

1. The traditional method used to link interventions to outcomes, even in the context of randomized designs, is based on OLS regression, in which variation in implementation is correlated with variation in outcomes. Another experimental approach uses instrumental variable analysis. This approach has been applied in many contexts by economists and is becoming increasingly popular for use with randomized experiments (for example, Angrist, Imbens and Rubin, 1996). It does not compare existing levels of student achievement and instructional practice. Instead it leverages the fact that a high-quality randomized experiment (or a well-executed regression discontinuity analysis) can produce unbiased estimates of program impacts on classroom instructional practices and on student test scores. The approach thus examines the association between outcomes that is implied by a pattern of program impacts. For example, in a randomized study, the analysis compares program-induced changes in student outcomes with program-induced changes in classroom instruction, where both changes are estimated using the randomized design. Under certain conditions, this analysis can provide internally valid (statistically consistent) estimates of the causal effect of classroom instruction on student performance. This methodology overcomes some of the problems in relational analysis (omitted variables and attenuation bias), although instrumental variable analysis depends on being able to show that there are no mediators additional to the classroom instruction that could account for the relationships with child outcomes.

2. The fact that parallel effects were not demonstrated on teacher ratings of children's attention and impulsivity may be related to power rather than to inconsistency in outcomes. Teacher ratings have been found to have higher correlations among children in the same classroom and center (ICCs) than do cognitive measures such as the PPVT. This means that only relatively large impacts can be detected for the teacher-reported outcomes,

View full report

Preview
Download

"index.pdf" (pdf, 307.32Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"apa.pdf" (pdf, 78.45Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"apb.pdf" (pdf, 401.85Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"apc.pdf" (pdf, 288.78Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"apd.pdf" (pdf, 90.04Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"ape.pdf" (pdf, 190.7Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"apf.pdf" (pdf, 68.04Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®