• In 1930, there were 260,000 schools in the USA (including approximately 150,000 ones with only a single teacher). In 2000, there were less than 95,000 schools almost none with only a single teacher. The average number of students per school size grew from 89 to 502.
  • The number of support staff per student has increased three-fold since 1950, from 1/83 to 1/27.
  • The number of teachers per student has increased two-fold since 1950, from 1/26 to 1/12.
  • The average number of years of experience for a teacher has increased 7 years since 1966, from 8 to 15. The average age of a teacher has also increased by 7 years over the same time period, from 37 to 44.

The statistics are from School Figures, which is published by the Hoover Institute. I hadn't heard of them before, but from reading their choice of statistics it's clear they're in the privatization/school-choice crowd. The idea of school choice had an appeal for me at one point, when I still believed that what made a school good could be measured by tests. School choice advocates are entirely right that if you place enough emphasis on competing for students via test scores and the school ratings based on them, you'll see scores go up. The problem with that approach is that any kind of test scores are basically broken as a measure of the success of schools.

For the first part, it's difficult to say whether test scores are truly rising or falling at all. There are a huge number of confounding factors involving demographic trends and changes in the tests themselves. Secondly, even assuming we could correctly tease out the true change in test scores, there's very little evidence that rising test scores mean better educated students. It's difficult to deny that students who score 1600 on the SATs are much more likely to be well educated and thoughtful than those who score 900. But as those 1600 point scoring students would tell you, corelation is not causation! It might be that on average tall students score higher on their SATs, but that doesn't mean giving students stilts will improve test scores.

This is all a roundabout way of saying that I don't think much of the book, although it has a lot of pretty graphs. They've managed to pull together a very large and impressive number of statistics, a few of which are meaningful and most of which are not. Other offenses include, but are not limited to: extrapolating trends from as few as two data points, representing public opinion polls as meaningful measures of something more than people's opinions, and failing to consider alternate explanations for trends. I shouldn't pick on them too much, since all of these are common crimes, but I wish for a book on the history and facts of education that I could actually recommend at some point.