Teachers use a variety of strategies to teach children to read. Reading instruction, for example, may focus on decoding skills or making meaning (i.e., understanding what they read). Activities may be teacher-managed, such as when a teacher introduces a list of vocabulary words to a class or child managed, such as when children engage in independent reading. We have known for a while that some teachers are better able to individualize instruction and produce larger gains in reading achievement for their students relative to other teachers. We have not, however, been able to capitalize on this knowledge because we did not know (a) the critical differences in instruction that produced the effect and (b) how to help less effective teachers do what the more effective teachers were doing.

In an earlier study of first grade students, researchers from the University of Michigan observed that students who made the most progress over the course of the year had teachers who individualized instruction in very specific ways. For example, children who began first grade with weaker letter-word recognition skills and received more teacher-managed instruction that focused on decoding skills made greater progress. Children with stronger vocabulary skills at the beginning of the year made more progress if they spent more time engaged in child-managed meaning-focused activities (e.g., independent reading). The researchers carefully developed an algorithm that accurately predicted how much students learned in one year based on students' fall reading scores and the proportion of time spent in different types of instructional activities. Armed with this insight and a grant from NCER, this team, now from Florida State University and University of Michigan, developed a software program that utilizes students' beginning of the year reading scores to develop an instructional profile of the type, duration, and timing of instruction that is recommended for each child and assigns children with similar profiles to reading groups for classroom instruction. The instructional profile was designed so that teachers can apply the recommended instructional profile to the reading curriculum provided by their school in order to plan instructional activities for each reading group.

In a randomized controlled evaluation, teachers were randomly assigned to use the algorithm-guided individualized reading instruction program or to plan reading instruction in their normal fashion. The evaluation included an ethnically and economically diverse sample of first grade students in 10 schools. Relative to children in the control classrooms, first grade students in the treatment group made greater gains in reading comprehension - in fact, there was about a two-month difference in grade equivalents between the groups. In classrooms in which teachers made greater use of the software program, almost every student, even those who started with weaker vocabulary scores, were reading at grade level by the end of the year (grade equivalent of 2.0).

Connor, C. M., Morrison, F. J., Fishman, B. J., Schatschneider, C., & Underwood, P. (2007). Algorithm-guided individualized reading instruction. Science, 315, 1-3.