Setting
The original study took place in 11 U. S urban public elementary schools in the Pacific Northwest that were known to have large numbers of students who preformed below proficiency on the state reading tests.
Study sample
For the randomized sample, intervention characteristics are: 53% male, 0% English Language Learners, 67% free and reduced price lunch, 4% special education, 18% Asian, 59% black, 15% Hispanic, and 15% mixed/other race. For the comparison group in the randomized sample, the characteristics are: 55% male, 0% English Language Learners, 64% free and reduced price lunch, 5% special education, 22% Asian, 35% black, 8% Hispanic, and 4% mixed/other race.
Intervention Group
The intervention included a set of 108 scripted lessons involving letter-sound correspondences, phoneme decoding, irregular words, spelling, and oral reading practice. Each one-on-one tutoring session included four to eight short components along with a decodable storybook for oral reading practice. Sessions were 30 minutes long, four days a week, for every week between the fall pretest and spring post-test of the study. Intervention students received an average of 66.3 tutoring sessions and completed an average of 60.03 lessons. Implementation fidelity was reported as a 4.49 average on a five-point scale where five equals always implements correctly and one equals never implements correctly.
Comparison Group
Comparison group students did not receive any supplemental tutoring. They received regular classroom instruction. The authors conducted three formal observations of classroom literacy instruction blocks to document typical classroom instruction.
Support for implementation
Lessons were scripted and assigned to children based on performance on a master test. Paraeducators had an initial 2 hour training with followup as needed throughout the intervention. Less experienced tutors had an extra 0.5 to 3 hours (M=1 hour) of coaching during the intervention. Researchers conducted 240 fidelity observations with an average of 9.6 per tutor. On a 5 point scale with 5 =always implements correctly, the mean for the 25 tutors was 4.49.