Features that make Direct Instruction effective: • Only 10% of the material is new, while the remaining 90% is a review of previously taught content. • Students are grouped based on their skill levels, which are determined by assessments administered before commencing the Direct Instruction program. • The emphasis is on the student's pace by either slowing down, reteaching, or accelerating through easily understood material. • The external validity of Direct Instruction has been tested, and the program is research-based. Debates about the efficacy of DI have raged before the final results of
Project Follow Through were published; however, there is substantial empirical research supporting its effectiveness. A meta-analysis published by Adams and Engelmann (1996), a chief architect of the DI program, finds a "mean effect size average per study...(as) more than .75, which confirms that the overall effect is substantial." A 2018 meta-analysis by Stockard et al. found an average effect on test scores of approximately 0.6 standard deviations. In some special education programs, it is used in a
resource room with small groups of students. Some research has shown there is benefit with this model. Direct Instruction is used with students from every population segment (with regard to poverty, culture, and race). In
Project Follow Through, the DI model was ranked first in achievement for poor students, students who were not poor, urban students, rural students,
African American students,
Hispanic students, and
Native American students. Today, many of the Bureau of Indian Affair's highest-performing schools use Direct Instruction materials. The Baltimore Curriculum Project has many schools with Free and Reduced Lunch Rates above 75% serving student populations that are more than 90% African American. These schools have shown strong achievement gains using Direct Instruction. Meta-analysis of 85
single-subject design studies comparing direct instruction to other teaching strategies found the effects to be substantial for students with learning disabilities; however, when qualified by IQ and reading levels,
Strategy Instruction (SI) had better effects for the high IQ group. For the low-IQ discrepancy groups, higher effect sizes were yielded for a Combined DI and SI Model when compared to all competing models. With the exception of handwriting, DI's effects were all above 0.8 (i.e., reading and mathematics). John Hattie's Visible Learning: A Synthesis of Over 800 Meta-Analyses Relating to Achievement (2009) summarizes the results of four meta-analyses that examined Direct Instruction. These analyses incorporated 304 studies of over 42,000 students. Across all of these students, the average effect size was 0.59 and was significantly larger than those of any other curriculum Hattie studied. Direct Instruction is recognized as one of two effective models of comprehensive school reform and, in many cases, can be integrated into a tiered model system to address students with developing problems. The findings from
Project Follow Through, conducted in a variety of selected communities throughout the United States, suggest that Direct Instruction is the most effective model for teaching academic skills and for affective outcomes (e.g.
self-esteem) of children. Recent large-scale studies (1997–2003), such as the Baltimore Curriculum Project, show that it is possible to help schools that are in the lowest twenty percentile with respect to academic achievement steadily improve until they are performing well above average. In some cases, school achievement improved from the 16th percentile to above the 90th percentile. ==Criticism==