Promoting Reading Achievement and Countering the “Fourth-Grade Slump”: The Impact of Direct Instruction on Reading Achievement in Fifth Grade

Abstract

Previous research has documented a substantial decline of standardized test scores of children from low-income backgrounds, relative to more advantaged peers, in later elementary grades, the so-called “fourth-grade slump.” This article examines changes in reading achievement from first to fifth grade for students in a large urban school system with a high proportion of students from economically deprived backgrounds. Students received first-grade reading instruction from Direct Instruction (DI), Open Court, or a mixture of reading curricula. Results indicate that students in schools using DI had significantly greater gains in both reading vocabulary and comprehension than students in the two other settings and that their average levels of achievement in fifth grade were above the national norms, thus countering the fourth-grade slump.

ACKNOWLEDGMENTS

I thank Cristy Coughlin, Kurt Engelmann, Zigfried Engelmann, Daniel Johnston, Jerry Silbert, and Gary Davis for very helpful comments on earlier drafts of this article. Any errors remaining are my sole responsibility.

Notes

¹We were also unable to find such longitudinal studies of other reading curricula.

²Even though we examined data from the same school system as Mac Iver and Kemper (2002), and Addison and Yakimowski (2003), our work differs from theirs by using multivariate statistics to examine changes from first to fifth grade for all of the students exposed to DI in the system, rather than a subsample, and by comparing the impact of exposure to three different curricular programs in first grade.

³Five of the 16 schools that introduced DI ceased their association with NIFDI soon after implementation and began to work with another provider of implementation support. To maintain an adequate sample size, the results for all 16 schools are combined for this analysis. Separate results for these two subsets of the DI schools are available upon request from the author and do not alter the substantive findings reported here.

⁴To preserve degrees of freedom and keep the sample size as large as possible, the two cohorts of DI students were grouped together for the analysis presented here. Results are virtually identical when they are separated. A methodological appendix to this article (Appendix A) provides descriptive results separating the two cohorts and details of the multivariate analyses separating the cohorts are available upon request from the author.

⁵Mac Iver and Kemper (2002) analyzed data from some of the schools included in this analysis and found that the DI schools had a lower rate of retention (“failing” a grade) than other schools in the system. This could conceivably result in a larger proportion of potentially “low achieving” or “higher risk” students in the fifth grade in the DI schools and thus provides a conservative test of the impact of DI on changes in achievement from first to fifth grade. However, when students who were in first grade in 1997 and were in fifth grade in 2003 (six years after first grade, rather than five) were included in the multivariate analyses, substantive results were virtually identical to those reported in this article. These results are also available upon request from the author.

⁶There was no way to control for the possibility that a student had attended multiple other schools between first and fifth grade. It is possible that some of the students in the panel sample were in the targeted schools in both first and fifth grade, but had attended other schools in the interim. Assuming that such children would have less of the “full treatment,” this would bias results in a conservative direction.

⁷McLean, O'Neal, and Barnette (2000) cautioned that effect sizes calculated with NCE scores are inherently smaller than those with other metrics. Thus, the effect sizes presented might be a conservative estimate.

Note. NIFDI = National Institute for Direct Instruction; DI = Direct Instruction; BCPSS = Baltimore City Public School System.

Note. The percentiles in Table 3 correspond to the average (M) Normal Curve Equivalent values given in Table 1. They may be interpreted as the percentile at which an average student in a given group and grade would score. Median values for all scores are also given in the Appendix to this article.

Note. Simple one-way analysis of variance comparing scores between the three groups for each measure and grade also indicate significant differences: for first grade vocabulary, F = 75.991, p < .001; for first grade comprehension, F = 24.528, p < .001; for fifth grade vocabulary, F = 6.69, p = .001; for fifth grade comprehension, F = 2.98, p = .05.

Note. No coefficient is included for Open Court because it is the omitted category for the dummy variables. FRL = free and reduced-price lunch; DI = direct instruction; BIC = Bayesian information criterion. LL = −2log likelihood statistic.

⁸We also tested interaction effects, specifically the interaction of group and school characteristics and the interaction of group and first-grade achievement. Neither of these sets of interaction effects was significant.

Note. DI = direct instruction; BCPSS = Baltimore City Public School System. These figures omit students who were in schools that began DI after they were in first grade (n = 341). Control students in first grade in 1998 had a mixed reading curriculum; control students in first grade in 1999 had Open Court.

Note. The “not retained” group includes all other students within the system but, as noted in the text, may include children who were retained in earlier years. Ns for each sub-group are given in Table A-1. NCE = Normal Curve Equivalent; DI = Direct Instruction; BCPSS = Baltimore City Public School System.

Note. DI = Direct Instruction; BCPSS = Baltimore City Public School System.