The study took place in 71 schools in three sites over two years: 14 schools in Nashville,
Tennessee; 36 schools in Minnesota; and 21 schools in south Texas. The final analytic sample
included 67 schools.
Project staff first recruited schools to obtain balanced samples on site-specific factors: Title I
status in Nashville, Title I status and whether the school offered half-day or full-day kindergarten
in Minnesota, and the proportion of limited English proficiency students in the schools in south
Texas. Teachers were recruited within the selected schools, and 224 teachers participated
over the two study years (55 teachers participated in both years, for a total of 279 teacheryears).
Within each participating school, teachers were randomly assigned to one of the four
conditions: control, workshop, workshop plus booster, or workshop plus booster and helper.
The study does not report the number of teachers in each condition.Researchers obtained parental consent for more than 90% of the students in the classrooms of study teachers. These students were pretested, and 12 students were selected from each class: four children with the lowest reading scores, four children with the highest scores, and four children with scores in the middle of the score distribution. The consented study sample included 3,171 kindergarten students, with 668 in the control condition, 968 in the workshop condition, 931 in the booster condition, and 604 in the helper condition. The final hierarchical linear modeling (HLM) analysis sample included only 2,959 students and 259 teachers. The WWC could not calculate attrition by condition based on the information provided in the study. However, based on reasonable assumptions about how to attribute overall attrition to groups, the study is assumed to have low differential attrition.
Twenty-four percent of the students in the study were English language learners, 62% were
eligible for free or reduced-price lunch, 50% were female, and 5% had Individualized Education
Plans. Of the students in the sample, 40% were Hispanic, 26% were non-Hispanic White,
25% were African American, 5% were Asian, and 3% were of other ethnicities.
The study included three treatment conditions: (1) a day-long training workshop (K-PALS), (2) the
workshop plus two follow-up booster sessions (K-PALS + Booster), and (3) the workshop and
booster sessions plus weekly technical assistance provided by a graduate student (K-PALS +
Booster + Helper). Although the treatment conditions vary by the amount of training and support
received by teachers, the K-PALS intervention was the same in all three treatment conditions.
Students were paired by their teachers and then worked through structured lessons during
35-minute sessions implemented four times per week in this study. Stronger readers were paired
with weaker readers, and pairings were maintained for four to six weeks before being reorganized.
Within each pair, students took turns acting as the reader and the coach. The classroom teacher monitored the pairs and provided feedback as necessary. Program materials, including a teacher manual and all student worksheets, were provided by K-PALS.
The comparison was a business-as-usual counterfactual. Comparison teachers did not implement the intervention and did not receive any additional training.
The primary outcome is Rapid Letter Sounds, an alphabetics measure developed by Levy
and Lysunchuk (1997). All study students were tested approximately three weeks before the
intervention began and again 20 weeks later. For a more detailed description of the outcome
measure, see Appendix B.
Support for implementation
All teachers in the three treatment groups attended a day-long training workshop before the
intervention began. For the K-PALS + Booster treatment group, two follow-up booster sessions
were also provided to allow teachers to review program procedures and to identify and solve
implementation issues. Teachers in the K-PALS + Booster + Helper treatment group attended
the training workshop and booster sessions and also had weekly technical assistance provided
by a trained graduate assistant. Average implementation fidelity, measured at two points during
implementation by the project coordinator, was 86%.