Setting the Record Straight
Federal officials are holding schools to impossible standards based on misinterpretations of the research.
Richard L. Allington
Proponents of the federal Reading First mandates of the No Child Left Behind Act have routinely misrepresented and exaggerated what the research shows about effective classroom reading instruction and early reading interventions. The legislative impetus for the new laws and regulations resulted from a recent policy consensus that "90–95 percent of poor readers can be reading on grade level if provided with appropriate instruction." Yet there is no evidence that classroom instruction of any sort can come close to meeting the needs of the readers who struggle most. Research does show that expert, individual tutoring produces on-level reading achievement with many struggling readers. But is that how we should define "appropriate instruction"? If so, it will come with a hefty price tag.
But questions persist. What evidence supports the assertion that 90–95 percent of poor readers could be reading on grade level if only educators would offer appropriate instruction? And what exactly would it cost to provide such instruction to every student who needs it?
Interpreting the Data
Locating the source of the 90–95 percent figure is easy. G. Reid Lyon of the National Institute of Child Health and Human Development (NICHD) provided that figure in his testimony before the U.S. Senate Committee on Labor and Human Resources in April 1998. That figure has stuck like glue. Quoting Lyon (1998),
We have learned that for 90 to 95 percent of poor readers, prevention and early intervention programs that combine instruction in phoneme awareness, phonics, fluency development, and reading comprehension strategies, provided by well-trained teachers, can increase reading skills to average reading levels. (p. 7)
What research did Lyon use to establish this success rate? He provided no citations in his testimony, but in that same year he coauthored a paper (Fletcher & Lyon, 1998) citing two studies (Torgesen, Wagner, & Rashotte, 1997; Vellutino et al., 1996) that supported his assertion.
Vellutino and colleagues studied approximately 200 at-risk primary grade students from six suburban school districts. They divided the students into two groups: treatment and control. The treatment students received individual, expert tutoring for one semester, typically for 30 minutes per day, for an average total of 70–80 sessions. Some students continued to receive tutoring for a longer period. Almost half (44.7 percent) of the tutored students achieved average reading levels (the 45th percentile) after tutoring.
Torgesen and colleagues provided 20-minute tutoring sessions four days a week for two and one-half years to approximately 100 primary grade students. Teachers and aides provided daily tutoring on an alternating basis. Tutored students received one of three possible interventions, each of which produced similar outcomes on the word reading tests.
Although the Vellutino and Torgesen interventions did demonstrate the power of intensive, expert tutoring, the outcomes indicated that these interventions raised roughly 50 percent of the poor readers—not 90–95 percent—to the 45th percentile or above, or to average reading levels.
Did Lyon misinterpret the data in the two studies? I suggest this because the intensive, expert tutoring of struggling readers (in addition to other efforts the schools initiated) reduced the proportion of poor readers in the school populations to approximately 10 percent of all students. In the Torgesen study, the very poor readers after intervention accounted for approximately 2.5 percent of the total population, and the students reading below the 45th percentile accounted for approximately 8 percent. In the Vellutino study, 3 percent of the original suburban students in the larger sampling population had very poor reading skills after tutoring, with 9 percent of the total population still falling below the 45th percentile. Using the 45th percentile criterion, 91 percent of all readers would be at average achievement levels.
But 90 percent of all students reading at or above average achievement levels is quite a different matter from 90 percent of poor readers achieving average levels of proficiency. To clarify, imagine Normal Elementary School. It has 1,000 students whose reading achievement just happens to be perfectly distributed along the traditional bell curve. Thus, 450 students fall below the 45th percentile, and 200 fall below the 20th percentile (a typical cutoff point for identification as a struggling reader). Getting 90 percent of the struggling reader population above the 45th percentile would require a program that solved the problems of 180 of the 200 struggling readers. But getting 90 percent of the total population (1,000) above the 45th percentile would require a program that solved the reading problems of only 100 struggling readers. In the first scenario, after the tutoring intervention, the school has 20 students (200 - 180) struggling to learn to read; in the second, the school has 100 struggling readers (1,000 - 900). Simply put, it is harder and more expensive to design interventions that reduce the pool of poor readers to 2 percent of the population (20 students) than to design one that leaves 10 percent of the population (100 students) still struggling.
The two studies indicate that when expert, intensive remedial tutoring supplements good classroom reading instruction, half of the poorest readers in any given suburban elementary school will still lag behind their peers, unable to read grade-level texts independently. Moreover, Vellutino and colleagues removed from the pool of poor readers all students with measured IQs below 90. Approximately 30 percent of all students have measured IQs below 90, according to the bell curve distribution. In other words, this study demonstrated that intensive, expert tutoring could accelerate literacy development in half of the struggling readers with IQs of 90 or higher, resulting in average levels of reading proficiency in this group by the end of 2nd grade.
Both studies demonstrate that many struggling readers can achieve average levels of reading proficiency; that these students are not learning-disabled in the traditional sense but are simply instructionally needy; and that expert, intensive tutoring is a powerful intervention. But neither study shows that 90–95 percent of poor readers can achieve average reading levels with appropriate instruction, even when that appropriate instruction is expert, intensive tutoring.
An Expensive Solution
The Vellutino and Torgesen studies demonstrate that expert tutoring can be highly effective as an early intervention tool (see also Shanahan, 1998, and U.S. Department of Education, 2003). But schools don't normally provide such tutoring to struggling readers. Why not? Let's look at the costs involved.
In both studies, the students received 30–75 hours of expert tutoring; students in the Torgesen study received an additional 30 hours of tutoring from a paraprofessional. The expert tutors were typically certified teachers, including several with M.S. degrees and advanced certification in reading remediation. Such a teacher would cost a school approximately $50,000 annually (National Center for Education Statistics, 2003).
Tutors following the Vellutino model would work with 10 students each day. In our Normal Elementary School example, the program would require approximately 20 full-time tutors for a single semester or 10 full-time tutors for an entire year to teach the 200 students who qualified for tutoring. Tutors following the Torgesen model would work with four more students each day for an additional one and one-half years, which would mean hiring additional paraprofessionals. Either model would cost the school at least $500,000 annually (10 full-time teachers with a salary of $50,000 each).
In the NICHD studies, poor readers who did not qualify or were not selected for tutoring still received small-group remedial or special education instruction, as did some of the tutored students who completed the program. We will assume that existing revenues at Normal Elementary will continue to fund those services. Even so, developing the necessary tutoring expertise, adding a supervisory component, and creating tutoring spaces would most likely add to the cost, which would be substantially higher than $500,000 per year.
But many schools do not have such a normally distributed student population. Consider a recent study of students enrolled in urban, high-poverty schools (McGill-Franzen, Allington, Yokoi, & Brooks, 1999). The average pre-test performance of this group of randomly selected urban kindergarten students was the 13th percentile, well below the cutoff that both Vellutino and Torgesen used for tutoring eligibility.
Consider the tutoring costs for a school such as this, which we will call Abnormal Elementary. Let's say the school serves 1,000 students in a high-poverty neighborhood and two-thirds of the students—667— qualify for tutoring. That would require 34 additional full-time expert tutors, at a minimum cost to the school of $1.7 million annually (34 × $50,000). And even with this added tutoring, we could still expect approximately half of the tutored students' measured reading achievement to fall below the 45th percentile.
One final consideration. Although Vellutino and colleagues and Torgesen and colleagues provide powerful testimony about the potential of early, intensive, expert instructional intervention, we have no data that suggest how permanent the tutoring effects will be. If we generalize the data from Reading Recovery (Hiebert, 1994), we can assume that many students will remain on this normal achievement track. But early intervention is not a vaccine that protects students from further difficulties: Without continued intensive instructional support, many students will gradually fall behind. Few schools seem to have the resources to provide intensive tutoring for students who get off track in the later grades. Ensuring that all 6th graders read on grade level would cost even more.
It's time to face the fact that some students will need expert, intensive intervention for sustained periods of time—possibly throughout their entire school careers—if they are to attain and maintain on-level reading proficiencies. But we haven't yet developed interventions that ensure that all students will be reading on grade level.
What the Evidence Really Says
The implied premise of recent federal legislation emphasizing evidence-based instruction—that 90–95 percent of poor readers, or 98 percent of all students, would be reading on grade level if only teachers would follow the research—exaggerates what the evidence actually reports. No intervention has raised the achievement of 90 percent of poor readers to the 50th percentile. Moreover, no research suggests that classroom teachers can help 90–95 percent of students acquire grade-level reading proficiencies by learning more about phonology, using a scripted curriculum, teaching systematic phonics, or following some "proven" program. Programs that most reliably accelerated the literacy development of struggling readers relied on costly, expert, intensive tutoring, which raised the achievement of only about half of the poor readers to average (45th percentile) levels.
Given the potential costs of initiating a tutorial intervention for all struggling readers and maintaining student on-level performance, no one seems to be advocating—much less funding—research-based tutorial interventions. The current emphasis is on buying new, "scientific" reading materials (Kame'enui & Simmons, 2002), which is surely a less expensive option.
Struggling readers are instructionally needy. Classroom teachers will never have the time to provide the one-to-one support that so many of these students require. Research has shown that tutoring is an effective intervention that can provide this one-to-one support and raise student achievement. It is, however, costly. If legislators and other policymakers are going to mandate adequate yearly progress on the basis of research that measured the effects of individual tutoring, then they should fully fund that research-based tutoring for all struggling readers.
Either that or admit to the public that we plan on leaving many children behind.
Fletcher, J., & Lyon, G. R. (1998). Reading: A research-based approach. In W. Evers (Ed.), What's gone wrong in America's classrooms. Stanford, CA: Hoover Institute Press.
Hiebert, E. H. (1994). Reading Recovery in the United States: What difference does it make to an age cohort? Educational Researcher, 23(9), 15–25.
Kame'enui, E. J., & Simmons, D. (2002). Consumer's guide to evaluating a core reading program. Institute for Development of Educational Achievement, University of Oregon. Available: http://reading.uoregon.edu/appendices/con_guide_3.1.03.pdf
Lyon, G. R. (1998, April 28). Overview of reading and literacy initiatives: Statement of G. Reid Lyon. Testimony before the Senate Committee on Labor and Human Resources. Washington, DC. Available: http://184.108.40.206/publications/pubs/jeffords.htm
McGill-Franzen, A., Allington, R. L., Yokoi, L., & Brooks, G. (1999). Putting books in the room seems necessary but not sufficient. Journal of Educational Research, 93(2), 67–74.
National Center for Education Statistics. (2003). Digest of education statistics tables and figures 2001 [Online]. Available: http://nces.ed.gov/programs/digest/d01/dt079.asp
Shanahan, T. (1998). On the effectiveness and limitations of tutoring. In P. D. Pearson & A. Iran-Nejad (Eds.), Review of research in education, 23 (pp. 217–234). Washington, DC: American Educational Research Association.
Torgesen, J. K., Wagner, R. K., & Rashotte, C. A. (1997). Prevention and remediation of severe reading disabilities: Keeping the end in mind. Scientific Studies of Reading, 1(3), 217–234.
U.S. Department of Education. (2003). Identifying and implementing educational practices supported by rigorous evidence: A user-friendly guide [Online]. Available: www.ed.gov/rschstat/research/pubs/rigorousevid/index.html
Vellutino, F. R., Sipay, E. R., Small, S. G., Pratt, A., Chen, R., & Denckla, M. B. (1996). Cognitive profiles of difficult-to-remediate and readily remediated poor readers. Journal of Educational Psychology, 88(4), 601–638.
Richard L. Allington is the Irving and Rose Fien Professor of Education at the University of Florida, Gainesville, Florida, and Vice President of the International Reading Association; firstname.lastname@example.org.
INDEX OF NCLB OUTRAGES