Well, this was fascinating.
In an era when reading scores are continuing to fall, The 74 just ran an article about schools that were beating the odds in literacy rates. In a very large study, the authors compared predicted literacy rates based on each school’s poverty level to actual literacy rates for third graders in the school.
Let’s talk about how they designed the study:
- They looked at a large number of elementary schools — 41,883 individual schools.
- They defined poverty primarily by using the government’s Free and Reduced Lunch (FRL) guidelines.
- They looked only at 3rd grade reading scores since by this point a lot of the nuts and bolts of the learning-to-read process is completed and kids are transitioning to reading-to-learn. It’s also the year that most states measure literacy, often to decide whether a child needs to be retained, so the apparatus for assessment is in place.
- The study also included public charter schools. In some states, public charter schools have some of the highest poverty rates, so this is an important subset to include. They may also be valuable since some are doing truly innovative work in the classroom.
The sheer size of the sample is important because it lends validity to the findings, but the findings are less incisive than I’d like. They give rise to a lot more questions than provide answers. In several states, poverty and literacy were highly correlated. In Washington, Maryland, Georgia, Connecticut, Colorado, Nebraska, Minnesota and the District of Columbia, poverty strongly predicted literacy, with high poverty correlated with lower rates of literacy and low poverty the reverse. In other states like Kentucky, Nevada, West Virginia, North Dakota and New York showed a much weaker relationship between family income and literacy rates. Why this is is not clear. The authors speculate that it may lie with a fundamental difference between states, namely how they assess reading ability in grade 3. There is no national tool for measuring reading ability so districts and schools choose what they want to use. The study compared the poverty rate for each school and that’s important because some districts have a very uneven distribution of economically disadvantaged kids between schools, with some schools having a small percentage of FRL kids and other schools having almost nothing but FRL kids. The authors don’t really offer reasons for differences between schools in a district other than poverty, but it’s possible rates may have something to do with the rigor and engagement of the instruction or of the written curriculum in use, or the experience and competence of the teachers at each school*. There may also be clusters of SPED or ELL students in some schools that affect total scores.
One fascinating finding was this: Public charter schools made up 7% of the total sample but 11% of the exceptional schools (those in the top 5% of their state in terms of outscoring their expected reading proficiency). Many of these exceptional charters were in densely urban areas like New York City and Philadelphia and many had high populations of FRL students. Why they should be overrepresented in the exceptional tier is also not clear; it may also come down to the quality of the instruction/curriculum or to alignment with the test, or maybe completely different factors we’re not aware of.
Some caveats:
- We don’t have a common, national definition for what we mean by “literacy,” so tools to assess this may be (definitely are) measuring really different skills from school to school across the country.
- Because of this inconsistency, school-level data can’t really be compared state to state, or district to district within a state, or even school to school within a district. IF a district uses a common literacy assessment in grade 3 (and many don’t), it would be possible to compare schools in that district, but we can’t see the tools used in this study. We can, however, look at the correlation between poverty and literacy and see how strong that is in the various states.
- The study doesn’t control for English Language Learners or any other subgroups. Understanding the demographics of individual schools is important. A school’s poverty rate may be mid-range but its population of ELLs or SPED students might be higher than other buildings and that’s going to affect scores.
- If you’re looking for specific causality, you’re not going to find it here. Nothing in the study attempts to quantify why correlations are or are not strong or what may be causing differences between schools. Given how the study is designed, identifying causality isn’t really possible.
But it is interesting. For my fellow data nerds, you can go to the article and scroll down to the interactive dot-plot graph and search by state and by district. You can see, for example, how strong the correlation between poverty and literacy is in your state, you can look at individual schools in your or a neighboring district, and you can see who the high flyers are for your state. Note: not all schools in every district were included so there are some gaps in the data.
Will this give us information we need to stop the slide in reading scores? Probably not. The overall finding seems to be that often poverty correlates to literacy and this is something we’ve known for decades. But it might shape other research and lead to other studies that try to account for what we see here and offer some practical steps to fixing the national reading problem.
_______________________________________________
*Research has demonstrated that the higher a school’s poverty rate, the harder it is to attract fully certified staff and also to retain staff with more than 5 years experiences. At-risk students end up being taught by what are essentially at-risk teachers. Low poverty schools, on the other hand, attract the most experienced, full certified, highly competent teachers out there. Sometimes low poverty schools are even regarded as a kind of reward for years of experience.