Questions and Conclusions from a Discussion of Reading Recovery

by Dr. Patrick Groff

NRRF Board Member & Senior Advisor
Dr. Patrick Groff, Professor of Education Emeritus San Diego State University, has published over 325 books, monographs, and journal articles and is a nationally known expert in the field of reading instruction.


The effectiveness and cost of Reading Recoveryİ have become a subject for serious examination. School district response to the discussion contained in these excerpts will have a direct affect on California school district Language Arts curriculum expenditures and on the reading ability of California school children.

Are Gains From RRİ Enduring?

It is vitally important, of course, that for an experimental remedial reading program, such as RRİ, to be adopted and funded for regular, long-term use in schools, it first must demonstrate experimentally that it generates relatively large gains for students involved in it. Such gains are of little if any consequence, however, if they soon fade away, leaving the students who achieved them no better able to read than are students who had no RRİ tutoring. It therefore is essential for RRİ to prove that the initial gains in reading that it produces are lasting in nature.

It is clear that RRİ has failed to meet this test of its effectiveness. That is to say, several disinterested, independent critics of RRİ (Center, et. al, 1995; Glynn, et al., 1989; Groff, 1994; Ohio Department of Education, 1995; Shanahan & Barr, 1995; Wasik & Slavin, 1993) have pointed out that most of the reading improvement gains brought on by RR are temporary; they “wash-out” over time. This finding is underscored by the fact that students released from RRİ, as remediated, often read so poorly that they qualify for inclusion in other remedial reading programs. (Groff, 1994).

The most impressive of the studies so far of whether reading gains in RRİ endure or evaporate is the one recently commissioned by the Ohio Department of Education, and conducted by the consulting firm, Battelle, of Columbus, Ohio. The exceptionally well-designed Battelle study (Ohio, Department of Education, 1995) surveyed the permanency of RRİ reading scores in many Ohio schools over a four-year school period, 1990-1994.

The Battelle Study concluded that there were initial reading gains from RRİ, greater even than those made by students in other remedial reading programs in use in Ohio schools at the time. “The differences in achievement level [favoring RRİ], however, were not evident in the three subsequent years” of the RR students schooling (Ohio Department of Education, 1995, p. 73.). “The average score advantage of Reading Recovery students was not maintained at the end of the second grade,” nor on “tests for the third and fourth grades” (p.1).

Is Reading Recoveryİ Cost-Effective?

As Slavin, et al. (1993) convincingly demonstrate, one teacher-one student tutoring, per se, has been proved experimentally to be the most effective instructional arrangement known. This one-to-one tutoring also is the most expensive kind of teaching, of course. School districts thus almost always must sacrifice some other educational services to students in general, omit purchases of educational materials, equipment, supplies, and housing, and/or increase regular teachers’ workloads or delay their pay raises, in order to find the money necessary to provide tutoring for selected students.

Any school district or board of trustees contemplating the adoption of RRİ as a tutoring vehicle therefore must look beyond the question whether RR actually is the most effective program of its kind. In addition, they must consider carefully whether the financial costs of adopting RRİ outweigh its actual contributions toward the remediation of students’ reading handicaps.

In this regard, the present analysis of RRİ so far has suggested that for several reasons this tutoring program is not the most pedagogically-effective remedial reading tutoring program available. If this negative judgment of RRİ is accurate and convincing, it is double important that school officials inspect carefully the cost-effectiveness of RRİ.

Those who control whether purchase of RRİ will be made for use in schools should realize, first, that the promoters of RRİ typically downgrade its cost, depicting them as very reasonable, and therefore as apt payment for RRİ’s supposed great successes in overcoming students’ reading handicaps. For example, Dyer (1992) sets the teacher salary costs per student of RRİ at $2063. By comparison, he maintains, the cost of the federally-funded program, Title 1, is $4715. Here Dyer wrongly assumes that all Title 1 students need 5 years of remedial reading tutoring. He also conveniently ignores other costs of RRİ. This relative low cost of RRİ is claimed by other of its advocates. For example, the cost per student in McAllen, Texas is reported as $2538 (Salinas, et al., 1993).

In contrast to these figures, are ones more recently gathered in schools in Ohio (Ohio Department of Education, 1995). These schools estimated that the costs of RRİ are 50 per cent higher than other (unnamed) remedial reading programs that they used. Earlier on, a study of RRİ in the Canton, Ohio schools found, however, that “Reading Recovery (sic) is approximately four times as expensive as Chapter 1” (now called Title 1) over the same period of time–but is less effective (Fincher, 1988, p. 20). Fincher noted that the low estimates of the cost of RRİ by its advocates fail to take into account costs of fringe benefits to RRİ teachers, materials and supplies used in RRİ, teacher training, salaries and travel expenses of RRİ program officials, and other miscellaneous financial outlays.

Hiebert (1994, p. 22) agrees that estimates of the cost of RRİ by its proponents “represent a deflated figure per student because teacher benefits have been excluded,” along with start-up costs of training, and costs of training rooms. These underestimated costs of RRİ also are based on the dubious assumptions that RRİ is successful with all students, that students never require any remedial reading instruction after they exit RRİ, that each RRİ tutor serves sixteen students, and that none of the reading handicapped students not given RRİ ever will attain proficiency in reading, Hiebert (1994) adds. Taking all these ordinarily unreported costs and lack of careful oversight of RRİ into account, Hiebert (1994, p. 22) places the “cost per successful student [in RRİ] at $8333,” or $278 per hour of tutoring.

Shanahan and Barr’s (1995) estimate of the costs of RRİ are significantly lower than that of Hiebert (1994), but higher than those offered by the proponents of RRİ. Taking into consideration fewer of RRİ’s normally ignored, but necessary, financial outlays than did Heibert (1994), Shanahan and Barr put the cost of RRİ at $4625 per student. The addition of RRİ thus doubles the average cost of educating a student, or triples it, if one accepts Hiebert’s estimate in this regard.

Another way of deciding the economic practicality of RRİ is offered by Rasinski (1995). In his view, when investigating whether RRİ is fiscally feasible, it is necessary, first, to determine how many times larger were the reading gains generated by RRİ than were the average reading gains made by non-RRİ students. The reading gains from RRİ must double or triple (Shanahan/Barr) or Hiebert) those of regular classroom instruction if the extra cost that is incurred by RRİ is to be justified.

Using Rasinski’s formula, the reading gains made by students in the Pinnell, et al. (1994) investigation of RRİ thus appear to be too small to warrant the extra costs of RRİ. For example, on the two standardized tests given there, the RRİ reading scores surpassed the “direct instruction skills plan” (DRA) scores by only 9 percent and 9 percent, respectively. As noted, the DRA is designed for group teaching. We therefore need to know if RRİ reading gains would double or triple gains made with DRA group teaching.

Public Reaction to RRİ

Furthermore, it is likely that the high cost of RRİ creates a public relations problem for the schools. In this sense, it is predictable that people outside the educational establishment who learn of the high price of RRİ, and the severe contraction, over time, of reading gains initially obtained with it, will protest that expenditures for RRİ are not a wise use of the limited school funds that are now available.

This potential for public remonstration against the adoption of high-cost RRİ is exemplified in a 1995 letter from Ohio state senator Cooper Snyder (chair of his senate’s education committee) to the Ohio superintendent of education regarding the Battelle study of RRİ. As noted above, this study found that significant extra money spent on RRİ did not result in enduring reading gains for RRİ students. “To put it mildly, I am chagrined with the findings reported” by the Battelle study, Snyder wrote. To Snyder, “RRİ is nothing more nor less than a band-aid for the first grade.” I thus “am further dismayed to learn that the [Ohio] Department [of Education] apparently concludes Reading Recoveryİ is okay,” of the general public as to the need for RRİ. “Why aren’t we doing the [reading instruction] job right to begin with?” he asks, suspecting that “something has to be fundamentally and very basically wrong” in the way students ordinarily are taught to read.

The “something” that is “fundamentally and basically wrong” about reading instruction, to which Snyder refers, is the “Whole Language” approach to reading development that has been adopted in his state, and even more so in California. The introductory remarks of the present analysis of RRİ explain why California students now are the least capable readers in the nation, and thus are prime candidates for RRİ. Here it is noted that more schools in California have made the unfortunately wrong decision to adopt WL than have schools in other states. As a consequence, California students are now the least capable readers in the nation.

It is important, as well, that future reports from local school districts that proclaim the purported successes of RRİ (e.g., Holmes, 1994) carefully consider beforehand the critiques of RRİ as made in the present analysis of it, especially those regarding: (a) the evidence that tests used to decide students’ entry and exit from RRİ are not valid nor reliable for that purpose; (b) the empirical invalidity of certain RRİ practices, ones that are based on WL; (c) the lack of longevity of reading gains generated by RRİ; and (d) the need for a precise and comprehensive formula for deciding if RRİ is cost- effective, as compared with other remedial reading tutoring plans, with small group teaching in the regular classroom, or with remedial reading programs that use paraprofessional or volunteer tutors. In short, no longer acceptable at face value are statements from RRİ promoters that RRİ “remains cost- effective because of its short-term nature” (Swartz & Klein, 1994, p. 6). This is a far too simplistic view of the cost issue of RRİ, and therefore no longer can be tolerated.


The conclusions that may be drawn from the present analysis of the empirical validity of Reading Recoveryİ (RRİ) can be expressed in a series of Questions and Answers about this remedial reading tutoring program:

Q: Does RRİ Produce gains in reading for reading handicapped first grade students?
A: Yes, but only initially. The preliminary advances in reading ability generated by RRİ for these students soon disappear.
Q: Is RRİ based solidly on the relevant experimental research findings on reading development?
A: No. To the contrary, RRİ is a projection of the empirically unverified “Whole Language” approach to literacy development. The principles and practices of RRİ are very similar to those of Whole Language. Whole Language has been shown clearly to be a failed instructional innovation.
Q: Is RRİ the best remedial reading tutoring program now available?
A: Probably not. Theoretically, RRİ is inferior to competing tutoring programs since they conform more to the experimental research findings than does RRİ. Empirical evidence in this regard, gained from disinterested, independent studies of RRİ versus competing tutoring programs, is badly needed if this question is to be answered satisfactorily.
Q: What are the major reasons why studies of RRİ by its advocates always find that RRİ is highly effective?
A: These investigations typically do not compare RRİ with competing tutoring programs. In action, these studies often have been designed and implemented, either expressly or incidentally, so that they result in favorable outcomes for RRİ. Of note here: when disinterested researchers study RRİ, it is not found to be exceptionally effective.
Q: How closely do RRİ tutors follow the prescriptions for its conduct laid down by Clay (1993b)?
A: This issue has not been investigated so far. Thus, it may be that tutors in RRİ programs that produce exceptional reading gains do not carefully follow Clay’s directions in many respects. For example, these tutors may spend considerably more time on explicit instruction of word recognition skills than Clay recommends should be done. There thus may be some useful informal corrections made of the official RRİ procedures by its practitioners.
Q: Is RRİ the most cost-effective of all remedial reading tutoring programs?
A: Clearly not, since some competing tutoring programs use paraprofessionals or volunteers as tutors, and have a greater chance for success because they align their practices more closely to what the experimental research reveals about reading development than does RRİ. The fact that initial reading gains generated by RRİ soon fade away also negatively reflects on its costs-to- results ratio.
Q: Has RRİ become a commercial product?
A: Some RRİ proponents claim it is not (Swartz & Klein, 1994). However, the fact that the name, Reading Recoveryİ, now is trademarked signifies that it is considered a marketable item. Clay has profited from the increasingly large sales of her books on RRİ. Centers that charge fees for training RRİ tutors have grown increasingly numerous as more and more school districts have been sold on the idea they need to adopt RRİ. By 1992, there were such commercial enterprises in 38 states (Shanahan & Barr, 1995). As well, RRİ is advertised in much the same way as common consumer products are. That is, its advertisements stress its supposed advantages, while conveniently leaving undisclosed its shortcomings.
Q: Is RRİ the best way to reverse the current decline in students’ reading ability?
A: No remedial educational program is preferable to initial and regular teaching of reading that is based on pertinent empirical evidence, i.e., instruction that has the best chance to be effective. Thus, if school reading programs conform to the findings of experimental research on reading development, there would be much less need for RRİ, or for that matter, any other special education services.
Q: Is RRİ in its present form, as prescribed by Clay (1993b), the most desirable one?
A: No. It is clear that RRİ needs to modify the practices that Clay invented for it so that RRİ is in greater conformity with what the experimental research says about beginning reading development. However, many RRİ advocates likely will strongly resist this modification of RRİ. In this regard, they declare it their duty “to protect the integrity of the [RRİ] program” (Swartz & Klein, 1994, p. 6) from such reform.
Q: Should school boards vote to purchase RRİ purely on the basis of recommendations of school officials to do so?
A: At this point in time, only by putting blind trust in these recommendations, and by ignoring the evidence of RRİ’s pedagogical weaknesses, and its relative high costs, can school boards justify making this decision to adopt RRİ. School boards would act more wisely if they made sure, first, that the initial, regular teaching of reading in their schools closely conformed to the relevant empirical findings. They then should search out less expensive, more empirically relevant tutoring programs, than RRİ is, for students who make less than normal progress in reading.

Center, Y., et al. (1995) An experimental evaluation of Reading Recoveryİ. Reading Research Quarterly, 30, 240-263

Dyer, P. C. (1992) Reading Recoveryİ: A cost-effectiveness and educational outcomes analysis. ERS Spectrum, 10 (1), 10-19

Fincher, G. E. (1988). Reading Recoveryİ and Chapter 1: A three-year comparative study. Canton, OH: Canton, Ohio, City Schools.

Glynn, T., et al. (1989). Reading Recoveryİ in context. Wellington, New Zealand: New Zealand Department of Education.

Groff, P. (1994). Reading Recoveryİ: Educationally sound and cost-effective? Effective School Practices, 13 (1), 65-69

Hiebert, E. H. (1994). Reading Recoveryİ in the United States: What difference does it make to an age cohort? Educational Researcher, 23 (9), 15-25.

Holmes, J. (1994). Reading Recoveryİ site report. San Diego, CA: San Diego County Consortium, San Diego County Office of Education. Ohio Department of Education (1995). Longitudinal study of Reading Recoveryİ, 1990-91 through 1993-94.
Columbus, OH: Ohio Department of Education.

Pinnell, G. S., et al. (1994). Comparing instructional models for the literacy education of high-risk first graders. Reading Research Quarterly, 29, 8-39.

Rasinski, T. (1995). On the effects of Reading Recoveryİ. Reading Research Quarterly, 30, 264-270.

Salinas, A., et al. (1993). Reading Recoveryİ 1992-93 evaluation report. McAllen, TX: McAllen Independent School District.

Shanahan, T. & Barr, R. (1995). Reading Recoveryİ: An independent evaluation of the effects of an early instructional intervention for at-risk learners. Reading Research Quarterly, 30 (4) 958-996.

Slavin, R. E., et al. (1993). Preventing early school failure: Research, policy and practice. Boston, MA: Allyn and Bacon.

Snyder, C. (September 22, 1995). Letter to Dr. John Goff, Superintendent, Ohio Department of Education, Columbus, Ohio.

Swartz, S. L. & Klein, A. F. (1994). Reading Recovery: An overview. Literacy, Teaching and Learning, 1 (1), 3-7.

Wasik, B. S. & Slavin, R. E. (1993). Preventing early reading failure with one-to-one tutoring: A review of five programs.
Reading Research Quarterly, 28, 179-200.