Monday, September 13, 2004

previous entry | main | next entry | TrackBack (4)

Charter school update

Last month there was a kerfuffle when the New York Times splashed a shoddy American Federation of Teachers study suggesting charter schools were a buit on their front page. Click here for the roundup.

This month, EduWonk's Andy Rotherham alerts us to a more sophisticated study by Harvard economist Caroline M. Hoxby. This is the abstract:

This study compares the reading and mathematics proficiency of charter school students to that of their fellow students in neighboring public schools. Unlike previous studies, which include only a tiny fraction (3 percent) of charter school students, this study covers 99 percent of such students. The charter schools are compared to the schools that their students would most likely otherwise attend: the nearest regular public school and the nearest regular public school with a similar racial composition. In most cases, the two comparison schools are one and the same. Compared to students in the nearest regular public school, charter students are 4 percent more likely to be proficient in reading and 2 percent more likely to be proficient in math, on their state's exams. Compared to students in the nearest regular public school with a similar racial composition, charter students are 5 percent more likely to be proficient in reading and 3 percent more likely to be proficient in math. In states where charter schools are well-established, charter school students' proficiency "advantage" tends to be greater.

As Rotherham observes:

Rather than the NAEP sample data which has garnered so much attention, Hoxby was able to analyze almost the entire universe of 4th-graders attending charter schools and compare their achievement in reading and math on state assessments to students at the schools they most likely would have otherwise attended. Where 4th-grade data was not available she used 3rd-or 5th-grade data. It's a much more sophisticated study than the recent AFT report.

I await with bated breath the NYT's splashy front-pager on this charter school study.

UPDATE: That breath will be bated for quite some time.

posted by Dan on 09.13.04 at 11:51 AM


Well, just don't held your breath yet.

posted by: BigFire on 09.13.04 at 11:51 AM [permalink]

Dan seems to finally get it. The New York Times is a second rate publication. One should usually take anything it prints as a news story with a huge grain of salt.

posted by: David Thomson on 09.13.04 at 11:51 AM [permalink]

While there is obviously no (or a very small) margin of error--students doing 3-5% better is hardly something worthy of bragging rights (reminds me of people being happy because instead of horrendous job creation there is only lousy job creation, whoo-hoo, let me break out the champaign now). Basically the study says students in charter schools do a teeny-tiny bit better than their public school counterparts.

The Times reported on a study that had lousy methodology. It would be nice if they published an update, but 3-5% improvement is hardly something to crow about. Now, I am actually in favor of charter schools, but really, this is not a study worth jumping up and down about and saying "see-see!!!"

And David, as someone who used to work as a journalist, can you give me an example of a better papaer than the New York Times? Perhaps the Washington Times?

posted by: Kate on 09.13.04 at 11:51 AM [permalink]

Wow, a whole 3-5 percent better?

posted by: praktike on 09.13.04 at 11:51 AM [permalink]

No, not 3-5% better. 3-5% more likely to have achieved the standard of "proficiency". That's not the same. It might lead to the same policy conclusions, but the pedants among us would like to insist on the difference.

posted by: dsquared on 09.13.04 at 11:51 AM [permalink]

Pardon my ignorance on the subject, but why would the New York Times be prejudiced against charter schools? From the stats provided in the round-up, they serve primarily disadvantaged groups - wouldn't that be consistent with concerns about social equity?

I guess their objection might be on developing a parallel school system that might substitute, rather than complement; but from the Charter schools website, it doesn't seem like their private in any sense (again, I'm not at all familiar with charter schools.) I can understand why a Teacher's federation would oppose innovative new educational models (Last week's Nouvel Observateur had a great article about this tendency in France) but I don't understand why this would fit into the NY Times leftward stance.

Can somebody enlighten me?

posted by: Jamie on 09.13.04 at 11:51 AM [permalink]

(I maintain my own view, by the way, that pupils at charter schools are by definition drawn from the population of those pupils whose parents care about their education, which is a subset of the total population. Therefore, the best methodology for these studies would be to pair children who got into charter schools with children who applied but did not get in; I seem to remember that this study has been tried a few times, with mixed results but in general showing positive results.

Though I suspect that most of the positive results are Hawthorne effect on the teachers ... we have had a system of "parent choice" more or less equivalent to charter schools in the UK for ten years, without noticeable effect on the amount of griping about educational standards)

posted by: dsquared on 09.13.04 at 11:51 AM [permalink]

Teachers' unions (and teachers in general) hate anything that looks even a little bit like "payment by results". The entire psychology of being a teacher is that you get to judge other people, not that other people get to judge you. Ask Dan how much of his pay he would like to be dependent on students' teaching assessments :-)

posted by: dsquared on 09.13.04 at 11:51 AM [permalink]

Since charter schools are to a great extent self selecting the worse students out of their student body to only acheive a 2% to 5% better score does not say much for the charter schools.

posted by: spencer on 09.13.04 at 11:51 AM [permalink]

Spencer: My understanding is that charter schools are in general not selective (other than that, as I noted above, they select pupils by lottery from the population of pupils whose parents care about their education rather than from the population as a whole). And I disagree that 5% is a small improvement here; it suggests that in a typical school class, there is one child who does not become proficient but would do in a charter school.

posted by: dsquared on 09.13.04 at 11:51 AM [permalink]

So, it sounds like charter schools are a very modest success.

My question: what is the selection process for students at charter schools? If they are self-selected in any way, then it seems reasonable to expect that their students would do modestly better than average, no matter what. (It's not that average people won't try to do self selection; it's that the ones who don't give a damn will not.)

posted by: mac on 09.13.04 at 11:51 AM [permalink]

There's some evidence that students who switch to charter schools are actually below average in achievement. The proposed explanation is that students who are doing well in public schools usually stay. Those who don't fit in are the ones who go to charter schools. I teach in a public middle school,and this seems to be true of my town's charter school.

I think the main source of this argument is a book or article by Terry Moe. I don't consider him completely reliable--a real pro-school choice ideologue--but I think I've seen the findings confirmed by others.

posted by: Rob on 09.13.04 at 11:51 AM [permalink]

This is good stuff! Better social science, but as dsquared and others have pointed out, not really adequate.

1) Charter schools may have some control over whom they admit. Self-selection
2) Parents who go through the trouble of enrolling their kids in a charter school may be more interested in education than those who dump the kid off at the nearest public school. This is turn might influence the outcome. dsquared's test -- those that got in versus those that didn't -- for this variable, as in all cases the parents were motivated to get their kids into the charter school. But, obviously, the kids who didn't get in would be a less able group in all likelihood.

3) Maybe the best way to get at this would be to match charter schools with similar socioeconomic data with those that have no charter school in a reasonable commuting distance.

The Harvard econo must have a great data set -- put it on the web for a seminominal download fee!

posted by: stari_momak on 09.13.04 at 11:51 AM [permalink]

A brief scan of an NBER search reverals that this woman has been the author of a number of papers on this topic. I'm not doubting her research, but I think I need to read more research on this, specifically from other researchers.

posted by: Brian on 09.13.04 at 11:51 AM [permalink]

Self-selection bias is inherent in charter schools for the reasons outlined by other posters above. Charter schools are a good idea but they are really better suited to serve niche populations with innovative or focused curriculum. The more charter school populations and teaching methodology resemble the local public school the more likely their results will mirror those of the local public school.

And secondly, if grade inflation at 1st tier universities is any clue to what happens when departments take student assessments seriously then tying teacher pay to aggregate parent happiness is apt to create a universe where the lowest acceptable grade is an "A-".

posted by: mark safranski on 09.13.04 at 11:51 AM [permalink]

By the way, I've always considered it to be the height of silliness to complain about what a newspaper chooses to put on its front page. If you don't like it, buy a different newspaper.

(Of course, this is slightly unfair of me; it's not Dan's fault that he lives in a country of 225m people with only two decent newspapers!)

posted by: dsquared on 09.13.04 at 11:51 AM [permalink]

“And David, as someone who used to work as a journalist, can you give me an example of a better paper than the New York Times? Perhaps the Washington Times?”

The Washington Times is far more accurate than the New York Times. Its weird owner, Revered Moon, thankfully has nothing whatsoever to do with running the paper. However, the Washington Times also lacks the financial resources to truly compete with the much larger entity. This is why the New York Times is so dangerous. Not only is it overall a dishonest publication---but nobody else can even begin to compete with its deep pockets.

The NY Times no longer earns the trust of even those who describe themselves as politically moderate. I encourage everyone to visit the respective blogs of Roger L. Simon and Glenn Reynolds. Neither gentleman is a conservative of the National Review variety, but both hold the NY Times in general contempt. Daniel Drezner is only now starting to wake up to reality. Should you never read the NY Times? Sadly, that is virtually impossible. There are too many guest writers and opinion pieces worthy of your attention. It’s the so called straight stories that usually require one to be skeptical.

posted by: David Thomson on 09.13.04 at 11:51 AM [permalink]

I was paid good money to do oppo research on a Hoxby expert-witness report, and it didn't take me long to earn it. As a start, she (more likely, some grad student assistant) stated that California had the highest per capita income. It's Connecticut. California isn't even in the top ten. Corrected, her analysis supported our position at least as well as hers. It went downhill from there. All of the methodology was flawed.

I have no doubt that BOTH reports are statistical garbage concocted to support pre-existing opinions. (I asked if I could look at our side's expert witness report, and the answer was not a quick "Sure!")

posted by: Andrew J. Lazarus on 09.13.04 at 11:51 AM [permalink]

dsquared got it right. Hoxby is capable of much more careful impact evaluation methods. From what I can tell from the short paper all she's done is pair scores from charter schools to nearby schools. There is no before and after and there is no effort to correct for selection. Parents who move their kids are different from parents who do not. Hence the 3-5% may be explained entirely by having more involved parents than by anything the charter school is doing.
Certainly Hoxby is capable of a far more sophisticated methodology. I think she just didn't have the needed data.

posted by: zzi on 09.13.04 at 11:51 AM [permalink]

One the reasons this could be significant is that so many in the education establishment, ie, teachers' unions, school boards, are against charter schools. In Ohio, the "Coalition for Public Education" has filed a lawsuit claiming, among other things, that charter schools violate Ohio's Constition by 1)extending the State's financial support and credit in joint ventures with for-profit management companies;
2)siphoning locally voted tax levy dollars from public school districts;
3)causing diminished academic standards and 4)diverting funds from city school districts and frustrating their efforts to educate students. The Harvard study, while not showing that charter schools are superior, refutes, at least in part, that charter schools "cause diminished academic standards."

posted by: Rick on 09.13.04 at 11:51 AM [permalink]

The simple fact is that without knowing whether or not charter school students are better students going in to the school (and previous commenters have argued both ways) we don't know whether Hoxby's conclusions are newsworthy. If they are better (as I suspect they are due to parental interest), then the fact that they score better on tests says nothing about charter schools. Only a very well constructed time series study can show us anything meaningful about charter school performance.

posted by: Stuart on 09.13.04 at 11:51 AM [permalink]

"The Washington Times is far more accurate than the New York Times."

It's so worried about accuracy, it doesn't even run a corrections column.

posted by: Brian on 09.13.04 at 11:51 AM [permalink]

Charter schools can't pick and choose students on the basis of academics, race, etc. Overall, charter students are more likely than other public school students to be black, since many charters are created to serve underserved communities and high-risk students. (If you start a school for drop-outs, you can't exclude A students but you're not likely to get many to apply.) Previous studies have found charter students start out behind comparable students but make faster progress. However, there's not a lot of data, and there many kinds of charter schools with very different results.

posted by: Joanne Jacobs on 09.13.04 at 11:51 AM [permalink]


posted by: Max B. Sawicky on 09.13.04 at 11:51 AM [permalink]

Sorry for empty post. Wouldn\'t take my link:

posted by: Max B. Sawicky on 09.13.04 at 11:51 AM [permalink]

Post a Comment:


Email Address:



Remember your info?