All charter schools are not alike. They're not supposed to be: The idea is to experiment with different ideas to see what works. But most studies of charter school effectiveness don't distinguish between wildly different schools: Back to Basics Elementary and Classic College Prep are lumped in with Waldorfish Learning Community, EthnoEsteem Escuela and Virtual Home Study for Alienated Youth.

Rand's new state-commissioned study of California charters doesn't look closely at curriculum, but it does distinguish between start-ups and charters created by converting an existing public school to charter status. It also compares classroom-based schools with schools that supervise home-based independent study. The differences are significant, it turns out.

Controlling for student characteristics, classroom-based start-up charters are more effective than conventional public schools; start-up students earn higher scores in elementary reading and in secondary reading and math. Furthermore, established classroom-based start-ups do better than brand-new ones, which suggests the gap may widen as more start-ups hit their stride.

Classroom-based conversion charters do about as well as conventional schools, while home-study charters are considerably less effective.

It's not surprising that converted schools aren't wildly different from district-run schools: Inertia (my favorite law of physics) prevails. Teachers have voted to convert; most stay with the school. Parents keep enrolling their children because it's the nearest school, not realizing that it's now a school of choice. It takes time for a conversion charter to create a new, distinct identity.

Some schools go charter to prevent change: Teachers, fearing a meddling district office will interfere with a good thing, vote to convert their school to charter status.

The lower results at non-classroom charters could indicate the model isn't very strong, but Rand researchers were cautious about drawing too many conclusions. Students who enroll in home-study programs are the kids who couldn't make it work in a traditional classroom. Almost certainly their problems aren't reflected completely in the socioeconomic factors that Rand used to control for student characteristics.

As many other studies have shown, charters don't skim the socioeconomic elite. California charter students are more disadvantaged, and more likely to be black, than the norm.

Parents participate more in school activities at charters.

Generally, charters spend as much time as conventional schools on various subjects, but there are some differences. Elementary charters are much less likely to offer bilingual education but more likely to teach a foreign language; charters also spend more time on fine arts. At the middle school level, charter students spend more time on math. Computer science is stressed a bit more at charter high schools.

Much of the reporting on the Rand study followed the money: California charters are producing similar gains for less money.

Classroom-based start-ups usually are in the worst financial shape: Most start-ups have to use operating expenses to cover the cost of classroom space, which can be very pricey in California. That's supposed to change: State law now says districts must provide equivalent facilities or funding to students who attend charters. But in 1999-2002, when Rand was conducting the study, paying for facilities was a drain on start-ups' funding, taking as much as 20 percent of the budget.

In addition, charters also get less than their share of categorical funds for federal and state programs. In some cases, the administrative burden of seeking and accounting for categorical funding was more than a small charter could handle.

Not surprisingly, charters -- especially start-ups -- offer fewer special programs of all kinds. Disabled students are likely to be mainstreamed. So are gifted students.

Charters -- especially start-ups -- raised considerably more private donations than conventional public schools. However, even with private money, start-ups had less to spend.

Of course, charters also hire more inexperienced and uncredentialed teachers, who are cheaper. Charters are more likely to provide shadowing, mentoring and other professional development help for teachers.

The Rand study jibes with other recent research, including a Hoover Institution study that showed charter schools are improving at a faster rate on California's Academic Performance Index.

Charters aren't a "silver bullet," Rand researchers warned. Nor a "cure-all." No kidding.

I'm writing a book on a start-up charter school, San Jose's Downtown College Prep, which targets low-achieving Hispanic students with college aspirations. The school's pioneer class is now entering 12th grade. In a year, the school will know how many graduates make it to four-year colleges: 100 percent is the goal.

I've seen all the issues in the Rand study in real life. Starting a school from scratch is much harder than converting an existing school, yet it's much easier to innovate when there's no status quo.

Joanne Jacobs is writing Ride the Carrot Salad: How Two Grumpy Optimists Built a Charter School. She blogs on education and other issues at www.JoanneJacobs.com.