QB social science: School District Income v. HSNCT Placement

[Matt Weiner]

Correlation...to be determined?



Credit to Evan Silberman, Christian Carter, and an army of data gatherers for constructing this graph.

[Down and out in Quintana Roo]

I'd like to see this broken down by county instead.

But that's very interesting.

[JackGlerum]

Nice work.

[grashid]

I was wondering. The closest to one is near the 50,000 range, but obviously 1 is TJ, and I thought either STPickrell or Byko gave data comparing Dorman's median income to TJ's and TJ's was a bit higher than 50,000.

[BuzzerZen]

Caveats about this chart:
1. Median income data is from the zip code the school resides in. This is a reasonable metric, but less accurate than free and reduced lunch percentage, for gauging the income bracket of a school.
2. The chart in Matt's post includes private and magnet schools, for which zip code-based data is unlikely to be representative.
3. SCIENCE!

[Down and out in Quintana Roo]

^^

Exactly why county-data would make a little more sense.

But the free/reduced lunch data would be the best of all.

[closesesame]

I will say that for "magnets" like TJ for example, since we draw from so many different neighborhoods, the median income of the school demographics is considerably different from the demographics of the neighborhood. Median family income in 22312 is probably $60000 or so, whereas TJ's median household income is easily towards $100000. In fact it is this lack of economic diversity that has driven the Fairfax County School Board to implement all sorts of policies and programs aimed at increasing minority attendance at TJ. So, like Evan said, take data for magnets with a grain of salt.

[Sir Thopas]

I will say that for "magnets" like TJ for example, since we draw from so many different neighborhoods, the median income of the school demographics is considerably different from the demographics of the neighborhood. Median family income in 22312 is probably $60000 or so, whereas TJ's median household income is easily towards $100000. In fact it is this lack of economic diversity that has driven the Fairfax County School Board to implement all sorts of policies and programs aimed at increasing minority attendance at TJ. So, like Evan said, take data for magnets with a grain of salt.

On the other hand, Hunter is skewed the other way probably because it's in a rich neighborhood, but attracts kids from across New York City.

[First Chairman]

Nice work! I am not sure that school district income is necessarily the right metric, especially when you look at the magnet schools. However, I presume an easier metric would be placement vs. games played during the year, especially games played vs. teams who also played at the HSNCT or NSC-qualifying teams. That may make things a bit more definitive with the hypothesis: you actually get better at quizbowl at HSNCT if you have enough money to play games.

So another challenge, which may wind up with the same results: school district income as measured by budget per student in high school vs. AI Bykowski rating. The other thing is to find out a control confirmation that school district income DOES affect, such as graduation rates from a high school so I know that you are using a valid metric that could correlate to a high school's commitment to its qb team.

But I will say, that's a nice set of data to put up at least for the time being. It's not perfect, but it's a start.

[NoahMinkCHS]

Wow, I'm actually really excited somebody did this. Thanks for compiling this data. That said...

Caveats about this chart: ...
2. The chart in Matt's post includes private and magnet schools, for which zip code-based data is unlikely to be representative.

Umm... that's a bit of a problem, don't you think? Even accepting ZIP code income data as a reasonable, but flawed, metric, leaving non-attendance zone schools in makes the data so noisy as to be completely meaningless. Can we see the graph without magnet and private schools? Then "no correlation" might mean something; as it is, it's just an assertion backed up by an irrelevant graph...

[BuzzerZen]

Wow, I'm actually really excited somebody did this. Thanks for compiling this data. That said...

Caveats about this chart: ...
2. The chart in Matt's post includes private and magnet schools, for which zip code-based data is unlikely to be representative.

Umm... that's a bit of a problem, don't you think? Even accepting ZIP code income data as a reasonable, but flawed, metric, leaving non-attendance zone schools in makes the data so noisy as to be completely meaningless. Can we see the graph without magnet and private schools? Then "no correlation" might mean something; as it is, it's just an assertion backed up by an irrelevant graph...

In fact you will be able to, but this is ongoing, and we all sort of lost momentum on this. We'll presumably continue collecting data and update this thread.

[Stained Diviner]

Is the HSNCT ranking based on the first day results only?

[STPickrell]

I was wondering. The closest to one is near the 50,000 range, but obviously 1 is TJ, and I thought either STPickrell or Byko gave data comparing Dorman's median income to TJ's and TJ's was a bit higher than 50,000.

I was the one who provided data comparing Dorman's median income to the TJ attendance area (Fauquier, Fairfax, Arlington, Loudoun, and Prince William Counties, and the Cities of Fairfax and Falls Church.)

TJ's physical zip code is in a middle class part of Fairfax County. Yet, I must agree with those who say zip code data is more or less irrelevant for private and magnet schools. I look forward to seeing the graph stripped of private and magnet schools.

As an aside: If I recall correctly, fewer than 1% of its students are on free/reduced lunch meals. (The highest in Northern VA is Mount Vernon). However this wealth doesn't seem to translate into gold-plated faucets. Only New Trier gets to have those.

Another monkey wrench -- what to make of schools like Princess Anne or Montgomery Blair, which are 80-90% general enrollment but have a magnet program of some sort?

I think Matt's main point -- that coaching and team dedication are infinitely more important than being rich -- remains true.

[Down and out in Quintana Roo]

Nobody wants to talk about it, but i would also love to see data related to minority representation in schools as well, particularly Hispanic and African American enrollment.

[First Chairman]

[Moved to Theory where I think it belongs better than the "my school is better than yours" section.

I'm not sure that no one wants to talk about it, but I don't know if anyone wants to address the socioeconomic disparities of school funding. To an extent there is a limited association with school zip code and funding level. But I will agree that proportion of historically underrepresented in the student body or even general "performance metrics" used to assess so-called schools of excellence would be of interest. Of course, there are flaws in those metrics too.

Again, I know the educators on this forum know of the metrics better than the college kids do (unless you study in those dreaded Colleges of Education as described on this board).

[cvdwightw]

The biggest problem with relating all of these metrics to quizbowl is that we are dealing with a ridiculously small sample size. Unless you play in the NAQT Small School division, your school's A team likely makes up less than 1% of your school's population. If you go to a large school, your entire program may not make up 1% of the school's population. Saying "School A has 80% of kids on free/reduced lunch" means nothing if the 1% of the school in quizbowl draws almost entirely from the kids who are not on free/reduced lunch. Saying "this school is failing No Child Left Behind" means even less (yes, I realize that means "less than nothing"), since there are about a gazillion different parts to "passing", no school gets any additional funding to work towards those standards, and schools can fail due to no fault of their own (OMG we have only 5% minority students and 80% of them didn't show up to take the test so we're going to fail the "minority students taking/passing test" quotas). Some of these metrics might mean something at a school with a large (relative to school population) quizbowl program that draws somewhat representatively from the entire school population. But to the best of my knowledge, neither one of these occurs at any high school. Therefore we are comparing averages against a very small group of students who are much more likely to be "above average" or not in the group we're comparing schools by, and I would be extremely hesitant to accept any "conclusions" drawn from our "data".

[Sir Thopas]

Dwight, this thread was made in response to claims that rich schools are inherently better at QB.

[dschafer]

Is the source for this graph available? I'd love to see the actual numbers.

[rchschem]

The obvious solution is to average the ZIP codes for the regions that feed students to a school.

Weighted averages, of course. Let me know if I can help anymore.

[dtaylor4]

The obvious solution is to average the ZIP codes for the regions that feed students to a school.

Weighted averages, of course. Let me know if I can help anymore.

Unless such data is published already (which I doubt), one would have to find the addresses of every student at a given school. Not only would this likely involve some form of privacy invasion, the time that it would take to do this is not worth it.

[Mechanical Beasts]

The obvious solution is to average the ZIP codes for the regions that feed students to a school.

Weighted averages, of course. Let me know if I can help anymore.

Unless such data is published already (which I doubt), one would have to find the addresses of every student at a given school. Not only would this likely involve some form of privacy invasion, the time that it would take to do this is not worth it.

Not to mention that would be a little useless, because if you're able to do that, then you should just ignore all the zip codes of students who don't play quiz bowl.

Not to mention the fact that now you're incorporating a BUNCH of zip codes that are probably not too well representative of the team's/school's/students' resources, instead of just a few.

[Stained Diviner]

Some of these methodological criticisms would be valid if you were trying to use this data for some sophisticated purpose. (If so, the biggest problem would be that you only got through about two-thirds of the teams. It also might be better to look at the top teams in Byko's system, since the HSNCT teams self-select to a large degree and money is a big factor in deciding which teams go and which teams don't.)

However, the overall point is that you can have a good quizbowl team if your district isn't full of rich kids. This data is good enough to support that conclusion if not prove it. Another way to support it would be to come up with a list of 20-30 top programs whose families generally are not wealthy and give a brief description of their demographics, especially their economic demographics.

Being rich can help though. My team is represent by the highest dot on the graph, and we were the only team to power a tossup with the answer, "Penn--he's married to my dad's cousin."

[rchschem]

The obvious solution is to average the ZIP codes for the regions that feed students to a school.

Weighted averages, of course. Let me know if I can help anymore.

Unless such data is published already (which I doubt), one would have to find the addresses of every student at a given school. Not only would this likely involve some form of privacy invasion, the time that it would take to do this is not worth it.

Not to mention that would be a little useless, because if you're able to do that, then you should just ignore all the zip codes of students who don't play quiz bowl.

Sorry I forgot the icon. I couldn't imagine how my suggestion could have been taken seriously.

[Mechanical Beasts]

Sorry I forgot the icon. I couldn't imagine how my suggestion could have been taken seriously.

See, I tend to go overboard on projects. A couple years ago, I would have gotten halfway done and then realized that my data was meaningless.

[jrbarry]

I have always wondered if the educational level of a school's attendance boundary might be instructive or at least interesting in such considerations. Once again, this wouldn't mean much for magnet schools or private schools.

[Angry Babies in Love]

Another monkey wrench -- what to make of schools like Princess Anne or Montgomery Blair, which are 80-90% general enrollment but have a magnet program of some sort?

RM is also that way (to a lesser extent), but on our A-team last year, I believe 3 out of the four members were "magnetized", if you will. So even though a school can be 80-90% original jurisdiction, the quizbowl is most likely to be comprised of magnet people. The members of a QB team are not a good representation of magnet pop. %.