Skip Navigation
Reading First Impact Study Final Report

NCEE 2009-4034
October 2008

Exhibit ES.4.  Estimated Impacts on Key Components of Scientifically Based Reading Instruction (SBRI): Spring 2007
Domain Actual Mean With Reading First Estimated Mean Without Reading First Impact Effect Size of Impact Statistical Significance of Impact (p-value)
Professional Development (PD) in SBRI     
Amount of PD in reading received by teachers (hours)a25.84 13.71 12.13* 0.51* (<0.001)
Teacher receipt of PD in the five essential components of reading instruction (0-5)a 4.30 3.75 0.55* 0.31* (0.010)
Teacher receipt of coaching (proportion)a 0.83 0.63 0.20* 0.41* (<0.001)
Amount of time dedicated to serving as K-3 reading coach (percent)b,c 91.06 57.57 33.49* 1.03* (<0.001)
Amount of Reading Instruction     
Minutes of reading instruction per daya 105.71 87.24 18.47* 0.63* (<0.001)
Supports for Struggling Readers      
Availability of differentiated instructional materials for struggling readers (proportion)b 0.98 0.97 0.01 0.15 (0.661)
Provision of extra classroom practice for struggling readers (0-4)a 3.79 3.59 0.19* 0.20* (0.018)
Use of Assessments     
Use of assessments to inform classroom practice (0-3)a 2.63 2.45 0.18 0.19 (0.090)
a Classroom level outcome
b School level outcome
c The response rates for RF and nonRF reading coach surveys were statistically significantly different (p=0.037). Reading first schools were more likely to have had reading coaches and to have returned reading coach surveys.
d Missing data rates ranged from 0.1 to 3.3 percent for teacher survey outcomes (RF: 0.1 to 1.0 percent; non-RF: 0 to 4.9 percent) and 1.3 to 2.8 percent for reading coach and/or principal survey outcomes (RF: 0 to 1.6 percent; non-RF: 2.7 to 4.1 percent). Survey constructs (i.e., those outcomes comprised of more than one survey item) were computed only for observations with complete data, with one qualification: for the construct "minutes spent on reading instruction per day," the mean was calculated as the total number of minutes reported for last week (over a maximum of 5 days) divided by the number of days with non-missing values. Only those teacher surveys with missing data for all 5 days were missing 0.9 percent).
The complete Reading First Impact Study sample includes 248 schools from 18 sites (17 districts and 1 state) located in 13 states. 125 schools are Reading First schools and 123 are non-Reading First schools.
The effect size of the impact is the impact divided by the actual standard deviation of the outcome for the non-Reading First Schools.
Values in the "Actual Mean with Reading First" column are actual, unadjusted values for Reading First schools; values in the "Estimated Mean without Reading First" column represent the best estimates of what would have happened in RF schools absent RF funding and are calculated by subtracting the impact estimates from the RF schools' actual mean values.
A two-tailed test of significance was used; statistically significant findings at the p≤.05 level are indicated by *.
EXHIBIT READS: The observed mean amount of professional development in reading received by teachers with Reading First was 25.84 hours. The estimated mean amount of professional development in reading received by teachers without Reading First was 13.71 hours. This impact of 12.13 hours was statistically significantly (p<.001).
SOURCES: RFIS, Teacher, Reading Coach, and Principal Surveys, spring 2007