Legend
	map
	hat
	act
	keel

ReformLab home

Introduction

As a computer nerd and future educator, I wanted to put together these skills and create a computer simulation of school reform. Of course, it has become ever more clear to me that school reform is so complex, a decent simulation, if possible at all, would take years of work. Also, it is hard enough for reformers to agree on qualitative, much less quantitative, assumptions .

The approach I have taken is to attempt to simulate just a few aspects of reform. On the opening screen there are a number of tabs. The assumptions page is where you display your keel as quantify your assumptions. The levers page is where the levers of reform are and hence where the acts happen; you as reformer may adjust these as you see fit. There is a graph tab, which attempts to show the result of the reform. This result is shown rather crudely. The graph shows an extremely typical Boston seventh grade math class measured by MCAS score, with lowest scores on the left and highest on the right. (I use the Boston averages for 2006 and some information about the Edwards Middle School.) Then there is a "Try it" button, which is when you make the reforms specified by the levers and see what happens to student performance according to your assumptions. After each "Try it" reform, you see another colored line representing the result of that reform .

Assumptions

This is where you get to specify the assumptions you want the program to operate under. There are severe limitations to the kinds of assumptions you can make, however; that represents my keel less than the limits of what I can do in this amount of time. They are set to my wild guesses about where they should be.

Low performer boost

How much better the lower third of students will perform when going from heterogenous grouping to 100% homogeneous grouping (whatever that means exactly). Can be negative if this third will do worse under ability grouping. The assumption is that this is linear so for example going from 0% to 50% ability grouping would have half the effect.

Mid performer boost

Same for middle third.

High performer boost

Same for high third.

Class size boost

The performance boost per student of having 5 fewer students per class.

Salary boost

The performance boost per student of increasing teacher salaries by $5000. This can be from increased teacher satisfaction, Attracting better candidates, etc..

Merit pay boost

The performance boost per student of giving the principal discretion over an average $5000 merit pay per teacher. This would presumably come from motivating good teachers to stay, better teachers to join, and bad teachers to leave. Make this negative if you believe this will lower morale or lower student performance some other way.

Levers

Ability grouping

This lever controls how much ability grouping we do in this school. It defaults to 10%, which is my wild guess based on resource rooms at the Edwards. The exact meaning of this number is up to you; it is tied to the low/mid/high performance boosts of the assumptions . Depending on assumptions, ability grouping is either a solution to a problem (if we assume that this rising tide lifts all boats) or a management strategy for a dilemma (if we assume that it benefits some and hurts some).

Class size

This is the average class size. It defaults to around 20, which seems to be about what the Edwards school averages. (The school's 2006 report works out to 451/37 or about 12, but I don't believe it!.) The dilemma here is mainly between class size and the budget; watch the budget display as you adjust this.

Salary

This is the average non-merit-based teacher salary. The default, $55,000, is eyeballed from the middle of the Boston Teachers' Union contract. Generally the "only" thing keeping this from going higher is the budget.

Merit Pay

This is the average merit pay per teacher. This defaults to zero per the BTU contract. In this program you are not subject to this contract, but of course in reality the union contract would have to be changed to allow this, which could have all sorts of other consequences . A teacher/reformer would have to tread carefully here . If you balance a merit pay increase with a salary decrease it doesn't cost anything, and depending on assumptions, that may make things better or worse .

Budget

The budget in this program is not a lever but rather a consequence of some of the other choices. My formula simply assumes that the budget is twice the amount spent on teachers' salaries (including other salaries, facilities, etc.) . In real life of course it often goes the other way, when politicians present the school committee with a budget and then they decide what to do.

Limitations and Potential Enhancements

This software is so limited it's hard to know where to begin. Maybe I should begin at the end, which is that there are many possible ways to evaluate an education, depending on one's assumptions, including purpose . The fact that a criterion is measurable does not justify its use as a measure of educational success. Nevertheless the state and NCLB put a lot of weight on this test, and the fact that each test is measured by a single number made my life easier.

The choice of levers was somewhat arbitrary (though influenced by what could be quantified) and only a tiny subset of the levers we have discussed in class. Also the brute simplicity of each lever is generally not warranted. For example, does 50% ability grouping mean that half of the kids are ability grouped, or all the kids are ability grouped half the time, or all the kids are ability grouped but into half as many groups as are possible? I have chosen fairly interdependent levers . For example, if there were a professional development lever, cranking it up would have to affect other levers, e.g increasing class size because of the reduced time in front of students for teachers.

Certainly the assumptions are limited both in the number of assumptions and in their stark linearity. For a simple example, changing a class from 30 to 25 students may not have much effect, but changing a class from 10 to 5 might have a fairly dramatic one. My program doesn't show that.

It might be nice to have a choice of hats with different levers for each hat ; for example principals may have some control over ability grouping but probably none over class size. The assumptions page might influence which hats go with which levers. For example, depending on one's view of civic action, someone wearing a parent hat may be viewed as controlling no levers or having some control over many. On a more prosaic lever, depending on the district's administrative structure, a principal may or may not have power over salaries and merit pay.

Conclusion

Because of all of the program's simplifications, it is not clear if the program tells us much. And because the user can make assumptions you can pretty much make it tell you whatever you want. For example, I can make optimistic assumptions about ability grouping and get a nice improvement from that reform. An opponent of ability grouping could make different assumptions and watch a disaster unfold.

Nevertheless, I usually find that when I express an idea in a computer program I understand it better. In this case, the appreciation I have already gotten of the complexity of reform was made particularly vivid.

References

Boston Public Schools. (2006). Clarence R edwards middle school SY 2005-2006 from http://www.boston.k12.ma.us/schools/RC522.pdf

Boston Teachers' Union. (2006). BTU/BPS contract (2006 - 2010) from http://btu.org/leftnavbar/downloadforms.html

Bradshaw, L. (1996). Java 2D graph package version 2.4., 2008, from http://www.sci.usq.edu.au/staff/leighb/graph/

Massachusetts Department of Education. (2006). 2006 MCAS technical report from http://www.mcasservicecenter.com/documents/MA/12-18%20NoDraftFooter%202006%20Tech%20Rpt.pdf