Skip Navigation
Reading First Impact Study Final Report

NCEE 2009-4038
November 2008

Exploratory Analyses of Variations in Impacts and Relationships among Outcomes

This report also presents results from exploratory analyses that examine some hypotheses about factors that might account for the pattern of observed impacts presented above. These exploratory analyses are based on analyses of subgroups of students, schools, grade levels, and/or years of data collection. The information is provided as possible avenues for further exploration or for improving Reading First or programs like Reading First. However, the study was not designed to provide a rigorous test of these hypotheses, and therefore the results are only suggestive. Findings from these exploratory analyses include the following:

  • Data collected during three school years (2004-05, 2005-06 and 2006-07) were used to examine variation over time in program impacts. No consistent pattern of differential impacts over time was established.
  • No relationship was found between the number of years a student was exposed to RF and student reading achievement.
  • There was no statistically significant variation in impacts across sites in the study, either by grade or overall, for reading instruction or for reading comprehension.
  • Correlational analyses, which are outside the causal framework of the main impact analyses presented in the report, indicate a positive and statistically significant association between time spent on the five essential components of reading instruction promoted by the program and students' reading comprehension. A one-minute increase in time devoted to instruction in the five dimensions per daily reading block was associated with a 0.07 point increase in scaled score points in first grade, and a 0.06 point increase in second grade. This relationship does not hold for models that include other potential mediators of student achievement. However, due to data limitations, these latter models could only be run on a subset of the data; thus, we do not know whether the differences in the findings across models are due to changes in the sample or changes in the model specification itself.