The (Lack of ) Experimental Evidence Supporting the Use of Accelerated Reader

Stephen Krashen
Journal of Children's Literature (2003) vol .29 (2): 9, pp. 16-30.

The purpose of this paper is to review the experimental evidence on Accelerated Reader (AR), probably the best-known reading management program. Accelerated Reader consists of the following elements:

If we wish to study the effect of these programs scientifically, we should compare groups that are very similar to each other, and that receive treatments that differ only with respect to the factor of interest. Thus, in the case of AR, it would be useful to study the impact of treatments that differ with respect to each of the four components.

The research literature has already provided us with some of what we need. There is consistent evidence that those who have more access to books read more (Krashen, 1993), and that those students who are provided with more time to do recreational reading (e.g. sustained silent reading programs) show better gains in reading achievement than comparison students. The effect is especially strong when such programs are allowed to last for one year or longer (Krashen, 1993; 2001; Pilgreen, 2000). On the basis of this research it would be expected that reading management programs that provide reading time and access to books will indeed result in gains in reading comprehension. The issue is whether the addition of tests and rewards makes any difference.

In addition, it is important to determine the long-term effects of rewards. A classic argument against the use of rewards is that it affects only short-term behavior (see McLoyd, 1979, discussed below).

This survey of the research is an update of a previous review by McQuillan (1997) on the effect of incentives, but focuses exclusively on the effect of AR on tests of reading comprehension. It begins with a review of studies that utilize standardized tests to measure student progress. Standardized test scores allow us to compare the progress of the group receiving the treatment with a norm, usually a national norm. This kind of information is useful, but comparison with a national norm is often not appropriate. It is far preferable to compare the treated subjects with a group that is similar to them in terms of academic achievement, poverty, language background and other variables that can strongly influence performance.

The next category of studies attempts to compare those who participated in reading management programs with comparison students who did not. This solves the problem raised in discussion of the previous category of studies, but do not solve the problem of which aspect of AR is responsible for gains, if indeed gains exist.

We will see that very few studies attempted to provide a comparison group that was also engaged in recreational reading. The lack of these studies is a serious gap. It is firmly established that providing access to books and time for free reading is effective. AR differs from free reading only because it adds tests and rewards. If there is no evidence providing clear support for the use of these factors, there is no evidence in support of AR.

Performance on AR Students on Standardized Tests of Reading Comprehension

Turner (1993) examined the impact of AR in a New Jersey middle school. The treatment included one hour per week of sustained silent reading, trips to the public library, and additional literature added to the language arts class. The school involved had an adequate school library (19.4 books per student). Those in grades 6 and 7 at the start of the study had the treatment for two years, those in grade 8 had the treatment for one year. AR included the usual tests and prizes, including pizza and ice cream parties. In addition, "... the names of the top scorer in fifth, sixth, seventh and eighth grade were etched into a permanent plaque outside the library ..."(p. 59).

1