It has been a rocky year for Reading First, the No Child Left Behind Act initiative intended to improve reading instruction in the primary grades. The same could also be said for other closely allied federal ventures related to reading education. Formal complaints that the government is not treating some educational programs fairly in the Reading First marketplace have led to an investigation by the U.S. Government Accountability Office. A cynic might say that the complaining education vendors probably aren’t getting as much federally sponsored business as they would like. But I think we should view these most recent complaints in context, as an extension of the list of concerns over federal reading efforts that have accumulated over the past six years.
Begin with the 2000 National Reading Panel report, which identified the five instructional factors at the heart of Reading First: phonemic awareness, phonics, fluency, vocabulary, and comprehension. As soon as the report appeared there were outcries that the panel had been prematurely selective in its coverage of topics (the decision to look at research on these five factors was made very early in the process, perhaps before the panel’s first meeting) and in the methodologies it was willing to accept (experimental and quasi-experimental evaluations were examined, with virtually all other data ignored).
Too often, advocates for Reading First have defended the scientific buttressing of the program by at least implying that these five factors are the only aspects of reading instruction that enjoy scientific support. They fail to acknowledge that these were the only elements that could have been deemed as evidence-based, given the NRP’s process of deciding what factors to examine in advance of any look at the evidence.
One defense of the NRP focus was that future panel reports would get around to other components of reading instruction. But frustration with this defense also mounted this year, when the U.S. Department of Education decided against releasing a long-promised report on reading interventions for English-as-a-second-language students, admitting it was of poor quality. The country continues to wait for another long-promised report from the department on early-childhood reading education. And adding to the frustration, the department’s What Works Clearinghouse has been promising a report on beginning reading that, as of this writing, is nowhere in sight.
Equally concerning is the possibility that a critical analysis in the original National Reading Panel report, the one about phonics instruction, may have been biased high, given the program’s emphasis. The particular statistical procedure used to analyze all the phonics-instruction studies the NRP considered, meta-analysis, requires the analyst to make a number of decisions. The reading panel made some decisions and concluded that phonics instruction has a moderate impact on grade 1 reading. In another analysis of the same studies, however, Rutgers University professor Gregory Camilli and his colleagues concluded that phonics instruction has a much smaller impact on beginning-reading achievement. In addition, there is now evidence that grade 1 phonics instruction may be bad for some children, relative to the alternatives.
Fred Morrison of the University of Michigan and Connie Juel of Stanford University both demonstrated that the phonologically oriented, word-recognition emphases are likely to maximize the reading achievement of weak entering grade 1 readers, but that for better-reading entering grade 1 students, achievement is likely to be maximized by participation in more holistic programs, such as ones emphasizing more reading of literature and writing. Real progress has been made in understanding how to create balanced beginning-literacy programs that can provide differentiated instruction depending on learners’ needs (programs providing different amounts of sound- and word-level skills instruction, literature, and other literacy experiences). Sadly, such balanced instruction seems an anathema to many of those who identify with the strong phonological, sound- and word-level skills models of Reading First.
Reading First’s narrow focus on basic skills is increasingly difficult to accept because many states, responding to pressure from the No Child Left Behind law, have put in place standards for primary-grades reading instruction (and literacy instruction more generally) that go well beyond students’ acquisition of the five Reading First factors. States want students to be familiar with a wide variety of genres and authors, and to be reading in the full range of content covered in the elementary curriculum (math, science, social studies). They want even primary-grades students to be critical but enthusiastic readers of text, and to be making progress in composition and oral-discussion skills. Yet variety in literature, critical understanding, motivation, writing, and discussion skills is nowhere to be found in the Reading First mix. This forces recipient schools, in some cases, to decide between the skills focus of Reading First and the broader state standards. And they risk adequate-yearly-progress sanctions if their students do poorly on state tests.
Then, of course, there is nothing at all in Reading First past the grade 3 level, despite great needs for better literacy education in the upper-elementary, middle school, and high school grades. Since the original NRP report, there have been some very telling analyses showing that not all children at risk for reading failure first manifest their problems in the primary grades, a key assumption of Reading First. There is just so much missing from the current federal support for reading education.
One of the complaints that Congress has received about Reading First is from the Reading Recovery program, which has a good beef. Reading First favors some instructional programs known to produce only small positive effects on beginning-reading achievement, such as stand-alone phonics programs. Reading Recovery also produces small positive effects on beginning-reading achievement, and yet receives very little Reading First funding and strong public criticism from Reading First advocates. In short, Reading First seems to magnify the small effects of some programs consistent with its preferred phonological model of reading, while failing to acknowledge same-size effects in other programs.
Skepticism about the effectiveness of the Reading First program as a whole has been growing since the summer. Despite the Bush administration’s strong endorsement of the practice of evaluating programs by true experimentation (that is, randomized field trials), no interpretable randomized field trials that I am aware of have been provided for the Reading First program. A federally funded evaluation is to be reported in 2006, although it is one using a design that falls well short of a randomized, true experiment (an approach known as regression discontinuity analysis). The best data currently available on the program—found on state Web sites and presented at conferences, rather than in the peer-reviewed journals demanded by the government for other reading research—include small gains after a year or two of Reading First, relative to performance before the program was implemented. Given the fact of increasing state standards in general, perhaps the gains seen in Reading First schools would have been obtained anyway. There is no way to know, based on the designs and data I have seen.
Variety in literature, critical understanding, motivation, writing, and discussion skills is nowhere to be found in the Reading First mix.
More positively, there are ongoing experimental evaluations of specific full-program interventions used in some Reading First sites. One problem with these, however, is that programs consistent with the phonological perspective seem to be favored in government-funded evaluations. The country needs to know about the full range of available programs. Those who are attempting to develop literacy interventions should have the opportunity to evaluate them. In other words, all of the competing perspectives in the current literacy-instruction mix should have access to federal resources to test their positions, since the very existence of a test of a program will, most likely, play a role in determining its eligibility for future Reading First-type funding efforts.
A broader problem with evaluations of the specific interventions—and with the federally funded evaluation of the overall Reading First program—is that the focus is too much on whether a program “works,” rather than how or why it works. Science probes deeper than the “what works” question, for only by understanding how and why an intervention works can we adapt it to new populations and settings effectively.
A constructive response by the government to these growing concerns about Reading First and related federal programs might be to view them as critical information about the first-generation effort at putting federally supported, evidence-based reading instruction in place. Then a better second-generation effort could be crafted, one that benefited from the reasonable objections about what has occurred during this first-generation effort. This could be accomplished by inviting the most astute critics of the current effort to be part of reforming and rethinking the federal programs in reading education and reading education research.