Doug Fuchs, Vanderbilt University
Mark Berends, Vanderbilt University
Loulee Yen, Vanderbilt University
Kristen McMaster, University of Minnesota
Laura Saenz, University of Texas-Pan American
Abstract: Across 5 years, we are exploring necessary levels of on-site teacher assistance to scale-up a research-validated reading intervention-Peer Assisted Learning Strategies (PALS). We target 2 grade levels at which challenges to scaling-up may differ substantially: kindergarten, where an important reading objective is word-level skills; and 4th grade, where a major challenge is to transition from learning to read to reading to learn.
In Year 1 (2004-2005), we conducted a 17-week, randomized control trial in 6 of the poorest counties in south Texas; in Minneapolis, St. Paul, and Bloomington, Minnesota; and in Nashville, TN. At each site, about 50 kindergarten teachers (with 1000 students) were randomly assigned within their schools to 4 conditions: no-PALS (level 1); a 1-day PALS workshop (level 2); a 1-day PALS workshop followed by several "booster" sessions (level 3); and a 1-day PALS workshop, followed by several booster sessions, and once-per-week mentoring (level 4). On all PALS teachers (levels 2, 3, 4) and students, we conducted systematic observations of their PALS implementation twice during the year. We also tested sub-groups of high-achieving, average-achieving, and low-achieving (including special needs) kindergartners in each class prior to and immediately following the intervention on a battery of phonological and reading tasks.
We applied 3-level HLM (student nested within class; classes nested within schools) to posttreatment phonological awareness and reading scores, corrected for pretreatment scores, to explore the importance of the respective levels of teacher assistance (1 = controls; 2, 3, 4 = increasingly intensive support) and implementation site (Nashville, Minnesota, Texas). Results indicated scaling-up effects (2, 3, 4> 1 across the three sites) for phonological awareness skills. For word reading measures, there were teacher assistance level x site interactions, whereby 2, 3, 4> 1 for Nashville. This pattern was less consistent in Minnesota and Texas. For reading in connected text: 3, 4> 1, but 2 = 1 in Nashville. Thus, in Nashville, more intensive levels of teacher support promoted reading in connected text.