JREE published a long-term evaluation of Reading Recovery (RR) - a tutoring program for 1st graders - that found adverse effects on grade 3-4 reading achievement, and received wide coverage (e.g., recent NY Times). Quick take: This is not a reliable finding, due to sample loss (attrition) of over 70% and other problems.

​

Program:

​

• RR provides one-on-one tutoring to struggling readers in 1st grade. It has previously been found in a high-quality, federally-funded RCT (N=9,784 students) to produce large gains in reading achievement in 1st grade, but the RCT didn't measure longer-term impacts in later grades.

 

Study Design & Findings:

​

• The new study used a comparison-group design (not RCT) to measure RR's long-term impact, & found sizable *adverse* effects on grade 3-4 reading scores. But the study lost 73-83% of its sample by grade 3-4, creating "unacceptable risk of bias" under established standards (WWC).

• The study had other substantial problems - e.g., the treatment group's baseline test scores were far below the comparison group's (~0.85 SDs), so the study relied heavily on statistical modeling to try to equate the two groups - an approach that's been found unreliable.

 

Comment:

​

• It's unfortunate that the NY Times, NPR, & others would give this study largely uncritical coverage.

• Meanwhile, a large, high-quality RCT of RR with long-term follow-up has never been conducted. We need one. It's important to know if the sizable effects found in 1st grade fade out over time, or lead to longer-term educational gains.