The Corporation for National and Community Service | 2014
D. Random Assignment of Students Within Schools
All students at the 23 sampled schools identified by Fall benchmark scores as eligible for MRC services (i.e., Tier 2)
were randomly assigned to either the MRC program (i.e., treatment) or control group at the beginning of the first
semester prior to the start of tutoring.
13F
14
During the first few weeks of the first semester, AmeriCorps members
conducted the Fall CBM assessments on students who the school’s staff identified as potentially eligible for MRC
services.
14F
15
Scores on these assessments determined students’ eligibility for MRC services. The total evaluation
sample size was determined by the number of students that AmeriCorps members at a school were capable of
assessing on a weekly basis.
15F
16
Students were added to the sample by grade in pairs, whereby each student in each
grade within a school was matched with another student based upon their Fall benchmark score. Students within
pairs were then randomly assigned to either the program or control condition. This matched pair design ensured that
students in the program and control groups had similar Fall benchmark scores at the start of the school year. A figure
illustrating the randomization process is provided in Appendix A.2.
When a school had more eligible students than the AmeriCorps members’ had capacity to assess, the evaluation
team recommended that the school follow their usual procedures for selecting students for the program. Most
schools rank order eligible students within grade by benchmark score and provide tutoring to students closest to the
benchmark first. Therefore, in instances where the number of students eligible for MRC services was greater than the
number of students the AmeriCorps member(s) could assess on a weekly basis, a subset of students closest to
benchmark was selected, and the students beyond the members’ capacity to assess were excluded from the
evaluation sample. Furthermore, in an effort to obtain as equal numbers of students in each grade as possible, the
evaluation team paired students by similar benchmark score and added the two students closest to benchmark to the
study sample in each grade (e.g., 2 kindergartners, 2 first grade students, 2 second grade students, and 2 third grade
students for a total N of 8), repeating the process iteratively until the maximum number of students that could be
assessed at the school was reached.
16F
17
In some schools, the number of eligible students differed by grade; therefore,
the total number of students selected to participate in the evaluation also differed by grade.
17F
18
After students were
14
Students who participated in the MRC program in previous years were eligible for participation in the evaluation. The important eligibility
criteria for students was not whether they had received MRC services in the past, but instead whether they were eligible to receive services at
the beginning of the 2012-2013 school year. Furthermore, since students were randomly assigned to condition, it was equally likely that a
student who previously received services would be assigned to the program and control groups. As such, each group should have a roughly
equal number of students who had and had not participated previously in the MRC program.
15
Processes for identifying students can somewhat vary across schools. Some schools assess all students in grades K-3, while others may
use previous years’ test scores or other more subjective means for identifying students to be assessed. Schools were asked to use the same
procedures they typically employ to identify students for Fall benchmark testing.
16
On average, one full-time member is able to assess 30 students each week.
17
The matched-pair random assignment design is described in section III.D.
18
Note that as reported in Table III.2, the number of second grade students eligible for MRC services was lower than any other grade.
IMPACT EVALUATION OF THE MINNESOTA READING CORPS K-3 PROGRAM Page 18