Anonymous wrote:This is all true if you choose a school based only on test scores. Does anyone really do that?

I think so from reading some of the threads on this forum.

Anonymous wrote:OP again. I must follow-up to this "if the IB families agree to send their children to Hardy, Hardy will quickly look like Deal."

Even if IB families don't coordinate well, there are indications that increasing numbers of IB students are already heading to Hardy. Each additional student makes future coordination all the easier. So, it's not like the families need everyone to sign off at once; piece-meal increases in IB and feeder enrollment will get the job done. It will just take a little longer.

I didn't say there is no other possible solution. Its just not the textbook solution.

Anonymous wrote:

Anonymous wrote:See this shows how stoopid economists are. They don't understand the important things in life, like school uniforms, they think it is all about test scores and academics.

Suppose my kid comes home wearing a uniform, and my neighbor sees it. They will think my kid goes to a ghetto school. I am supposed to start telling them about standard deviations and confounding variables?

Really? If you wear a uniform people will think your kid goes to a ghetto school? Are you kidding?

Well if I wear a uniform and my kid does not, we are probably a military family and our kid goes to FCPS.

OP again. I must follow-up to this "if the IB families agree to send their children to Hardy, Hardy will quickly look like Deal."

Even if IB families don't coordinate well, there are indications that increasing numbers of IB students are already heading to Hardy. Each additional student makes future coordination all the easier. So, it's not like the families need everyone to sign off at once; piece-meal increases in IB and feeder enrollment will get the job done. It will just take a little longer.

Anonymous wrote:

Anonymous wrote:See this shows how stoopid economists are. They don't understand the important things in life, like school uniforms, they think it is all about test scores and academics.

Suppose my kid comes home wearing a uniform, and my neighbor sees it. They will think my kid goes to a ghetto school. I am supposed to start telling them about standard deviations and confounding variables?

Really? If you wear a uniform people will think your kid goes to a ghetto school? Are you kidding?

I think this person is being facetious.

OP here.

As it turns out, I'm a game theorist. So, yes, I COMPLETELY familiar with such concepts as the prisoners' dilemma.

You're confusing matters. The tension in a PD is that everyone has a strictly dominant strategy: fink. Translated here, this is "going private" is the strictly dominant strategy. No one believes, not even yourself.

Instead, what you're actually describing is a coordination game. Stag Hunt is a good example. If you and I agree to work together, we can bring down a large deer and eat like kings. If I work alone, I can catch some rabbits. You too. But if I decide to hunt stag while you go after rabbits, I go hungry while you dine on Bugs Bunny.

Stag Hunt is more applicable here. That's the crux of the matter: if the IB families agree to send their children to Hardy, Hardy will quickly look like Deal. If you look below the surface, it already looks like Deal in several key ways.


An unrelated poster asked about "why would I send my kid to a lesser-quality school (Hardy) when I could just send them somewhere better (private)?" There is a difference in cost, you know. That should be a sufficient answer, but there are other compensating differentials as well.

Moreover, my entire point was that if Wilson is good enough for your child, Hardy is most certainly good enough for him too. If Deal is good enough for your child, you can make a solid case that Hardy is good enough for him too. That's what the data say.

Anonymous wrote:See this shows how stoopid economists are. They don't understand the important things in life, like school uniforms, they think it is all about test scores and academics.

Suppose my kid comes home wearing a uniform, and my neighbor sees it. They will think my kid goes to a ghetto school. I am supposed to start telling them about standard deviations and confounding variables?

Really? If you wear a uniform people will think your kid goes to a ghetto school? Are you kidding?

Anonymous wrote:
Notwithstanding that I fully believe that my kid will do well anywhere, why on earth would I send my kid to the lower performing school, if I had a choice?

The implication of the above is that there are costs to attending such a school for a high SES student (or a white high SES student) that have nothing to do with the student't test scores. One is that there are academic negatives not covered by the tests. More likely is that there are social unpleasantnesses apart from academics. "My kid aced the scores, and eventually went to Dartmouth, where he regaled his classmates with tales of how he was beaten up every day at DCPS"

To the extent that those other costs are relatively minor (not like in the hypothetical quote above) the reason to send the kid to this school is that a family does not want to move, and does not want to spend the money on private school.

I'm just going to respond to this part:

anonymous wrote: Since this is important, I’ll belabor the point: in DC, “white” is a clean proxy for high income. This does not mean all high income people are white. But what it does mean is that if you’re high income, whether white or AA, the best predictor of your child’s scores is given by the “white” average since this average represents only high income students.

This is not true. In DC "white" is a clean proxy for being the beneficiary of systematic racism. There are so many studies that say that AA kids from highly educated, affluent homes where parents own books, and read to them, and value their education, are not protected from racism. They are subjected to disproportionate discipline, and low expectations, which leads to achievement gaps that continue to exist even when income, parental education, time spent reading, and other factors are taken into account.

Similarly, there are plenty of white kids growing up in DC whose parents don't take them to the library, or have homes full of books, or give a shit about their education. But because of their skin color, teachers and others treat them as if they were growing up in households that do these things, and hold them to the same high expectations.

"I’m an economist and I find the Hardy discussion fascinating because the two “sides” are so directly opposed."

It is fascinating because it is such a perfect example of a prisoners' dilemma. At least 90% of the reasons advanced (and some kept hidden) for IB families to not send correlate with the demographics. Maybe 100% of the legitimate reasons. If all IB families sent, (or even all those IB families for whom private is not an easy choice against even the best DCPS MS) and most of the IB families who move to the suburbs out of fear of Hardy stayed, then Hardy the things motivating them to avoid Hardy would disappear. The textbook optimal solution would be either for all of them to make a legally binding contract to send their kids to Hardy, or for an outside force to coerce them (making it impossible for them to use private schools or to move). Neither of those is possible however, so all that can be done is moral suasion "we are going, you try too" (not necessarily compelling) or DCPS inducements "what a great new special program" which may involve resource conflicts.

Great analysis, and thank you. But what about the uniforms???

Kidding, kidding.

But to be serious, let me ask you this. Suppose I am white or black or yellow or whatever, but I meet all the bullet points for whiteness above.

And suppose we are discussing a hypothetic school that is 100 percent low income or non white or whatever descriptor we are using, where the kids all score terribly on the standardized tests.

And suppose I agree with your prediction that my "functionally white" kid will destroy the tests because we read at home and all that good stuff.

Let's also assume that I have money and thus options. I an afford to move, pay private, whatever.

Notwithstanding that I fully believe that my kid will do well anywhere, why on earth would I send my kid to the lower performing school, if I had a choice?

For years now, everyone has known that "white" (defined as you do above) kids do well wherever they go in DCPS. But why send your kid to a school with many struggling students when you could afford to send them to a school with few or no struggling students?

Also, seeing as you are an economist, you should consider the prisoners' dilemma that is unfolding here at the grade 5 to 6 transition, which to me is the most interesting thing about this whole conversation.

Thanks again for a carefully researched thread starter.

This is all true if you choose a school based only on test scores. Does anyone really do that?

You understand that Hardy was at a different location in 2006 / 2007 due to renovations. The student body was VERY different

See this shows how stoopid economists are. They don't understand the important things in life, like school uniforms, they think it is all about test scores and academics.

Suppose my kid comes home wearing a uniform, and my neighbor sees it. They will think my kid goes to a ghetto school. I am supposed to start telling them about standard deviations and confounding variables?

This is a long post. It starts with disclosures and ends with predictions. I will try to check in periodically over the coming day/days to answer questions.

Disclosure: I have no children at Hardy, Deal or Wilson. I have no children at a Hardy feeder. I have no immediately vested interest in any of these schools.
I’m an economist and I find the Hardy discussion fascinating because the two “sides” are so directly opposed. I decided to see what the data say.

Data: all data come from http://dcps.dc.gov/DCPS/About+DCPS/DCPS+Data/DCPS+Data+Sets. In particular, I will be using the DCCAS scores from 2006-2007 through 2013-2014. I will discuss the overall scores for reading and math as well as the scores for white students and the (implied) non-white scores. The numbers discussed are the percentage of students scoring advanced or proficient on the DCCAS reading and math exams. If there is a way to post the data in a clean format, I'd happily do that.

Whiteness: I will be using “white” as a proxy for several things. From an objective perspective, all white students in DC appear similar. They generally:
• Come from families with higher incomes;
• Come from homes that value education highly;
• Are likely to have books on the shelves and spend time reading together from an early age;
• Have meals together where the school day is discussed;
• Have parents who take an active role in homework;
• Come to school after a night sleeping with heat and after a morning with food in their bellies.

We don’t care about “whiteness” directly, but we care about the things bulleted above. If you’re African-American but the bullets above describe your household, then you should expect your children to achieve DCCAS scores of “white” students. That is, if those bullets describe your AA household, it is the “white” scores that are applicable to you, not the AA scores.

Since this is important, I’ll belabor the point: in DC, “white” is a clean proxy for high income. This does not mean all high income people are white. But what it does mean is that if you’re high income, whether white or AA, the best predictor of your child’s scores is given by the “white” average since this average represents only high income students.

Nonwhite: while “white” can be taken to mean high income, we cannot assume non-white is not high income. So, in scores broken down by non-white, please have caution in interpreting these numbers. They may be meaningless for drawing conclusions.


So, onto conclusions.

1. “White” scores at all three schools (Deal, Hardy, Wilson) are in the 90s. (That is, 90+ percent of white students score advanced or proficient on the DCCAS for reading and math.)
a. This has been the case since 2006 (at least).
2. “White” students perform as well at Hardy as they do at Wilson.
3. “White” scores are lower (though still in the 90s) at Hardy than they are at Deal. I haven’t checked, but I suspect the difference is statistically indistinguishable given the sample sizes. (Hardy’s “white” scores are sometimes higher than Deal’s “white” scores, by the way.)
4. There is a race gap at all three schools. Non-white scores are roughly 20 points lower than white scores at Deal and roughly 30 points lower than white scores at Hardy and Wilson.
5. Non-white scores are better at Hardy than Wilson. They are better still at Deal. Deal’s non-white scores are in the 70s, Hardy’s non-white scores are in the 60s, and Wilson’s non-white scores are in the 50s. I suspect but cannot confirm that non-whites perform better at Deal because non-whites at Deal feeders (like Shephard) are all described well by the bulleted points above. That is, many non-whites at Deal probably appear pretty similar to whites at Deal. Without a feeder school composed primarily of high income African-Americans, I doubt this is the case at Hardy.
6. Hardy is the least white of the three schools. Deal has never been below 28% white and is now consistently around 43% white. Wilson has never been below 21% white and is now 28% white. Hardy, on the other hand, hasn’t been above 11% since 2007 and has been 10% or below for the last 5 years.


7. The difference in overall scores between Hardy and Deal is overwhelmingly due to composition effects (i.e., demographics). White students score similarly in the two schools (consistently above 90% proficient or advanced). Non-whites score better at Deal than Hardy, but the main driver of the overall scores is the school composition. (See the final sentences of point 5 above for a guess why non-whites perform better at Deal than Hardy.)
8. If the percentages of white and non-white students at Hardy matched those of Deal, Hardy’s overall numbers would be about 10 percentage points higher for each category. For example, the percentage of students scoring proficient or advanced at Hardy in 2012-2013 would be 70% for reading and 77% for math instead of the observed percentages of 62% and 67%. This is without making any changes to performance (i.e., assuming no cohort or spillover effects). Deal’s numbers are still better (83% and 88%), but the gap is cut in half right there.
9. The above bullet understates the gains (in scores) to Hardy from having identical demographics to Deal. If Hardy’s percentages of white and non-white students matched those of Deal in 2008-09 or 2009-10, the schools would have the same overall DCCAS scores. There was no gap once one accounts for demographics. Since a gap does remain currently once demographics are controlled for, this suggests there are spillover effects benefiting Deal. It is reasonable to expect the same spillover effects to benefit Hardy as its demographics change.
10. Deal’s overall DCCAS scores have increased from the low 70s to the mid-to-high 80s over the sample period. This is driven primarily by increased percentages of “white” students – going from 29% to 43% over the timeframe. Using this as a guide, if Hardy should become more in-bounds (i.e., more “white”), a similar transformation would occur.

Main, actionable takeaways:

By the numbers, Hardy at present is at least as good as Wilson at present. If you care about DCCAS scores and you’re willing to send your child to Wilson, you should be willing to send your child to Hardy. “White” students perform as well at Hardy than at Wilson and non-white students perform better at Hardy than Wilson.

Hardy, today, is pretty similar to Deal of six or seven years ago. Deal has been stellar for the last four years or so. With increases in the IB percentage of students attending Hardy, the best forecast is that Hardy should become stellar just as quickly. (For “white” students, it is already stellar, much like Deal years ago. The overall numbers will catch up as the composition changes.)

If you believe that Murch is actually as good as Mann, then you should believe that Hardy is actually as good as Deal. The argument of the Murch supporters is that Murch’s lower overall numbers are explained by the composition of the Murch class. Looking at just “white” scores, Murch performs as well as Mann. An absolutely identical argument can be made for Hardy and Deal.

So, what should you expect for your child? If the bullet points at the beginning of this post describe your household – regardless of race – the white scores are the best predictor for your child. If you’re IB to Hardy, look at the white scores. If you’re IB to, say, Brent or Ross, but OOB to Hardy, look at the white scores. If you’re OOB at a Hardy feeder, chances are the bullet points describe your household. This means, look at the white scores to evaluate how your child will likely do at Hardy. It doesn’t matter if you’re AA or white, the relevant numbers are the white scores if those bullets apply to you.