Public charter schools, and KIPP in particular, have experienced persistent critiques that when charter schools produce positive performance outcomes, the results are driven by assumed charter school policies that counsel out low-performing students and restrict replacement enrollment (see Kahlenberg and Miron). This student attrition issue for charter schools is complicated. On the one hand, there are many charter school operators that strive to create schools that meet the needs of all students. From KIPP’s annual report: “At KIPP, we are committed to keeping our students with us because we believe that every student can thrive in our schools.” On the other hand, student mobility and the selection of schools based on personal preferences are inherent to school choice policies. Matt Di Carlo at the Shanker Blog (not the typical defender of charter schools) points this out: “Within-district attrition – students changing schools, often based on ‘fit’ or performance - is a defining feature of school choice, not an aberration.” And he questions why supporters of school choice are unable to clearly articulate and stand behind the fact that student mobility and non-random selection of schools will be a part of choice-based school systems.
Because KIPP is aiming to create an alternative educational model to the traditional public schools, a model that will serve any and all students—not just a small niche of students—it has been important for KIPP to show that its impact on student performance is not due to harder to educate students leaving for other schools.
Enter a new study by Mathematica that takes a very thorough look at attrition for 19 KIPP middle schools in nine states plus the District of Columbia and comparison middle schools in geographically relevant school districts (read more about the study here). There have been other studies that have taken a look at attrition in KIPP schools (San Francisco Bay Area KIPP schools, KIPP Lynn). However, the Mathematica study examines attrition exclusively and dives deep. Here are some of the findings:
- Cumulative and grade level attrition rates were similar for KIPP middle schools and comparison traditional public middle schools.
- The characteristics of students (i.e., race, FRL, baseline test scores) were the same for students who left KIPP as for students who left comparison traditional public middle schools.
- In terms of late entry enrollment, traditional public schools admitted a higher average number of students than KIPP middle schools. However, when looked at in terms of proportionality to total enrollment by grade level, there were no differences between KIPP middle schools and traditional public schools in terms of late entry enrollment.
- The demographics of late arrivers to KIPP did differ from students who entered traditional public schools after the middle school entry grade level. KIPP late arrivers were more likely to have higher baseline scores, less likely to qualify for FRL, and less likely to be male.
The study finds that there may be some selective replacement of students in KIPP middle schools. The question, then, is whether the demographic changes due to attrition and late entry explain the impact KIPP charter schools have on student performance. Based on the literature on the size of peer effects (i.e., did the change in peer demographics create a learning environment more conducive to student performance gains), the authors conclude that attrition and late entry of students explain no more than a quarter of the KIPP impact on test scores.
Let’s put that in perspective. The Betts and Tang meta-analysis of charter school effects found that KIPP schools had estimated effect sizes for reading and math at 0.096 and 0.223, respectively. This means that a student attending a KIPP charter school, compared with a traditional public school, would move from the 50th percentile to the 54th and 59th percentiles in a single year. If attrition accounts for roughly 25 percent of the effect, then the effect sizes for KIPP would drop to 0.072 for reading and 0.167 for math. The meta-analysis found that middle school charters (excluding KIPP schools) had effects sizes of 0.011 for reading and 0.055 for math. So, even when attrition is accounted for, KIPP middle schools outperform their traditional public school counterparts and have effect sizes three to seven times larger than other charter school middle schools.
Evidence is mounting for the scalability of the KIPP charter school model.
Image: A KIPP charter school in the Bronx (By Leila Haddouche, via Flickr Creative Commons)