Turner presented results of the CTBS reading comprehension test, administered in April 1991, and again in 1992 and 1993. The 1991 test can be considered the pretest. For grades 6 and 7, the 1993 test is the posttest, for grade 8 the 1992 test can be considered a posttest. Those who were sixth graders at the start of the study improved 10.5 percentile points relative to norms in two years, seventh graders grew less than two percentiles in two years, and eighth graders grew 10 percentiles in one year. I performed a matched pair t-test on the data (table 1). The gains were statistically significant for grades 6 and 8, but not for grade 7. The small sample size and the unusual variability in scores of the seventh graders on the pretest may have played a role here (note that the standard deviation for the pretest is high, sd = 24.5).

Table 1: Gains made in three grade levels using AR

From: Turner (1993).
Standard deviation in parentheses.
d = effect size (mean of post test minus mean of pretest/pooled standard deviation)

Turner claims that "increasing reading activity had a positive relationship with improved reading comprehension" (p. 85), noting that those students who earned 40 or more points made better gains on the CTBS (in terms of NCEs, normal curve equivalents) than those who earned fewer.

To determine whether more points earned resulted in more gains, I performed a more detailed analysis, based on data supplied in Turner's paper. Multiple regression allows us to measure the effect of AR points earned on CTBS gains, controlling for the effects of previous reading ability. The results are surprising: For grades 6 and 8, those who earned more points made lower gains, but the effect was not statistically significant (for grade 6, b = -.134, p = .11; for grade 8, b = -.165, p = .24). For grade 7, those who earned more points made significantly greater gains (b = .128, = .02). This analysis shows that AR had a positive effect in grade 7, but not in the other grades, contrary to Turner's conclusion.

Potter (1994) reported on a middle school program that used "an accelerated reader program" and the Electronic Bookshelf. After two years, students improved 2.4 percentiles on the California Test of Basic Skills (CTBS), and Chapter I students improved 6.87 NCE's (Normal Curve Equivalents). Potter also noted that after four years "the school is still using the same program ... with success" (p. 245). Potter does not explain why he reported school-wide scores for two years, Chapter I scores for one year, and does not provide scores for the entire four years. Nor are we told why NCE's are used in one case and percentiles in the other. Nor were means, standard deviations, or sample sizes provided. Nor are we told the name of the middle school; thus, there is no way to confirm this data.

Goodman (1999) reported on the impact of an Accelerated Reader program in a middle school in Arizona. Students were in grades 7 and 8, ages 12 to 15. AR was accompanied by an increase in the number of books available (an increase of 2000 books for a total of 282 students), which could have had a large impact, as about half (46%) of the students were classified as "economically disadvantaged" (p. 18). It was also clear that students were given special reading time in the classroom and library (pp. 13-14). More access to books was also provided through public libraries, during the summer. The duration of the study was one full year.

The results are presented in table 2 in terms of grade-equivalent scores. These gains are not impressive. In one year, these students gained only three months in reading comprehension.

2