The first week after becoming president, George W. Bush sent Congress an educational reform “blueprint” that included a reading initiative mirroring policies that supposedly produced a Texas educational “miracle” when Bush was governor.
Bush promised to eliminate the nation’s “reading deficit” by “ensuring that every child can read by the third grade.” To do so, he proposed applying “the findings of years of scientific research on reading” to “all schools in America.” Bush stressed that these research findings were “now available,” especially in the recently published report of the National Reading Panel (NRP), which reviewed “100,000 studies on how students learn to read” and provided a guide for “scientifically-based reading instruction.” Bush said his legislation would provide funds for reading instruction — but only if the instruction were “scientifically based.”
Never has this nation had a more scientifically-minded president.
Shortly after his reading proposal, Bush rejected a policy for reducing carbon dioxide emissions in power plants because, he said, the science is “still incomplete.” He also terminated plans to reduce the amount of arsenic in drinking water because, he said, he wanted to determine if the proposed reductions were “supported by the best available science.” He went on to withdraw the United States from the international agreement to reduce global warming because, he said, the “state of scientific knowledge of the causes of, and solutions to global climate change” was “incomplete.”
Although many have questioned Bush’s intellectual ability to do the job, these decisions demonstrate that we have a president who evidently has studied an array of scientific literature and has been able to formulate various policy judgments based on the empirical evidence. One might conclude that the reading research must indeed be sound for Bush to have given it his one empirical imprimatur.
BUSH AND ‘THE BASICS’
The reading education Bush has in mind can be gleaned from remarks made in 1996 when he advised Texas teachers to get “back to the basics” — that is, return to traditional education that focuses on basic skills, basic facts, and a traditional curriculum, and reject all classroom reforms that include children’s perspectives and thereby reduce the authority of the teacher.
“The building blocks of knowledge,” Bush explained, “were the same yesterday and will be the same tomorrow. We do not need trendy new theories or fancy experiments or feel-good curriculums. The basics work. If drill gets the job done, then rote is right.”
Although advocates of what is now being called “scientifically-based” reading instruction might take issue with the “rote” comment, Bush accurately described the essentials of this teaching. “Trendy” is code for whole language, cooperative learning, meaning-emphasis learning (reading instruction that stresses comprehension and teaches skills as part of students’ reading for meaning), and critical literacy (reading instruction that examines and questions values, assumptions, and ideologies in written material) — that is, anything that is an alternative to the “basics.”
“Basics,” meanwhile, is code for beginning reading instruction that emphasizes exclusively the explicit, direct, and systematic instruction of skills and minimizes the need for meaning and comprehension until the skills are learned. Under such an approach, teachers follow preestablished reading programs that move children through a step-wise process from small parts of language to larger ones. While advocates of this instruction insist it is “balanced,” that is, instruction balances reading for meaning with learning skills for reading, a close look reveals that the comprehension end of the seesaw remains close to the ground for a long time.
THE SCIENTIFIC ARISTOCRACY
Demands for “scientifically based” reading education precede the current legislation. Since the mid-1990s, reading research funded by the National Institute of Child Health and Human Development (NICHD), a branch of the National Institutes of Health, has been identified as the scientific gold standard justifying scripted skills-emphasis instruction (that is, instruction using a reading program that has prescribed, sequential lessons that teachers and students must follow). The “Chief” (actual title) of the reading research division, Reid Lyon, has been a persistent spokesman for scripted instruction in policy hearings across the country. Lyon and NICHD-supported researchers have usually had an easy time doing so because alternative views at these hearings have been few, if any. Lyon has been a chief adviser to Bush and Secretary of Education Rod Paige, and could become, as the Baltimore Sun reported, “the administration’s reading chief.” In the Bush reading legislation, the reading division of NICHD is included in the processes for disseminating information about the reading legislation, for setting standards for “scientific” research, and for reviewing and judging applications.
Last year, NICHD arranged for Congress to “request” that it form a panel to report on the best scientific information on reading. Given NICHD’s dominant role in selecting the panel and administrating its work, its conclusions were predictable. Representative of the media reports announcing the panel’s findings was Education Week, whose headline read, “Reading Panel Urges Phonics For All in K-6.” Although the report urged more than phonics, the headline was generally correct because the panel stressed the need for an early mastery of sound-symbol connections and similar skills through explicit, systematic, direct instruction. Implicit in the panel’s report and explicit in most media reports was the rejection of a whole language approach to literacy (www.nationalreadingpanel.org).
In Misreading Reading: The Bad Science That Hurts Children (Heinemann, 2000), I reviewed the NICHD reading research and related studies and documented the shoddy research in this work and its failure to support skills-emphasis instruction. With the publication of the National Reading Panel report, nothing has been added that refutes my critique. Before providing a few illustrations of the report’s deficiencies, I want to return to Bush’s description of the report.
THE ‘100,000’ STUDIES
Bush’s legislative proposal states that the panel reviewed 100,000 studies on reading, a number that has been repeated in the media. Unmentioned is the fact that the number does not remotely describe the actual number of studies used in the panel’s analysis.
The panel did begin by looking at the research studies on reading and found that approximately 100,000 had been published since 1966. However, the panel used several criteria for extensive pruning of this number. For example, the studies had to deal with reading development from preschool through high school, and therefore excluded studies on many reading topics such as literacy skills in occupations or brain functioning and literacy activity. The studies also had to meet the Panel’s definition of “scientific research,” which was limited almost entirely — and narrowly — to quantitative, experimental studies.
Perhaps most important, the panel used only studies relating to the instructional topics which the panel majority had decided were key areas of good reading instruction. As Joanne Yatvin, an Oregon principal, wrote in her “Minority View” dissenting from the report’s conclusions, “From the beginning, the Panel chose to conceptualize and review the field narrowly, in accordance with the philosophical orientation and research interests of the majority of its members.”
This questionable procedure eliminated a variety of important instructional issues contained in the research literature, such as the relationship of writing and reading, meaning-emphasis instruction, the interconnection of emotions and literacy learning, or approaches for responding to children’s individual literacy needs. This panel of experts had no qualms about exercising a priori its judgment on what is central to successful reading instruction, and ignoring contrary views of other leading reading researchers on what comprises the best way to learn to read. The panel knew best.
After pruning the “100,000” studies, the numbers that remained were: 52 studies on phonemic awareness; 38 studies for phonics; 14 for fluency; and 203 for 16 categories of comprehension instruction, that is about 12 or 13 studies on average in each category. Certainly these numbers are sufficient for drawing some reasonable conclusions. But they are far from the much-heralded “100,000 Studies” and, most importantly, not sufficient numbers for establishing restrictive national policy. There were members of Bush’s education advisory committee who were familiar with the NRP report and knew the 100,000 figure was a misrepresentation. But given their interest in foisting their brand of teaching on the nation, it is not surprising that no one told the President that the figure would mislead the public and Congress.
APPRAISING THE RESEARCH
The primary method of evaluating the research was a meta-analysis, a statistical method that pools a group of studies and estimates the average effect something has on something else, in this case, the effect of aspects of instruction on achievement. A meta-analysis can provide useful information, but it also can have substantial deficiencies, such as not uncovering the quality of and reasoning in the studies pooled. Although the panel’s conclusions were derived from its meta-analysis, its appraisal of the quality of the research and its reasoning on behalf of its a priori decisions about instruction can be gleaned from the report’s detailed descriptions and interpretations of a number of studies. Because of space limitations, I can only discuss two that appear at the beginning of the report’s first section. I believe they are representative of the panel’s entire appraisal of its studies.
CORRELATION AND CAUSE
The panel concluded that “results of the experimental studies allow the Panel to infer that [phonological awareness] training was the cause of improvement in students'” reading. (Phonological awareness refers to the ability to separate and manipulate speech sounds mentally and orally, as when blending or separating phonemes in order to identify words.) The report states, for example, that one study1 showed that [phonological awareness] was the top predictor along with letter knowledge” (knowing the names of letters) of later reading achievement. The panel’s description suggests that these two predictors are “causes.”
However, the accuracy of this conclusion becomes questionable when one includes the third strongest predictor of reading achievement. The panel did not mention this predictor, possibly because it has no apparent relationship to written language. That predictor is the degree of success on a “finger localization” test, in which a child whose vision is blocked identifies which of her or his fingers an adult has touched. Despite its predictive correlation with future reading achievement, finger localization skill in itself could not be considered “causal” to learning to read, and no educator would suggest finger location training as a beginning reading method.
Zip codes are also good “predictors” of academic achievement. A student’s zip code (an indication of family income and education, quality of schools in the area, a child’s access to educational experiences, etc.) is strongly correlated with future school success. However, this correlation does not make zip codes a cause of academic achievement. Unfortunately, given the failure of some reading researchers to understand the difference between correlation and causation, it is possible that one of them might concoct an experiment to see if the reading achievement of children in a poor, urban area with terrible schools improves if they are given the zip code of children in an affluent, suburban area with excellent schools.
The NRP report also fails to mention that the researchers who did the study offered an explicit caveat about confusing correlation with causation. Yes, they did find that letter knowledge was a strong predictor of future reading, but they emphasized its predictive strength did not mean that beginning readers needed to know letter names in order to get off to a fast, secure start in reading.
Although “knowledge of letter names has been traditionally considered the single best predictor of reading achievement, there appears to be no evidence that letter-name knowledge facilitates reading acquisition,” the researchers said in their report.
Letter-name knowledge is likely to be part of and represent experience with early written language that contributes to literacy attainment. That knowledge — like the knowledge of various skills, such as phonological awareness — can be considered to be a “marker” of these experiences and accomplishments. None of this complexity and these distinctions are captured in the report’s simplistic summary that the study “showed that phonological awareness was the top predictor along with letter knowledge.”
NO NEED FOR SKILLS TRAINING
A striking example of congruence between the panel’s interpretation of the research and its a priori conclusions about the best reading instruction is its description of a study supposedly showing that a phonological awareness “treatment group” was “compared to one no-treatment group” and the “effect size was impressive.”2 This provides evidence, wrote the panel, that the training program being studied could be “used effectively in American classrooms.”
This group comparison was correct as far as it went, but the report did not tell readers that there were two control groups, not one. Besides the one control the panel described, another control group learned phonological awareness in an informal, “as needed” way similar to the way skills are taught in a whole language approach. In other words, we have in this study an opportunity to compare children learning these phonological awareness skills either through an explicit skills training program or an implicit, as-needed, approach. Although this comparison would have allowed the panel to delve into the question of how these skills should be taught and learned, it did not do so. Therefore, we will.
At the end of the school year, the skills training group did significantly better on phonological tests, but the researchers found there were “no significant differences between the [skills training and implicit teaching] groups on the tests of word reading and spelling.” (Interestingly, the NRP report neglected to mention this finding.) The researchers who did the study observed, moreover, that “the significantly superior scores achieved by the training group in this study on tasks of phonemic awareness suggest that this group should also achieve higher scores on the reading tasks, but this was not in fact the case.” Again, the conclusions were not quoted in the NRP report.
The researchers went on to propose that the “writing experiences” of the informal learning group might have accounted for their reading success. On average, they “wrote longer stories than either” the training group or the “normal” kindergarten group. These conclusions too were omitted from the report.
In other words, the study showed that children can learn rhyme, syllable synthesis, word reading, and spelling without needing an explicit skills training program. Furthermore, extensive writing activities are likely to be effective for attaining the literacy knowledge for which the researchers tested. This study also lends some support to a holistic written language approach insofar as it indicates that phonological skills can be readily learned within a rich array of reading and writing activities. And it further demonstrates that there not only is no need for a step-wise approach to literacy learning, but that such an approach can reduce time spent on essential and productive literacy activities, such as “writing experiences.”
THE ‘SCIENTIFIC METHOD’
Despite the lack of scientific evidence for a heavy emphasis on direct instruction of skills in beginning reading instruction, “scientifically-based” appears 31 times in the House version of Bush’s reading bill and almost as many in the Senate version. To be sure that no one reading the bills is confused about the meaning of “reading,” its “components” are enumerated in a list that starts with phonemic awareness and is followed by phonics, vocabulary, reading fluency, and, at the end of the list, reading comprehension.
For skills-first proponents, “comprehension” is always part of their definition of learning to read, but like this definition, the “components,” are always designed to be learned in a sequential, stepwise, systematic, script-ed, managerial instructional program. The list never runs in the opposite direction, and the components never are described as interactive from the start of learning to read.
The judicial coup d’Ztat that brought us the president has many permutations, one of which is seizing the power to design and enforce the standards of literacy education that will coerce schools into using “scientifically-based” reading programs, especially if they hope to receive federal money. In pursuing and exercising this power, talk about “scientifically-based” instruction has been no more than a string of infomercials justifying the legitimacy of that power. In legislative hearings, critical voices are locked out. But such is the way of the “scientific method” in the back rooms of Washington these days.
1. Share et al., 1984. “Sources of individual differences in reading acquisition,” Journal of Educational Psychology, 1309-1324.
2. Brennan & Ireson, 1997. “Training phonological awareness,” Reading and Writing, 241-263.