Middle School is Bad

There is a reason why there are so few private or charter middle schools

Most California school districts put their students through three types of school: an elementary school that starts in kindergarten (or preK these days), a middle school that starts in sixth or seventh grade (but sometimes fourth or fifth), and a high school that starts in ninth grade. 86% of 7th graders attending a district school are in a middle school of some sort and 11% are in a K-8 school. Other grade configurations are rarely found.

Charter and private schools do not follow this model. Among 7th graders, just 30% of those attending charter schools and only 4% of those attending private school are in middle schools. A large majority are in K-8 or K-12 schools but a chunk are also in schools that combine middle and high school grades. Public school districts almost never offer this grade configuration.

This is not just a curious accident of history. It has an effect on children’s education.

In the last post, we saw that the SBAC Math scores of San Francisco of 6th graders were lower than those of its 5th graders. That they were actually lower was a novelty but the weakness of 6th grade was not. Students in 6th grade in San Francisco always learn less than students in any other grade. It’s unlikely that the district is systematically assigning its least effective teachers to that grade so what’s the problem?

Fortunately, we have other districts to look at and we benefit from the variety of grade structures found in California school districts. 51% of 7th graders started middle school in 6th grade. Another 35% started middle school in 7th grade. Just under 12% are in a K-8 or K-12 school. We’ll forget about the 2% of students whose middle school started in 4th or 5th grades.

SBAC Scores by Grade Configuration

Most school districts have a standard grade configuration. In some, the elementary schools are K-5 and the middle schools are 6-8. In others, the elementary schools are K-6 and the middle schools are 7-8. There are a couple of districts that exclusively use K-8 schools but most K-8 schools are in districts where the dominant model is either K-5 and 6-8 or K-6 and 7-8.

Here’s a chart showing the average SBAC scores of students depending on the grade configuration of the schools they attend.

I attach no significance to the difference in scores at the 3rd grade level because demographic factors provide an obvious explanation. The Bay Area counties, which are all among the highest scoring in the state, all have a majority of their students in K-5 schools. The counties that are majority K-6 include Fresno, Kern (e.g. Bakersfield), Monterey, Orange, Sacramento, San Bernardino, Stanislaus, and Tulare. Of these, only Orange is traditionally high scoring. Similarly, the scores for students in K-8 schools are low because the districts with the greatest number of students in K-8 schools are Stockton, Manteca, and Compton.

What is interesting is the trajectory of scores after 3rd grade. The K-5 and 6-8 schools retain their lead over the K-6 and 7-8 schools at every grade except 6th grade. My interpretation is that starting middle school is disruptive to learning. The 6th graders who have just started middle school don’t progress as far as they would have if they’d still been in their elementary schools and this allows the 6th graders who are still in their elementary schools to catch up. The situation then reverses the following year. The kids whose middle schools start in seventh grade are the ones who don’t progress as far. The kids who are in their second year of middle school have adapted to their new schools, progress normally, and reopen the gap that had existed when everyone was in elementary school. The students in K-8 schools never experience this down year. In elementary school, their scores significantly lagged those of students in other grade configurations but, by the end of 8th grade, the gap has narrowed considerably.

Exactly the same pattern exists for Math scores.

The pattern become clearer if, instead of looking at the actual scaled scores for each grade, we look at the cumulative change in the scaled score since third grade. This enables us to control for the different grade 3 scores. Also, to avoid the allegation that the inclusion of charter school data is distorting the results, I’m just going to exclude all charter schools and focus just on students in district-operated schools.

We can see that students in K-5, K-6, and K-8 schools grow at almost exactly the same rates in grades 4 and 5. That is reassuring because I can’t think of any reason why they would be different. The divergence starts in 6th grade. The students who have started middle school suddenly learn less than those who remain in their elementary schools. The kids who start middle school in grade 7 immediately start to lose their lead over those who started middle school in grade 6. The kids in K-8 schools, who never have to switch to a new school, sail blissfully on and end up learning more than the others.

The data is not as clear cut for Math. K-8 students don’t do any better in 6th grade than students just starting in 6-8 middle schools and the jump for kids still in K-6 schools seems almost too big. Nevertheless, the core point remains true: sixth graders in K-6 schools do jump ahead of those starting middle school only to lose that advantage in 7th grade when they themselves start middle school.

Disadvantaged Students

We saw earlier that students in K-5 schools score higher in 3rd grade than students in K-6 or K-8 schools. I attributed this, casually and without proof, to demographic factors. Nevertheless, it’s natural to wonder whether all types of students are affected by the middle school transition. Here are the last two charts again, this time just looking at socioeconomically disadvantaged students. First ELA:

When we restrict it to disadvantaged students, the transition to middle school clearly affects Math performance too.

By the end of grade 8, students in K-8 schools have gained about 13 more points in ELA and 8 more points in Math than students in K-5 and 6-8 schools. To get a sense of the magnitude of this difference, recall that, in the last post, we saw that a 3 point difference in the average scale score was roughly equivalent to a one percentage point difference in the overall district proficiency rate. Raising the average score by 13 points would thus increase a district’s proficiency rate by over 4%. That’s a substantial effect for something as apparently innocuous as grade configuration.

Conclusions

I’m acutely aware that, although the results are suggestive, this does not constitute proof of the superiority of the K-8 configuration. The Calculation Notes below describe some of the weaknesses in my approach and there are probably others I haven’t thought of. Further study with better data would be required in order to definitively establish the superiority of K-8.

In the meantime, if I were running a school district, I’d be trying to bolster my K-8 schools. If I were running a middle school, I’d be trying to figure out what can be done to lessen the damage caused by the middle school transition. If the root cause is the difficulty of adjusting to a new environment with new classmates, perhaps middle schools could make a conscious effort to keep students from the same elementary school together in 6th grade and to keep classmates together for more subjects. Being surrounded by familiar classmates would provide more continuity and might produce better results than assigning students randomly to classes.

Calculation Notes

Most districts have either K-5 elementary schools and 6-8 middle schools or K-6 elementary schools and 7-8 middle schools. We’d expect that the number of students per grade in K-5 schools should be about the same as the number of students per grade in 6-8 schools because nearly all of those who graduate a K-5 school will move on to a 6-8 middle school. Similarly, the number of students in grade 6 of K-6 elementary schools should be about the same as the number of students in grade 7 of 7-8 middle schools. When I first ran the numbers, I found that 6-8 middle schools had about 20,000 more students than I expected and 7-8 middle schools had about 20,000 fewer. I had been classifying a school as 6-8 if it had SBAC scores only for grades 6, 7, and 8 but this obvious approach ignores some real-world complications:

• Most districts use either K-5 or K-6 but not both. A few districts have a mix which means some middle schools in those districts have a mix of K-5 and K-6 schools feeding in to them. The middle school’s 6th grade just contains kids from the K-5 feeders while the 7th grade contains kids from the K-6 feeders as well as those already in the school from the K-5 feeders. Bakersfield’s middle schools have this pattern. For example, Walter Stiern Middle goes from 121 in grade 6 to 305 in grade while Curran Middle goes from 232 in grade 6 to 398 in grade 7.

• Some middle schools are home to two different programs that have different grade spans. In Los Angeles, Mary McLeod Bethune Middle serves grades 7 and 8 but it is also home to a STEM Magnet program that serves grades 6-8. The school has 24 6th graders (from the Magnet program) and 413 7th graders (from both the Magnet program and the regular middle school).

• Some middle schools primarily serve grades 7 and 8 but also offer a small grade 6 “Academy” to prepare students for middle school. For example, Cesar Chavez Middle School in San Bernardino had 93 students in grade 6 but 502 in grade 7 and 491 in grade 8.

I ended up classifying such schools depending on which cohort was the bigger. A school with 120 6th graders and 300 7th graders was classified as a 7-8 school whereas a school with 180 6th graders and 300 7th graders was classified as a 6-8 school. These reclassifications affected about 10,000 students but didn’t materially change the average scores of any group.

There is a more fundamental problem. My analysis is comparing this year’s 6th graders as a group with this year’s 5th graders as a group. It would be much better to look at individual students and measure their actual growth from last year to this year, and group them according to whether or not they were attending the same school this year as last year.