Data analysis on the go

A recent Fast Company article describes Tableau’s latest app, Project Elastic, as a potential game-changer in mobile data analysis. The app would allow iPad users to move from an Excel file in Mail.app to an interactive chart with a few quick taps. Filters can be applied/removed with swipes, zoom levels are adjusted by pinch/spread gestures, and charts can be shared via email.

It’s a cleverly designed interface, but Elastic suffers from a critical problem: it assumes the user is accessing clean datasets.

Contrary to popular belief, the real work of a data scientist isn’t sexy. A large chunk, if not a majority, of a data scientist’s time is typically spent cleaning messy data. By encouraging users to jump straight from an email attachment to data exploration, Elastic can lead to garbage-in, garbage-out analyses. This risk is only compounded by the targeting of the app to casual analysts.

Elastic is an admirable effort by Tableau to bring data analysis to the iPad, but it will encourage a flawed approach to data analysis. To paraphrase Jeff Leek, good data science may look easy, but so does bad data science. Elastic looks cool, but the appealing interface is a mask for an insufficiently rigorous approach to data analysis.

I have serious doubts that quality data analysis can be performed from a tablet or mobile platform.

When asked about the potential for the iPad replacing the PC, Steve Jobs compared it to the replacement of trucks by cars:

“When we were an agrarian nation, all cars were trucks, because that’s what you needed on the farm. But as vehicles started to be used in the urban centers, cars got more popular. Innovations like automatic transmission and power steering and things that you didn’t care about in a truck as much started to become paramount in cars. … PCs are going to be like trucks. They’re still going to be around, they’re still going to have a lot of value, but they’re going to be used by one out of X people. … I think that we’re embarked on that.

Mobile computing is replacing PC’s for millions of people around the globe for tasks like email, photography, personal finance, and gaming, but there are still many jobs that ought not be done on a mobile device. Attempting to move data analysis to the iPad makes about as much sense as trying to move a piano with a Prius - just because it’s possible, doesn’t mean it should be done.

Bottom line: if someone emails you a spreadsheet, you should probably wait to get to a PC before attempting to dig in to the data.


Throw out CCJEF

A few months ago, it appeared that the CCJEF v. Rell trial was going to begin at the start of 2015. The CCJEF coalition1 claims that the state isn’t paying enough money to adequately educate Connecticut students. They’re hoping for courts to compel the legislature to spend an additional $2 billion on schools - essentially doubling the amount the state currently spends on their main grant to districts, ECS.2

All of this changed in November, when the Attorney General released emails showing that CCJEF’s leader, Dianne Kaplan deVries, asking several people to delete their emails to avoid subpoena. As a result, the state is now asking the courts to remove the coalition as a plaintiff.

I’m not by any means a legal scholar, so I can’t comment on the strength of the state’s argument in this particular instance, but I do hope the CCJEF is dismissed as soon as possible.

As long as the specter of a CCJEF-related court mandate looms over the legislature, there is little chance they will take up meaningful school finance reform. I can understand the predicament legislators are in at the moment - why should they stick their necks out on a political hot-button like school finance when a court ruling could make their work irrelevant?

Dismissing the CCJEF case would make it much more possible to enact meaningful school finace reform. Let’s hope it happens sooner rather than later.


  1. CCJEF includes the two teachers unions in Connecticut (CEA and AFT-CT), the Connecticut Association of School Boards (CABE), the Connecticut Association of Public School Superintendents (CAPSS), several towns, and a handful of students and parents. 

  2. Current projections put ECS spending in FY16-FY18 at $2.04 billion. 


Outdated funding systems are limiting options for families in Connecticut.

The Growth of School Choice in Connecticut, 1995-2015

For years, families in Connecticut demanded more public school options for their children. Over the last two decades, state funding for magnet and charter schools grew significantly.

choiceGrants

Even with this level of growth, thousands still end up unable to attend magnet schools or charter schools due to limited capacity. Wouldn’t it seem reasonable that lawmakers would work to expand the number of public school options available to families?

Unfortunately, Connecticut may be heading in the opposite direction. Some lawmakers are considering putting a moratorium on any further expansion of magnet schools and charter schools.


Choice funding in Connecticut: too big to exist as line-items?

Simply put, the current system of funding school choice is expensive and downright Byzantine. A student can have their public education funded at least 10 different ways based on the type of school they attend.1 As the chart below indicates, this complicated system of funding is now equivalent to more than 15% of the main grant to school districts, the Education Cost Sharing grant (ECS).

choiceVsECS

Freezing funding to schools of choice would be an easy way for legislators to avoid answering a simple question: why should more than 50,000 students have their public schools funded in different and inequitable methods?

What’s the solution?

It’s going to be a lean budget year with many tough short-term choices, but lawmakers need to think seriously about the future of our school finance system. I hope they keep these principles in mind:

  • All public school students should have their public education funded by a combination of the state and by their town of residence.
  • The ratio of state/town funding should be determined by the relative wealth of the town.
  • A student’s total level of funding should be determined by their learning needs.
  • The full funding for a student should follow them to the public school of their choice.

A funding system guided by these principles would benefit the state by focusing the school finance system on town wealth and student need instead of the label a public school carries. It would benefit towns by giving them a predictable level of costs - no matter what schools families choose, towns would contribute the same level of funding to their education. Most importantly, it would provide families with more choices for their children.


  1. As noted in a previous post, this was detailed in the ECS Task Force’s Final Report 


A Thought Experiment

What would happen if we actually followed the ECS formula?

There are two significant flaws in Connecticut’s current school funding formula, known as ECS (Education Cost Sharing):

  1. It does not reflect the actual needs of towns and students.
  2. It does not treat the thousands of students accessing school choice equitably.

I plan on addressing those concerns in subsequent posts, but for a moment, let’s take ECS at face value and assume that the underpinnings of the formula are correct. If political processes had not brought Connecticut off of the ECS formula and we distributed current funding according to the formula, how might that change how dollars flow to towns?

Right now, the ECS formula is somewhere around $650 million under-funded. That is to say, the legislature and governor have not found a way to follow the ECS formula as it was originally intended to operate. In lieu of actually following the formula, the money that flows from the state to towns to school districts has been manipulated by politicians each legislative session since the ECS formula was adopted.

What would happen if our elected representatives followed the distributive principles of the ECS formula, so that if the state could only contribute 80% of what the formula dictates, each town would receive 80% of what the ECS formula dictates they should receive?1 Let’s figure it out!


In fiscal year 2015 (FY15), ECS was about 75% funded, meaning that the state is paying about three quarters to the dollar of what the ECS formula actually dictates.

You’ll be shocked, shocked to learn that some towns are earning more than 75% of their ECS entitlement, while other earn less!

Which towns are doing better/worse than they should be doing if current dollars were distributed based on the ECS formula?

absdiff

Which towns are benefiting from the current flow of ECS funding?

TownAdj. FY15 Over-funding
Hartford*+$21,489,136
Bridgeport*+$19,716,791
New Haven*+$18,926,723
Naugatuck*+$4,757,282
Meriden*+$4,352,828
Groton+$4,183,143
Colchester+$3,355,802
Wolcott+$3,339,954
Norwich*+$3,217,643
Ledyard+$2,969,683

Unsurprisingly, the biggest beneficiaries of the current distribution of ECS dollars tend to be Alliance Districts, a collection of 30 low-performing districts the Malloy administration placed a priority on supporting with additional funding (identified in the table with a *).

Twelve towns are under-funded by at least $5 million according to the current ECS formula. West Hartford falls $20 million short. Danbury is $17 million short and Milford is more than $11 million under-funded:

TownAdj. FY15 Under-funding
West Hartford- $20,029,204
Danbury*-$17,489,274
Milford-$11,180,390
Norwalk*-$9,681,090
Stratford-$8,680,338
Shelton-$6,910,435
Branford-$6,389,163
Wethersfield-$6,246,413
Middletown*-$5,983,252
Southbury-$5,557,434
Hamden*-$5,269,591
Stamford*-$5,263,781

It’s surprising to see five Alliance Districts on this list (again, identified with a *). The combined under-funding of these five districts (Danbury, Norwalk, Middletown, Hamden, and Stamford) is about $43 million, which is less than the $60 million over-funding of the big three cities (Bridgeport, Hartford, and New Haven).


Looking at raw dollar differences is helpful, but Connecticut is a state with a few large towns and many medium and small towns. Let’s compare what towns receive now as a percentage of what they ought to receive if we followed the ECS formula with current appropriations:

pctDiffECS

Now we’re able to see which towns are really benefitting from the current flow of ECS funding. There are 53 towns in Connecticut that receive about 32% more ECS funding than they ought to, given the current level of total ECS appropriations. This includes affluent towns like Greenwich, Darien, Fairfield, Madison, and Westport.

Which towns get the short end of the stick? Well, there are a dozen towns that receive less than half what they ought to receive under the ECS formula:

TownFY15 Funding as % of Adj. FY15 Funding
Branford23.0%
Woodbury30.7%
Farmington30.9%
Southbury32.1%
Orange33.3%
Shelton43.3%
North Haven44.5%
Trumbull44.5%
Middlebury45.3%
Rocky Hill45.8%
Westbrook46.6%
West Hartford47.6%

The place I call home, Branford, gets less than a quarter on the dollar of what we should be getting according to ECS with the current level of total funding distributed via the ECS grant.


Again, I don’t think the current ECS formula is correct,2 but this thought experiment highlights the key challenges of reforming Connecticut’s school finance system if we kept total spending flat:

  • The “big three” Alliance Districts are doing much better than other Alliance Districts. Connecticut’s big three urban cities (Bridgeport, Hartford, and New Haven) are doing very well under the current flow of ECS dollars, which is in stark contrast to other high-needs towns that are significantly under-funded (Danbury, Norwalk, Middletown, Hamden, and Stamford). In order for the big three to gain funding, there needs to be either an increase in overall funding or adjustments to the ECS formula to more heavily weight for poverty.
  • There are many (mostly affluent) towns that are receiving more than their fair share. Many towns have already hit the ceiling on what they should be getting under a fully-funded ECS formula. Any funding reform that gives more weight to towns with high poverty would mean reduced funding for these towns.
  • Suburbs are getting a raw deal. While there are some Alliance Districts that are relatively under-funded, the most disadvantaged towns are middle-class suburbs like Branford, Orange, and West Hartford. These are towns that should be pushing for immediate school finance reform, since they would stand to see the largest relative increase in funding, even with no new money added to the system.

Instead of waiting for a court to compel the legislature to change ECS, leaders of middle-class towns and Alliance Districts should be working to develop an school finance “grand bargain” that could be implemented without significant additional revenue. This would likely need to include schools of choice (magnets, charters, etc.) along with some sort of regulatory relief to affluent towns that would stand to lose funding.

Will any leaders take on this challenge?


  1. I’m using a set of ECS funding targets from a handout at the ECS Task Force, slightly edited to prevent any districts from receiving less ECS funding under a fully-funded ECS formula than they get right now. If/when more up-to-date figures become available, I’d be happy to update this post accordingly. All code used to generate charts/data in this post is available on my GitHub page

  2. I would prefer a formula that included more recent wealth/population data and more weight for towns with highly concentrated poverty. 


What’s the problem with school finance in Connecticut?

School finance in Connecticut is broken.

The state’s main vehicle of support to public schools, the Education Cost Sharing (ECS) grant, sends $2 billion to local districts each year with the goal of equalizing the ability of communities to fund public education. Unfortunately, years of legislative tinkering with the ECS formula left Connecticut with a Byzantine approach to funding schools that takes more than 16 pages to explain.

On top of this labyrinthine system, non-traditional public schools (including magnet, charter, vocational/technical, and agriscience schools) are funded under entirely separate mechanisms. In practice, this means that one student could have their education funded up to ten different ways1.

This approach makes the bulk of Connecticut’s education aid unresponsive to the changing needs of communities and families across our state.


In the wake of the largest tax increase in state history and with major financial challenges looming, it’s not likely that our state will simply be able to address this problem with a significant cash infusion. We need to make difficult choices to improve how Connecticut allocates scarce resources for public education.

In the state with the largest achievement gap in the nation, we need a better way to provide a high-quality public education to every child. I believe that a more equitable and responsive school finance system is an necessary mechanism to achieve that goal.

It won’t be easy changing the status quo in the land of steady habits, but I’m going to do my part to move the conversation in a productive direction. In this space, I plan on:

  • Detailing the challenges and competing priorities involved in school finance reform.
  • Proposing policy approaches that take into account the educational challenges and fiscal realities facing our state.
  • Transparent analyses of school finance and demographic data. When possible, all code and data will be available on my GitHub page.

Connecticut’s school finance system is broken. We need to fix it my making some tough policy choices.

Let’s get to work.


  1. The ten ways of funding public school students in Connecticut include the ECS formula and the nine streams of choice funding found on p. 9 of the ECS Task Force’s Final Report