Better Off Ed

This makes me so angry. It’s a NYTimes story of a principal tasked with turning around a failing Bronx public school, now under investigation by the NY Department of Education. Santiago Taveras’s crime? Breaking rules in the best interest of kids. According to the Times (direct excerpts):

1. Taveras had changed certain courses on students’ transcripts from required credits to elective credits in cases where administrators had mistakenly enrolled the students in the same course twice. The [DOE] report cited Mr. Taveras as saying that he did not want the students to be punished for those mistakes. 

2. Mr. Taveras had in four cases changed students’ grades, from failing to passing, in violation of department policies. In one case, the student said she had a medical condition that had forced her to miss many days of school. She said that her global history teacher had agreed that if she completed a supplemental work packet and passed the Regents exam, she would get a passing grade. The girl did the work and passed the exam, but the teacher gave her a grade of “No Show.” The girl spoke to Mr. Taveras about it over the summer. He emailed the teacher, but did not hear back from her and changed the grade himself.

3. Mr. Taveras had told an assistant principal that he was going to give him a poor rating and recommend that he be terminated unless he found a job somewhere else. Mr. Taveras did give him a low rating. Then, after the assistant principal found a job at another school, he changed it.

Doesn’t he sound horrible? Seriously, to what citizen do these sound like crimes? What parent would be upset that the school took responsibility for its own errors and did not penalize the student? What parent is upset that a student with a medical condition, who did supplemental work to pass a class, actually passed that class? And what parent is mad that the principal took action to get a bad teacher out of the school as soon as possible instead of following the ill-constructed rules to the detriment of the students?

Unfortunately, the people at the NY DOE who classify Taveras’s case as “very serious” are not acting like parents. They’re acting like Pharisees. 

The Pharisees were the religious elite who criticized Jesus’s every move. They pointed out that Jesus broke lots of the rules that they had so perfectly memorized and followed (which is the reason they got to be in power). Jesus healed on the Sabbath (the day reserved for resting and worshipping the Lord) and they admonished him for it. Jesus responded pretty simply, 

“What kind of action suits the Sabbath best? Doing good or doing evil? Helping people or leaving them helpless?” Mark 3:4 MSG

The Bible continues, “No one said a word.” Well, duh.

The God the Pharisees supposedly followed was a God that commanded his people to love God and love their neighbors above all else (you would figure loving neighbors includes healing them when you have the power to do so). Jesus showed that when obeying a ‘rule’ hinders that ultimate command, the ultimate command takes priority. It’s just common sense.

As Jesus says to the Pharisees later, 

“Why do you use your rules to play fast and loose with God’s commands? God clearly says ‘Respect your father and mother’... But you weasel around that by saying, ‘Whoever wants to, can say to a father and mother, What I owed to you I’ve given to God’ That can hardly be called respecting a parent. You cancel God’s commands by your rules.” Matthew 15:3-6 MSG

The ultimate command of our states’ education departments should be: educate kids well. Sure, it makes sense to have some rules, but when following those rules is not in the best interest of children’s educations, the rules should not be followed to the letter. And certainly, principals that break rules to follow commands should not be punished.

Because what we really don’t want is an educational system with nothing but Pharisees: rule followers that have gotten so focused on the rules that they lose sight of the bigger picture: our children’s educations. Our principals and school leaders need to put students first and our departments of education should support them in doing so.

Anything else is just silly.

Every once in a while, Goldman Sachs sends out an “Alumni in the News” newsletter to all its employees and alumni. As an investment banking analyst at Goldman, I remember always reading that newsletter and thinking one day maybe some analyst could be reading about me and my work in education. Well, hey, that day is today, and I think it’s a great opportunity to talk about how much I love Goldman Sachs.

Really, I do. Working at Goldman as a healthcare investment banker was one of the best decisions I’ve ever made. When most people hear my story of “investment-banker turned inner-city teacher,” they always assume it follows the old “I was making lots of money but my soul was dead, because everyone around me was a greedy asshole, so I had to leave and pursue my passion” storyline.

That’s not my story. I came to Goldman Sachs to pursue my passion of education. I came to Goldman, because I knew it was the best two-year education I could receive - and not just with respect to finance. Goldman Sachs taught me how to be a professional; it taught me leadership; it taught me how to deal with my mistakes, deal with crises and how to just plain deal. And of course, it taught me how to effectively communicate an organization’s finances – an invaluable skill I’m now utilizing to help schools better optimize the $600 billion spent on education each year.

I often explain that I took my summer internship there because “at that time at Wharton, being an investment banker at Goldman Sachs was the coolest thing you could do… I hope it’s still not that way.” While I know it’s not the “coolest” first job anymore, I do think investment banking at Goldman may be the best first job. Here’s why:

You learn how to be a professional:

I don’t mean you learn how to dress formally and write a formal email; actually a bit of the opposite. Our best VPs and MDs (the ones most often interacting directly with the clients) were not the most uptight and formal; they were the ones that were able to be humorous and casual, but always commanding of respect. Part of being professional is making the opposite party feel both comfortable and respected. Being warm is okay. Hugs can even be okay in some situations. Basically, your tone and behavior must adapt to a given situation and a given client. Because in banking you’re exposed to many different situations with all types of clients, you quickly learn the range of acceptable behaviors and when to use them. While some people are more natural at this than others, this is not simply about being a people person; it’s a skill that’s learned over time with experience. I’ve seen summer interns transform in their professionalism over the course of their 10-week internship, and I’ve seen associates gain confidence and effectiveness with clients by emulating their VPs over time. I myself went from being petrified to speak in client meetings to comfortably calling up CFOs or other senior executives to work through the model or give them a deal update one on one.

You learn what good leadership is:

Is everyone at Goldman an awesome leader? Of course not. My favorite thing about investment banking is that you have lots of bosses. You have multiple teams, typically each with an associate, a VP, an MD, a partner, and some combination of those roles in different groups depending on the deal (for example equity capital markets if you’re working an IPO). You end up working with about a dozen associates, a dozen VPs, a dozen MDs, etc. In a lot of first jobs, you may be stuck with one boss and she may be great or he may be horrible. In banking, you have exposure to so many different personalities at the same role/rank/relationship to you, so you can easily compare. It’s incredible. Analysts quickly learn what types of behaviors make someone awesome to work with and what types of behaviors make someone awful to work with. And analysts compare notes with other analysts, processing and analyzing what makes a good associate, VP, MD, etc. constantly.

What did I learn about leadership? The best leaders were the ones that were servants. In fact, in all my subsequent nonprofit work since Goldman, I still have not experienced such servant leadership as I did at Goldman Sachs. These were the people who were supposed to care only about money, but never have I felt more cared for and protected by my bosses than at Goldman.

There were a few bosses (at every level), who acted like your superior, who didn’t mind having you wait in their office as they sat thinking about what work needed to be done for the client; who didn’t mind offering up extra analysis or a tighter deadline no one asked for, because they wanted to win points with their boss or client (completely not caring that it would have the junior team up all night); who would throw the junior team under the bus for a mistake. But those people didn’t last long. In order to survive long at Goldman, you have a to be a good leader, which means you have to be a servant leader. Servant leaders were those who respected their subordinates’ time by only calling meetings after they themselves had prepared decisive plans of action; who made the effort to understand the level of difficulty of each task they assigned, so as to minimize unnecessarily cumbersome work for the junior team; who “protected their team” by taking responsibility for any mistake their junior teams made.

As nerve-wracking as the role can be at times (you’re dealing with billion dollar deals and companies), I always knew I was protected. I always knew that if the client saw some incorrect number that was a result of my stupid mistake, someone more senior would rightly take responsibility for it.  And it didn’t make me more careless. Actually, it had the opposite effect. Because I knew my bosses were protecting me, I wanted to protect them by avoiding those mistakes at all costs. But mistakes were inevitable (which brings me to my next point).

You learn how to deal with your small mistakes, your mistakes that become crises, and other people’s mistakes:

As an analyst you are producing so much work every given day; you may have a draft of a pitch deck for one client, a sensitivity analysis for another, a new merger scenario for a deal, an email updating the team on the company’s recent earnings call. In any given client presentation, there may be 50 different charts and 50 different footnotes that all need to be right. Guess what? You cannot possibly get it all right, all the time. Often, you are running on two or three hours of sleep and a misplaced comma in a number in a footnote gets completely overlooked. Often, the mistake you make is really dumb and there’s no excuse for it. Instead of freaking out and trying to explain away a mistake or justify it or blame someone or something else, you learn to just say, “My bad; I’ll fix it right away.” You learn to not let it affect you emotionally, because it’s not productive, and you learn how to quickly fix and learn from your mistakes instead of dwell on them. Thankfully, the little mistakes almost always get caught by someone on your team before it gets to the client. But when you make a mistake that gets through to the client, you learn an even more valuable lesson about trust. Some mistakes may impact the entire deal and the client needs to know about them right away. Some mistakes may never get revisited, and so it’s not worth alerting a client and risk having them doubt you moving forward. Knowing when (and how) to alert a client to a mistake is a valuable skill.

I remember one deal in which I realized at the last hour that I had made a simple but significant mistake; it impacted the pricing of the deal. It had to do with a number that we had been using for nine months, that the CFO and the rest of our team had signed off on, but I had somehow just noticed was incorrect. I had a decision to make: was I going to tell our team and risk looking like an idiot or should I just ignore it? After all, no one else had caught it, so the deal could just go on as is without anyone ever noticing it. I was mortified, but I decided to do the right thing; I told my associate, who told our MD, who told our client.

And I wasn’t punished for it. Our MD was not exactly happy but he didn’t waste time with emotions. He went straight to a plan of action to make it right. He didn’t yell and scream “how could this happen?!” He didn’t patronize me as I corrected the mistake with lots of condescending questions making sure I was doing it right this time. He understood it was a mistake, that because he was the boss, he had ultimate responsibility for it, so he didn’t demean me. He didn’t focus on the mistake itself; he focused our team around fixing it. We told the client, the client understood, we fixed it, the deal got done, and in the end, everyone was happy. I didn’t get a big lecture about never making that mistake again. My MD knew I would never make that mistake again. He still trusted me, and still wanted to work with me on other deals. In turn, I worked harder and more joyfully for him than just about anyone. Joyfully? That’s my last point.

You learn how to plain deal:

All jobs are hard in their own ways. I’m not so presumptuous to think the long hours and sleepless nights means that investment banking is the hardest job out there (I was, after all, a charter school teacher). But investment banking gives you lots of moments where you are blind-sided with the unexpected, and all you want to do is kick and scream, but you have to stay positive. Your biggest client calls you personally on a Saturday with work they need as soon as possible and you have to miss your best friend’s birthday party (that happened to me). A deal is about to close and you have to cancel a vacation (that happened to my friend). You finally leave the office at a decent hour and your boss emails at 10:00pm with 5 hours of work to do before the next day (that happens to everyone). When you average three hours of sleep a night, you’re moody. But you learn there’s no sense in dwelling on the negative. You learn how to deal. You learn how to make jokes about whatever you can, and you learn how to smile when you want to cry.

Doesn’t it sound lovely? So many people say, “I could never do that. I could never work those hours; I could never deal with that much work or such high expectations or pressure.” To which I always say: no one can…until you get thrown into it and then you simply learn how to deal. Which is why investment banking is one the best first jobs you can get. 

Combine the nature of investment banking with the excellent leadership at Goldman, and I have nothing but fond memories of my two years there. Ok, that may be a stretch, but the good far outweigh the bad, and I would do it all over in a heartbeat. I am so very grateful for the people I worked with and learned so much from, and I still visit them whenever I’m in NYC.

In fact, last week in NYC, I had 15 minutes between meetings and squeezed in a coffee with my old assistant Mona (pictured below). Mona became my second mother. She answered my phone, scheduled my meetings and flights, but most importantly told me when I needed to call my actual mother, be nicer to my boyfriend, or brush my hair. Truly, Goldman set me up with all the life skills I would ever need. Best first job ever.

The recent signing of the No Child Left Behind rewrite had people (including me) applauding that it reduced the federal government’s role in education. Others, like the very thoughtful Citizen Stewart, bemoaned that it has decreased overall accountability which will hurt our most struggling students the most. Stewart has valid points, and I completely agree that accountability in education is crucial.

But the effective accountability created by No Child Left Behind was not through the federal government’s mandated interventions in failing schools; it was through the mandated standardized tests, which were required to be made publicly available. The latter remains in the rewrite. The former is scrapped, because it wasn’t particular effective.

See, the federal government can (and did) force districts and schools to reorganize when they’re failing, but they cannot (and did not) force districts or schools to do that well. 

No Child Left Behind set the goal that 100% of students would be proficient in reading and math by 2014. That goal was obviously not achieved. So what did the federal government do? Well, the federal government did not pull all its money from schools - that would have inspired outcry about districts unable to fund their lunch programs for poor kids, massive teacher layoffs that hurt students, etc, etc. 

What did the federal government do to hold states accountable for missing this goal by a landslide? Not much.

The most the federal government can hold states accountable to are certain actions, like “once a school has five years of lagging test scores, you must fire the principal and half the staff.” If a state isn’t competent enough to recognize a failing school and improve it until the federal government steps in, why do we think the state will manage the process of hiring a new principal and teachers well? States can just start another bad school, wait a few years for the federal government to tell them to do something else, and then do that. Rinse and repeat, all the while states still get federal funding. In order to receive federal funding, states just have to try to educate students, not actually produce educated students.

Actual accountability to results would require that the federal government reduce funding to states based on their test scores, but that’s not happening, because the federal government (and I would argue the general public) doesn’t have the stomach to reduce funding for poor students given any situation.

So if the federal government is going to keep funding states without true accountability, we don’t need the pretense of an 1,061 page bill describing what states must do in order to receive federal funds in the name of faux accountability. 

We need 3 pages that say this: 

Page One: Here’s how much states get per poor student (or Special Ed student, or English Language Learner, etc). 

Page Two: States must distribute funds to districts on a per pupil basis. Districts must distribute funds to schools on a per pupil basis.

Page Three: Districts must test their students and make their student achievement data (subgroups included) publicly available and accessible. They must also make financial data publicly available and accessible.

It’s crazy to give money away without ensuring it’s going to be spent effectively, but I would argue it’s crazier to think that the federal government could possibly be a good judge of how 100,000 schools allocate their funds and educate their students.

This rewrite acknowledges that, and lets states set their own accountability plans. But the reason the federal government is this involved in education in the first place is because states couldn’t be trusted to educate children (particular low income and minorities) effectively; it’s considered a civil rights issue.

Which is why if the federal government wants to support certain populations of students, they shouldn’t ask the states to hold the schools accountable, they should ask the parents.

By tying funding to a student, the only schools that get federal funding are the schools who have convinced parents that they are trustworthy enough to educate their kids.

Of course, this only works if that parent can make an informed decision about how well a school can educate his or her kid. We’re not fully there yet, but with the No Child Left Behind testing provisions (that remain in the newly passed rewrite), we’re closer. 

Before NCLB, the public could trust that a school that said 75% of its students were proficient in math was probably doing alright. NCLB’s requirement for schools to report by subgroup (like racial or special education or English Language Learner status) had the ability to uncover achievement gaps previously unknown. If that same school had 85% of its non-poor students but only 30% of its poor students achieve proficiency, the public would probably want to know what’s going on. 

If a poor mom seeing those numbers decided to send her kid to a different school that has a greater percentage of proficient poor students, and that better performing school gets the money associated with that poor student, you have a situation where transparency created real accountability. Washington doesn’t have to decide if the school is effective and will likely spend the money correctly, the parent of the child can.

But wait, you argue, lots of parents are stuck in situations with only bad options. Those with the choice of a school with 30% proficient students versus 35% proficient students don’t really have options. That’s why we need the federal government to step in when states don’t!

Look, we have a lot of schools right now that are massively underperforming. Unfortunately, a lot of them are in areas with other massively underperforming schools, leaving parents with little to no high-quality options. Even though everyone can see their scores and they’re abysmal, the schools don’t improve. Even when the federal government tells them they need to reorganize, the schools don’t improve. Why?

Because they’re trying as hard as they can. They will undergo reorganizations, invest more in professional development, hold up their “Plan for Turnaround” as evidence they’re on the right track. And they’ll tell the federal government, the state government, the district, and their parents that. A lot of times, all four will be convinced.

But I would argue of those four options, parents (yes, even less well-educated ones) are the most likely to call bullshit. Parents can call bullshit on leadership and teachers who they see and about whom they hear stories from their children every day. State and federal governments can mostly look at numbers and reports. Parents care more than any other constituency about the specific quality of their specific school. State and federal governments care mostly about looking like they’re caring.

The testing provisions of No Child Left Behind gave parents more power to call bullshit. Teachers and principals who seemed smart and caring enough couldn’t hide behind an average proficiency score when their low income and minority students were struggling. Additional transparency of academic and financial results will help parents even more. 

Yesterday, the reauthorization of No Child Left Behind, now named Every Student Succeeds Act or ESSA, passed the Senate 85-12, and will be signed into law by Obama today. It was as bi-partisan a bill as exists these days, with everyone from the American Federation of Teachers Union to the National Alliance for Public Charter Schools applauding. 

How are both sides happy?

Republicans are excited about the reduced federal oversight and Democrats are excited about… the reduced federal oversight.

Now, of course, the Dems would not use the language “reducing the federal role in education” to describe the benefits of this bill. That’s not a principle of the Democratic party, and unions will work to increase the federal role if it furthers their agenda. Since teachers unions give an overwhelming amount of their money to Democrats, Democrats are traditionally pretty happy to help further the union agenda.

Until Obama, that is. Obama had an education agenda that for the first time in Democrat history was in direct opposition to the union agenda: most notably Obama’s (and later his Ed Secretary Duncan’s) support of heavily tying teacher evaluations to student performance on standardized tests.

Ask any Republican if a good way to evaluate teachers is based on a student’s academic performance, they’d probably say yes, that’s practical. When confronted with the fact the unions hate the idea, Republicans don’t care (it actually probably makes them like the idea more).

But ask a Republican whether the executive branch should mandate (or the basic equivalent) how teachers are evaluated by schools, and they’d probably say, no, that’s not the President's, nor any branch of the federal government’s role. 

By working to increase the executive branch’s role in education with a traditionally Republican-sounding agenda, the Obama administration managed to unite both Democrats and Republicans together against itself and its implementation of NCLB policies.

Crazy right? Instead of making Dems happy (more federal money and power!) and Republicans happy too (Look! Don’t you agree with these solutions I’m prescribing! I’m being pro-kid and anti-union!), the whole thing backfired with everyone upset: unions and Dems about the specific policies ( “high-stakes testing”) and Republicans about the abuse of federal (and particularly executive) power.

So upset they were motivated to work together across party lines to get a bill passed.

To be clear, the bill doesn’t change the standardized tests themselves; it doesn’t remove required yearly testing, nor even reduce testing in a meaningful way.

The most notable rewrite of the bill removes the Adequate Yearly Progress (AYP) measures which mandated schools take certain specific steps (like fire half the staff) when their students (or subgroups of students) were not making adequate improvements to their test scores year to year.

Basically, the federal government can no longer tell schools how exactly they need to improve. They do mandate that states have their own accountability plans, though the particular language used in the bill leaves a lot left up to interpretation. In a state’s accountability plan, “substantial weight” must be given to quantitative measures, such as standardized test scores and graduation rates, and less weight given to subjective measures, such as surveys from stakeholders. 

So the unions claim victory that it removes the “high-stakes” out of “high-stakes testing” and the Republicans claim victory that it removes federal power and gives it back to the states.

The rest of the 1,061 pages (yes, 1,061 pages) handed out even more victories.

The Democrats got increased funding for the Title programs (which are largely directed at students in poverty and other high-risk students) and a $250 million allocation for PreK funding.

Charters got increased funding, more flexibility on startup grants (would have loved that when we started Magnolia!) and elimination of the Highly Qualified Teacher requirement (would have saved our teachers money, time and headaches pursuing state certifications when they already had Montessori certifications).

And legendary heavyweight boxer Jack Johnson received a posthumous “sense of the Senate” pardon. 

Did I mention it was 1,061 pages?

Of course, no one got everything they wanted. Ted Cruz said it didn’t reduce the fed’s role enough and complained it didn’t include Title I portability.

Is it perfect? Heck no, but it was a compromise. And it’s a step in the right direction of reducing federal oversight while still retaining NCLB’s greatest gift to accountability: transparency.

A few weeks back, Politico got under my skin for screwing up the math behind a basic graph. This week is more painful: a WSJ editorial misinforming its readers by conflating two different stories about Arizona’s education funding to further an agenda I tend to support.

Jon Gabriel’s WSJ editorial applauded Arizona Governor Doug Ducey for solving “one of the hairiest problems in Arizona politics: How to give more money to teachers—without raising taxes” by taking $2 billion from the state trust lands to put “directly into the classroom, rather than funnel it through layers of bureaucrats.”

Immediately upon reading that, I was curious to see how putting additional money “directly into the classroom” worked. I had a couple ideas:

Handing teachers cash to spend on their classrooms 

Restricting the use of funds to certain instructional uses

If Ducey had pulled off the first option, that would have been impressive. States allocate funds restricted to specific purposes all the time, so I was assuming it was option two. In 2000, Arizona created the Classroom Site Fund which are funds that can only be spent on teacher pay and a “menu” of instructional goals including reducing class sizes or teacher training.

But after reading the legislation, it seems that Ducey didn’t even pull off option two.

What Gabriel calls “putting money directly into the classroom” seems to actually be simply increasing overall state funding to districts.

As the Arizona Republic reports, the additional funds go to an increase to the “base amount the state pays per student by $173, to about $3,600.”

Unlike the Classroom Site Fund, which were additional funds earmarked for a specific instructional purpose, these additional funds get lumped in with the base level state funding: basically the money the state sends a district every month to educate its kids (a district gets its money through a combination of local monies, state monies and federal monies). In Texas, the base funding is called FSP money and districts can spend it however they like within certain reasonable parameters. Because it’s such a large percentage of total per pupil spend (on average, around half of a district’s total spend), the parameters could not possibly be “spend this only on classroom teachers and not on principals or accountants.” 

Unless very clearly noted in the legislation, I don’t see how this particular $173 addition to Arizona’s state base funding could be earmarked for classroom spend only. But I couldn’t find anywhere in the legislation, or in any article written about the legislation, about how the funds were specifically restricted to classroom use rather than administrative costs. 

The editorial is titled “Arizona’s End Run Around the Education Spending Lobby.” The article praises how Ducey fought and beat union powers and district administrators who opposed this legislation on the basis that it was disrupting the status quo and not giving bureaucrats their share of the money.

The arc of the editorial is this:

Arizona spends too much on non-classroom expenses. (In 2013 Arizona spent only 54% of school funds in the classroom, compared with the national average of 61%.) 

Increasing non-classroom spending that goes to “district paper-shufflers” is not good. 

Increasing classroom spending (particularly in a state that has the one of the lowest classroom per pupil spends in the nation) would be better.

Increasing classroom spending without having to raise taxes would be even better.

Hey look at this guy, Ducey! He found a way to increase classroom spending without raising taxes!

Those dang unions and bureaucrats (including Mesa’s superintendent) hate Ducey’s idea, because they want the money for themselves (non-classroom spending), rather than the children (classroom spending).

Ducey passed his plan through the legislator; he’s a master negotiator. We should try him with hostages.

The thing is, I’m not a union woman who reads this arc and immediately screams “Bullshit! Ducey’s an evil Republican!” I am prone to agree with points 1 through 4, and I’ve seen unions and bureaucrats act in their own self-interest over children’s interests too many times.

But when point 5 seemingly fell apart with a little bit of digging (funds don’t seem to be allocated directly to teachers or classrooms), points 6 and 7 didn’t follow (and points 1 through 4 became a bit irrelevant).

On further research, the Mesa Superintendent’s robocall campaign Gabriel cited (as evidence of point 6) was not engaged to turn parents against this particular legislation. The campaign was initiated back in February in response to Ducey’s 2015-2016 budget proposal that wanted districts to shift a portion of their non-classroom spending to classroom spending. The robocall campaign was unsuccessful. The budget that passed in March cut $123 million in non-classroom spending. That may be an interesting story about the “education spending lobby.”

But that had nothing to do with this legislation.

As Gabriel had mentioned, the impetus for these bills was a 2010 lawsuit that ruled the state had underfunded education by not keeping up with inflation. As the Arizona Republic reports, that extra $173 per pupil that Ducey’s bill granted is only 72% of the amount the court ruling prescribed. That’s why unions and democrats were upset. They wanted more money! (Go figure!) 

So the real story seems to be this:

Ducey had to settle a lawsuit to increase state funding of education

Instead of increasing taxes, Ducey proposed to take most of the money from state trust lands, and it passed

Some people are mad, because they think schools are still underfunded by the state 

Settling a contentious lawsuit and increasing funding without increasing taxes is a notable story in itself, so Gabriel could have stopped there with his praise. What bothers me is that he didn’t. He inserted the government-bureaucrats-selfishly-maintaining-the-status-quo narrative (which perhaps could have been told back in March with the passing of the 2015-2016 budget) to make it more exciting. And the WSJ ran with it.

Last week, Politico wrote about the additional $3 billion in economic stimulus money the Obama administration invested in the School Improvement Grants program. The results? Two thirds of schools made modest or no gains and a third actually got worse.  Awesome, right?

It’s another case of federal government educational investment failure, and I’m not surprised. What I am surprised by is Politico’s awful math.

I tweeted about it, and even though I have all of a dozen followers, I figured because it was such a blatant error, someone would notice it and they’d get around to changing it. No such luck. Perhaps they’ve been too preoccupied with other recent reporting errors.

Or maybe they really can’t do math. In which case, I’ll explain here:

The second two graphs above depicting student standardized test scores should not add up to 100%. 

When we’re talking about students taking two tests (reading and math), these are the possible groups that could add up to 100%: students who only passed reading, students who only passed math, students who passed both math and reading, students who passed neither math or reading.

It’s unclear what “doesn’t meet or exceed” means, but even if it means the percentage of students who passed neither math nor reading, we still need to know the percentage of students who passed both (I am 100% confident that number is not zero in both years).

The thing is, I’ve actually never seen standardized test scores displayed that way before, because usually states don’t give you that information. Usually, there is one graph for reading (% who passed, % who didn’t) and one graph for math (% who passed, % who didn’t). Both of those graphs individually add to 100%. 

But in no conceivable way can the Politico test score graphs above add to 100%.

Part of me feels silly for taking the ten minutes to write out what seems so very obvious. But the fact that Politico hasn’t yet corrected this is even more silly. 

Today, Nathaniel Popper of Dealbook described Goldman Sach’s social impact bond deal with a pre-k program in Utah and questioned its success metrics (fair), while subtly implying self-interested foul play by Goldman (unfair). I’m afraid that the former issue makes the latter seem all too likely and people will conclude social impact investing is another way private investment corrupts public education. They shouldn’t.

For full disclosure, I’ve worked for Goldman and yes, like every adult and company I’ve ever met, Goldman engages in self-interested activities that make them money. With regards to social impact investing, we should be okay with that. In the end, it’s good for kids.

Let’s first be very clear about what exactly Social Impact Investing (also known as Pay for Success) is in the context of this program (something Dealbook was not). Let’s start with the basics:

PreK isn’t free. And money to pay for it isn’t unlimited.

Someone has to pay for PreK. In 2014, Utah passed legislation to spend more on PreK programs in the state, but not enough to grant free universal PreK in the state (that starts in kindergarten). State money allocated for PreK is not unlimited; it’s a finite number (and it’s a low one in this case... $5 million), so only those programs deemed best by the state can receive money. This may be a very obvious point, but it’s very important. We often talk of government spending in education as if the sky is the limit. It’s not.

Given the limited funds, not all programs can be funded with government money.

If the government doesn’t pay, who can? Non-profits of course! Well, another bummer... Non-profits don’t have unlimited money either. Non-profits, just like the government, can pick which programs get funded and which do not. Sometimes that means that good programs cannot get funded due to lack of government and non-profit funds.

There is an alternative to government and non-profit funding: debt.

Oftentimes when states or districts don’t have the cash upfront to fund their budgets, they issue bonds. They can borrow at relatively low rates because generally, the government is pretty good at paying back its money and so there’s little risk. As an investor in government bonds, you’re okay with a relatively low rate of return on your money because there’s little risk involved.

Typically in education, bonds are issued for construction costs rather than programming costs. At some point, a government taps out of the debt it can prudently issue. Again, we have to think about the investors. No one wants to be the 100th in line to receive his interest back and risk not getting it. When the debt is tied to construction, however, the investor has a way of mitigating that risk: the buildings are collateral. The government doesn’t pay back the investor with interest? He’ll own some of their building(s) and have a chance to make his money back by selling it.

With educational programming, mitigating risk by taking collateral is a little more complicated. The government didn’t pay back the debt issued for that PreK program? The investor can’t own and sell kids’ future earning streams and he doesn’t want hundreds of used Elmo toys. There’s little way to mitigate risk, and so the best way is to just avoid investing!

No taxes, no bonds, no problem: enter social impact investing

In the case where all these alternatives have been exhausted, social impact bonds create an alternative funding stream. It is very important to understand that these bonds are a last resort. If a state or district can get taxpayer money or donations for free and never have to pay them back, they’d do that. If a state or district can issue cheap bonds that investors are confident they’ll be able to pay back (or have collateral in case they don’t), they’d do that. Without social impact bonds, exhausting the options of taxes, bonds and philanthropy would be the end of the road. Sorry, PreK kids, your taxpayers and your bond investors do not want to help you.

But social impact bond investors do.

Social impact bond investors are not simply evaluating risk and return. They add a third parameter: social impact. Generally, investors are willing to accept more risk in their investment if the monetary return is higher. If the monetary risk is held constant, social impact bond investors are willing to accept more risk if the potential social impact is higher. For example, a high risk investment that returns 5% in interest may be passed up by many investors that think they can get more return for that level of risk elsewhere. But a high risk investment that returns 5% in interest but has the chance to lift hundreds of kids out of poverty may be worth it for a certain set of investors that want to see their money being used for good (but don’t want to just give it away).

[It’s important to note that social impact investors aren’t just “sort of good” people, because they want to do good but also make money, and the “really good” people simply give their money away. Again, money doesn’t grow on trees, so to the extent that the “sort of good” people can make money on their investments and reinvest that money again and again creating more and more positive social impact, they could be doing more good than the “really good” people.]

Given this set of “sort of good” people and their investment proclivities, Goldman Sachs raised $140 million from them and started looking for investments that would both return money and create social impact with which they’d be satisfied. One such investment was the Utah High-Quality Preschool Program. The terms:

Initial Loan: Goldman’s fund loans $4.6 million to the program and J.B. Pritzker loans $2.4 million.

Use of Funds: This funding pays to provide free PreK to up to 3,700 low-income students.

How “Social Impact” is defined: By enrolling in this PreK program, a student is set up for success in Kindergarten and beyond. Specifically, a child who would typically enter Kindergarten as a Special Education student, no longer needs Special Education services. This is positive for the student and society.

How “Social Impact” is measured: Students take a diagnostic test before they start the program. The test, known as P.P.V.T., is used by the district to assess whether the student is high-risk for special education - a score below 70 labels them as such. If a student scores below 70 before the program, implying that they will probably enroll in Special Education come Kindergarten, but ends up doing so well in the PreK program that he does not need Special Education services in Kindergarten, the PreK program is successful for that student. (The NYTimes article is mainly about how bad this measure is. I don’t necessarily disagree and I’ll get back to this later).

How Repayment works: If a program truly reduces the number of students needing Special Education services, it is saving a district money: in this case, about $2,600 per kid per year. The PreK program is being funded by Goldman at less than $2,000 per kid per year. If the program costing $2,000 a year for two years saves a district $2,600 per year for 10 years, why wouldn’t the state or district pay for it themselves? It would literally pay for itself. Well, they don’t fund it themselves because they’re not sure it does actually save them money and they’re not willing to risk it or they simply don’t have the money now, and they cannot convince taxpayers or bond investors or donors to give them the money now.

If, however, there is proof that the program worked, the state or district itself (or other donors excited to throw money at successful programs after the fact) might be glad to retroactively fund it -- so they make a deal to repay the debt in the case that the program is successful.

Terms of Repayment: For every child that scores below 70 on the P.P.V.T. and labeled at-risk for Special Education, but does not need Special Education services in K-12, Goldman receives repayment. Over the next 12 years, Goldman receives 95% of the cost savings per child ($2,470 according to the Salt Lake Tribune) per year from kindergarten through sixth grade, plus 5% interest. After that, Goldman receives 40% of the cost savings per child until the debt is repaid.

In the first cohort of students, 110 were identified as high-risk for Special Education, but 109 students did not need Special Education services by kindergarten. Therefore, Goldman received 109 times $2,470, or about $260,000 this year.

But this payment isn’t profit Goldman receives. It’s simply the repayment of the debt. So when Popper says in the first sentences of the article that “the investment also resulted in a $260,000 payout for the Wall Street firm,” it is akin to saying my friend borrowed 5 bucks from me, I paid her back two and my friend got a two dollar “payout.” That’s a bit misleading.

The most Goldman will make off this investment is their 5% in interest. When I first read the article, I had assumed that if more students "avoided special education” according to test results, Goldman would somehow get paid more. That’s not the case. Goldman simply gets repaid sooner. By my calculations, if they keep receiving the same $260,000 principal payment per year per cohort, they’ll have to wait 8 years to receive their principal back. If results are less than stellar, they might have to wait 20 years. If results are completely horrible, Goldman loses its entire investment. That’s not exactly a great deal for Goldman.

Now, back to the contested issue of success metrics. Popper does a fine job of explaining all the ways the P.P.V.T. is not a great test. The mere fact that the test implies 99% of kids no longer need special education discredits it as an accurate measure of student achievement. But then again, what one number is an accurate measure.

If the world were a perfect place with all parties having perfect judgment and perfect intentions, the state could see the results of the program and deem it “worth our money” or “not worth our money.” But it’s not, and so social impact bonds have to use flawed metrics as a proxy for judgment. But this flawed metric is what the state agreed to. Like it or not, taxpayers let the state make all sorts of judgments about which social programs are worth our money with less than perfect metrics, or no metrics at all. 

At least with social impact bonds, if the program is an unmitigated failure, it doesn’t get funded at all. In the status quo of government social program funding, mitigated failures get funded all the time as long as government bureaucrat(s) believed it in. Governments (and thus taxpayers) have no way to get most money we spend on social programs back. With social impact investing, taxpayers at least have a chance to refuse to repay investments in programs that were failures.

Unequivocally, we should seek better metrics to judge success. But imperfections in metrics should not mean we should stick with the status quo. The conclusion shouldn’t be “it’s hard to structure social impact bonds, because it’s hard to determine measures of success in social programs” or “we should only do social impact investing when the measures of success are perfect.” If we applied that logic to status quo government spending, we’d be complaining about all government spending on all social programs (and the NYTimes is most certainly not doing that). 

Social impact investing should not be evaluated on its own merits, but rather as an alternative to the status quo. Given our status quo in education funding, it’s a pretty good alternative.

And I don’t mean it’s a good alternative, because it makes Goldman Sachs money. This is not a slam dunk for Goldman. I can promise you no one at Goldman was thinking of ways to increase the firm’s bottom line and came up with social impact investing. Goldman is creating a market that would otherwise not exist - risky investments in PreK programs whose results cannot be seen until years later. 

We moan and whine all the time about how nothing innovative happens in education, and this is a big reason why: government bureaucrats generally don’t want to take risks investing in new programs that could be super successful, or could be total flops that prevent them from getting re-elected.

Social impact investors, on the other hand, don’t mind risking their money to create innovation and positive social change. 

Last week, Ginia Bellafante of the NYTimes told the story of Jack Spatola, principal of PS 172 in Brooklyn for the past 31 years. Eighty-five percent of his students qualify for free and reduced lunch and last year, 98% of third, fourth and fifth graders passed the math test and 76% passed the language test. Those are pretty impressive results. People are noticing (obviously, the NYTimes did!) and next month, business guru and author Jim Collins of the Good To Great is visiting the school.

There may be something in the water at NYTimes, but this is the second education article in weeks that highlights resource allocation. Bellafante writes, “Mr. Spatola’s most notable innovation, though, may be the manner in which he comes up with the money to finance all the additional academic supports his school deploys in the absence of enormous Wall Street donations, which are available to many charter schools.”

Instead of spending up to $50,000 on the Education Department recommended curriculum, the school creates its own curriculm. Instead of expensive textbooks (Bellafante writes that NYC schools spent $100 million on textbooks last year), Spatola spends the money on primary texts, and fiction and nonfiction works selected by teachers and administrators; students use the internet to research.

Bellafante ends with the question: is this scalable? Do we have enough educators “hardworking enough to take on the challenges demanded by the kind of culture the school has established: [d]evising curriculums and maintaining notebooks on every single child…”

I’m not certain hard work is the x-factor here. I don’t know the full story beyond this article (and that’s a huge disclaimer, because test scores really don’t mean everything but for the point of argument, I am trusting Bellafante’s praise). But what I think is unique in Spatola is that he acts like an instructional leader, not an instructional manager.

Depending on the state, depending on the district, principals have varying levels of control of their budget. Almost all district schools have set pay schedules, where a principal cannot say that he wants Mr. Johnson who has 4 years of experience to get paid more than Ms. Bliss who has 10 years experience and a master’s degree because Mr. Johnson is a way better teacher. Many principals are limited in their hiring discretion, often being forced to take experienced but ineffective teachers (and of course, it can be extremely hard to fire a teacher). 

When principals do not have full control of selecting and retaining their staff, they cannot be leaders; they are stuck with being coaches that have to make the most of the team they’re given.

And when principals don’t have control over the instructional model (because the state or district mandates class sizes, specific curriculums, textbooks, building layouts), they are merely managers of a school designed for them.

Districts create pay scales and mandated curriculums, in part to save the principals from exerting judgment (a principal may dislike a teacher for bad reasons and pay her less for it and that would be unfair!). By preventing bad judgment, they also prevent good judgment (the principal can’t offer the best teacher more money to incentivize her to stay), and thus stifle innovation and excellence. 

Spatola is an example of a principal that takes himself seriously as the instructional leader of his school: exhibiting good judgment allocating the resources over which he has freedom. The fact that his high test scores and customized resource allocation gets him in the NYTimes implies he's an exception to the rule, and that's a crying shame. 

If I asked you what you think we as a country prioritize more: math or cheerleading, what would you say? 

What if I told you of a high school that spends $1,348 per student on cheerleading and less than a fourth of that - $328 - per student on math? What if I told you that school wasn’t an anomaly, but schools consistently spend much more per pupil on sports and electives than core subjects?

What if I told you schools in poor areas of a district often get less money than schools in the wealthier areas, even as we as a nation have decided to prioritize funding towards low income students in attempts to reduce the achievement gap? 

Marguerite Roza’s Educational Economics: Where Do Schools Funds Go? reports research by the Center on Reinventing Public Education. One of Roza’s big conclusions after analyzing district and school data: We don’t spend the most on what we prioritize most. Roza writes:

“If one subscribes to the textbook explanation than an organization’s resource allocation system is a manifestation of its strategic priorities, then here is what CRPE research has shown is most important to urban districts throughout the nation: middle and upper-class students, not poor students. Electives and athletics, not core subjects. Gifted and high-achieving students, not struggling students. With implicit priorities like these playing out at the school level, it is no surprise that the system produces such a weak link between spending and goals for student outcomes.”

How can such blatantly contradictory spending patterns persist? Roza explains: simple ignorance. “Education leaders simply do not know whether their investments support their stated objectives...”

And not because they don’t necessarily want to know, but rather because the systems in place make it all but impossible: “District budgeting and accounting practices make it incredibly difficult to identify detailed spending patterns. Frankly, the five-inch-thick binders of impenetrable financial information plunked down in front of school board members do not even include this essential information.”

She continues, “Despite extensive financial reporting requirements, most districts cannot answer basic questions that the public might have during these budget discussions: How much do remedial reading programs cost compared with tutoring? How much does the district spend on professional development and how much would it cost if restructured?”

So what to do? Roza calls for accounting systems that actually let school leaders get the answers they need: “[U]ntil district leaders recognize that their financial accounting systems are inadequate at the school level, they will continue to be “driving blind” and heading toward financial instability, even disaster...What is needed are solid, reliable accounting systems that inform resource allocation.”

It’s 2015. I don’t think that’s an unreasonable request. So many of the problems within education are complicated and messy and cannot be solved with good technology; this can. BookReport (my new venture) is committed to doing just that.

Beyond district and school leaders, the public should have this data too. Even in Texas, where high school football is king, for anyone who’s seen Friday Night Lights, I think we can all agree that Tim Riggins needed a math tutor more than Lila Garrity needed a new cheerleading uniform. 

Imagine what Tami Taylor would have done with data telling her where the school’s money actually went...

Last post, I summarized what I considered the thesis of Dale Russakoff’s book about the Zuckerberg-Booker reform effort in Newark.

I was wrong. In a recent interview with NPR, Russakoff said her biggest takeaway was about resource allocation.

Russakoff doesn’t think more money is the answer; Newark already spends over $20,000 per student, well above the national average of about $12,000 (though in line with other east coast cities). Rather, she sees a bloated bureaucracy where too much is spent at the central office rather than given to the schools to address particular student needs. As Joe Nocera writes in the NYT, “The KIPP charter network, which runs Spark, gets $16,400 per Spark pupil, of which $12,664 is devoted to the school. The district schools get $19,650 per pupil, but only $9,604 trickles down to the schools.”

Russakoff gives the powerful contrast between KIPP Spark (a charter school part of the KIPP charter network) and Brick Avon (a Newark district school).

At KIPP Spark: Principal Joanna Belcher said, “We designed the school and the budget purposely based on what kids are going through and what they need.” Russakoff writes: “To support students who struggled, Belcher placed two certified teachers in each kindergarten, as well as in every math and English class in grades one through three. Students who fell behind got small-group instruction from one teacher while the other led the lesson for those on grade level. For children who still couldn’t keep up, a full-time learning specialist - one for each grade - provided tutoring and other interventions.” 

“By contrast, Avon and all district schools had one teacher in each classroom. In kindergarten, there was also a classroom aide, who was not required to have graduated from college. Most district schools shared one to two specialists among all grades, if they had any at all.” 

At KIPP Spark, when a boy slugged teachers without warning, their lead social worker Sarah Dewy “created an ‘office’ for him - with his own child-sized desk and chair - in a corner of her work space, and she and Adams supervised his work while his mother went through the long process of having him classified for psychiatric treatment.”

At Avon, star kindergarten teacher Princess Williams had a student with “similarly violent tendencies (in his fiercest rages, he threw chairs, terrifying classmates), but it took the district bureaucracy almost eight months to deliver the support she requested.”

Princess Williams was committed to teaching at a district school, because she was convinced that charters didn’t serve the neediest of Newark’s kids. Yet after a year of struggling with the district to get aids into her classroom for severely disturbed kids throwing furniture in her classroom, Williams left Avon for KIPP Spark, saying that it wasn’t because charters were necessarily better, but “[b]udgets tell you a lot about values...We need to change the values of our district.” She wants to one day return to the district to help make that happen.

Pretty compelling case for better resource allocation right? Why isn’t it happening then? 

Because not everyone buys Russakoff’s story. New Jersey teacher blogger and Rutgers PhD candidate Mark Weber wrote a 22-page brief ‘debunking’ Russakoff with data from the NJ Department of Education. The data is correct, but I think Weber’s analysis of it is incomplete. Here’s what the data tells him, followed by what the data tells me.

Weber: How can Russakoff say that charters spend more on classrooms when the Newark public school district has a lower per pupil administrative cost than any charter school in the city?

This is a given, because the district is so much bigger (serves many more pupils) than any charter.  If Newark started paying their Superintendent $5 million a year, they would still have a lower admin cost than any other charter paying their Superintendent less than a tenth of that. That doesn’t mean Newark is therefore less bloated than the average charter; it’s just math.

Russakoff’s main argument isn’t about the aggregate amounts of money spent on instruction versus admin anyways. Rather, it’s the way the money is actually spent, which brings us to the next point:

Weber: KIPP Spark and BRICK Avon are not average schools. On average, Newark public schools have more instructional personnel and education support staff per pupil than most Newark charters.

Further, Weber notes that while BRICK Avon has an unusually low number of social workers on staff, they do have a psychologist and school counselor. To me, this data actually strengthens Russakoff’s point. The point isn’t which school has more employees serving kids. The point is which school is actually addressing student needs better. 

The Princess Williams example is only one classroom and one teacher’s perspective. But what Princess Williams and her students needed was a social worker, not a counselor or a psychologist. Maybe other students at Avon also needed a social worker - more than a counselor or a psychologist. Maybe, given the students that year, the school needed only 10 hours of contracted psychology work and a social worker. In that scenario, they could keep the total number of support staff and student support costs the same, but they would be addressing student needs better. 

For a district (or Weber) to argue that they are supporting students better simply because they have more support staff per kid is to pretend all students, schools, and situations require the exact same prescription and the exact same staffing model. Who really thinks that?

Weber: The Newark Public School District spends far more per pupil on education support services than any Newark charter school.

Weber admits that many charter schools (including KIPP) don’t report any spending in this category, which means that this is probably not a good data point. As we know, KIPP employs social workers, so having $0 in reported student services doesn’t mean they aren’t supporting students with money for social workers. In school accounting, there is some discretion about how a school can record a given transaction, so if we’re trying to compare spending on student support services, but KIPP is coding all their social workers to the instruction or admin category, the data in this form is not at all comparable or meaningful.

In fact, comparing aggregate per pupil data tells us very little. First, per pupil data is skewed when you are comparing a district with tens of thousands of students to a school with hundreds. But more importantly, in any given category of “Classroom Instruction” or “Educational Support Services” or “Administration,” the aggregate total doesn’t tell us much about how efficiently and effectively it was spent. According to the NJ’s Taxpayers’ Guide to Education Spending, included in “Classroom Instruction” is professional development and included in “Educational Support Services” are salaries for those reporting attendance. If a school spends $1 million on ineffective PD (and really, what other kind is there?) and hires two clerks to report attendance for $100,000 per year each, they’ll get credit for spending lots of money on “student instructional and support.” They shouldn’t.

We simply can’t make any conclusions until we dig deeper to the actual transactions, the actual staffing models and salaries, the actual services contracts. I applaud Weber for presenting us with the data (the raw data is not in an easily digestible format, so kudos to him for making it more so), but the conversation must extend beyond big, aggregate numbers.

In September 2010, when Mark Zuckerberg announced he was giving $100 million to Newark public schools to create a national model of education reform, I was an investment banker at Goldman Sachs (just about the only thing I did other than work was read my NY Times education alerts).

I remember having two thoughts:

How cool. Maybe if I stay here at GS, make tons of money, someday I can give it away to a school district to create major change in US education 

Maybe I should go to Newark and work for the district and help create major change in US education

Well, I’m glad I didn’t do either. Two hundred or so pages into Dale Russakoff’s new book, I’m convinced those were naive goals. The Prize: Who’s in Charge of America’s Schools chronicles Zuckerberg’s investment and the reform effort in Newark led by Mayor Cory Booker and Governor Chris Christie. It’s a sobering book about how hard tranformational change in a school district is. Five years later (Booker promised this change in five years), the district has improved, but no one is looking to Newark as a national model of transformational change. Still, I think our nation can learn a lot from the Newark example.

First, that reform must always be contextualized. In order to reform education in Newark, you need to, well, reform Newark’s schools. It seems obvious, but if Russakoff has a thesis, it’s that the national focus of Zuck’s donation and Booker’s political stardom coupled with the vision of creating a national model of education reform hindered their ability to gain community buy-in and ultimately produced less-than-promised results for Newark’s kids.

Some key missteps along the way:

Setting a five-year timeline. As part of the reform effort, Newark needed a new superintendent. Booker’s first choice was John King, then deputy education commission of NY with a background of leading top-performing charter schools in NYC and Boston. King carefully considered, spent a weekend in Palo Alto with Zuck, his wife and Sheryl Sandberg and then turned the offer down, because of the unrealistic timeline: "King’s view was that no one has achieved what they’re trying to achieve: build an urban school district serving high-poverty kids that gets uniformly strong outcomes...[He] believed it could take almost that long to change the system’s fundamental procedures and to raise expectations across the city for children and schools...King had questions about a five-year plan overseen by politicians who were likely to seek higher office.” 

King was right on both fronts. They couldn’t achieve their vision in five years, and the politicians in charge of that vision have both moved on to the national stage (Booker to the sentate and Christie to a presidential bid). 

So why didn’t the vision turn to reality? What was the plan?

Announcing the $100 million gift on Oprah before telling the community started the entire effort off on the wrong foot. It was Booker deciding that national attention surrounding the reform effort trumped community engagement and cooperation. It was gimmicky, and it was a bad call. As Ras Baraka (then high school principal and now mayor of Newark) said,“Quietly, a lot of people agree with a lot of this stuff... This dictatorial bullying is a surefire way to get people to say, ‘No, get out of here.’ It becomes my crew versus your crew, and it’s disrespectful to the people of Newark.”

The “my crew versus your crew” (teachers unions vs ‘reform-minded’ superintendents; charters vs traditional district schools; the public vs the big corporations) saga wastes so much time in education reform, making even the most common-sense of reforms somehow controversial. The Oprah announcement fueled that fire.  

Spending $21 million on expensive NYC consultants with no ties to Newark didn’t bode well with the public, nor was it particularly effective. 

Almost $3 million went to consultants for a “fact-based” needs assessment of Newark, but this came a month after Booker had already sent Zuckerberg the blueprint, “Creating a National Model of Educational Transformation” created by McKinsey pro bono consultants. The plan was for Newark to “coordinate a critical mass of local and national partners with proven models of excellence to deliver high impact programs and best practices.” Sounds awesome when you’re trying to get people excited, but it’s not exactly a plan for Newark. Russakoff notes, “The language of national models left little room for attention to the unique problems of Newark, its schools, or its children.”

Indeed, the consultants weren’t great at solving the problems unique to Newark. As one example, more than $3 million went to Parthenon group, in part to create “data dashboard” for principals. Great in theory (and should definitely be part of a national model!), but Russakoff writes, “unlike the dashboards of corporate executives, which were updated continuously, the principals’ dashboards...were updated once a year when new test score data arrived. Several principals said privately that they could have assembled the data themselves.”

The national model language was fine in theory (it was exciting and inspiring for sure!).

But focusing on creating something that scales on a national level before you have created something that works on the local level is silly. And it’s simply not the way anything great - great companies or great school systems - are built.

Zuckerberg didn’t create Facebook trying to be the world’s biggest social media site right out of the gate. He created a Facebook for Harvard, then for Ivy Leagues, then for colleges, then for everyone else. This took years and years.

The KIPP charter schools didn’t start because Dave Levin and Mike Feinberg set out to create a national model for no-excuses charters. They first focused on teaching their classroom of Houston kids really well. Then a whole school of Houston kids and New York kids. Then more and more schools of kids across the country. 

In Newark, the Zuckerberg-Booker intiative was more interested in creating a blueprint for America than a plan for Newark. But a blueprint to transform America’s schools just didn’t make sense. Quite frankly, a blueprint for the Newark district didn’t even make sense.

To transform a district, you need to transform its schools. Transformation happens school by school, teacher by teacher, student by student. It’s messy and hard work.

But if there were any spots of hope in the book, they were the leaders and teachers in the two schools Russakoff traced: BRICK Avon (a public school that was committed to being a district school while seeking some flexibility from the district) and KIPP SPARK Academy. I think I cried a half dozen times between the stories of the KIPP principal taking a parent who had been beaten by her boyfriend to the hospital, the teachers who organized a carpool for her kids to ensure they got to school, the teachers and leaders in both schools who did whatever it took to care for their students. 

After years of hype and $100M later, adaptive learning platform Knewton is live. Knewton’s CEO Jose Ferreira told Wired “this robot tutor can essentially read your mind.”

I’ve criticized Ferreira’s and Knewton’s bold claims before here and here. Today I tested out the platform to see if the reality matches the rhetoric.

Am I surprised it doesn’t? Of course not. But I am surprised it’s so far from the rhetoric.

I pretended to be a student and selected a quiz on Evaluating Algebraic Expressions. I got a couple of textbook looking questions that I answered correctly. Then I got this question:

I decided to see what would happen if I made a common student error - evaluating the expression left to right - without regard for the parenthesis. So I did 4 times 2 times -1 which equals -8 then added 1 for  -7. As a teacher/tutor, if I saw -7 on a student’s paper for this question, I would immediately assume they made this error, and walk them through how that error led to an incorrect answer. That’s not called reading someone’s mind; that’s called understanding pedagogy and student misconception.

But as you can see above, when I got this (very) common wrong answer, my mind-reading tutor told me “No, but keep trying.” Brilliant.

I mean really, Knewton? This is the really easy stuff. How hard would it be to program your tutor to say, “Did you multiply 4 times -1 then add 1? Dear child, when there are parenthesis, you need to first evaluate the expression inside the parenthesis first, then go about calculating left to right...” Have it speak in a British accent, and I’m in robot love.

Perhaps worse are the instructional videos scattered throughout. First, I wasn’t quite sure how or when they decided to give me a video, because for the first few problems, I got a video after every question, even if I got the answer right. Do you know what’s really annoying? Getting an answer right and then watching a 2 minute video about how to solve the problem. Of course, it allows you to always skip the video, but again, if they are reading my mind, can’t they understand I don’t want a video after a correct answer?

Secondly, the videos are pretty lame. Here’s a snapshot of one of them.

So yeah, not what I would expect $100mm later. Khan has been doing this for years.

Perhaps Knewton’s argument would be that they are trying out all types of videos from all types of teachers and then they will learn which ones kids like based on...BIG DATA. Because big data tells us everything... OR, what they could do is get a great teacher that understands pedagogy to curate or create top-notch videos and then use student data on those curated/created videos. If Knewton in fact did that, and this video got through the high-quality curation/creation screening, then I don't have a lot of faith in their pedagogical judgment as an organization. Which is nothing new, I suppose.

But it’s a good reminder that when thinking about edtech, content is king. You can have the fanciest algorithm around, but if your questions and videos aren’t engaging, kids are not going to learn the way they would with excellent content.

“This very obvious fact – that each generation is taught by an earlier generation – must be kept very firmly in mind. The beliefs which boys fresh from school now hold are largely the beliefs of the Twenties. The beliefs which boys from school will in the Sixties will be largely those of the undergraduates of today. The moment we forget this we begin to talk nonsense about education. We talk of the views of contemporary adolescence as if some peculiarity in contemporary adolescence had produced them out of itself. In reality, they are usually a delayed result – for the mental world also has its time-bombs – of obsolete adolescence, now middle-aged and dominating its form room. Hence the futility of many schemes for education. None can give to another what he does not possess himself.” -- CS Lewis, “God on the Dock,” 1946

Here, CS Lewis is addressing those who were perplexed that Christianity among England’s schoolchildren was waning. Lewis argues this is inevitable: “To blame the schoolmaster of the last ten years for it would be ridiculous. The majority of them failed to hand on Christianity because they had it not: will you blame…a stone because it yields no blood?”

Isn’t it the same about math in schools today? We bemoan that so few kids have the skills for the "21st century economy”: creativity, critical thinking, perseverance. We need kids that are able to be programmers and engineers, but we have a bunch of kids walking around saying they are bad at math.

Because we have a bunch of adults, and even worse, a bunch of teachers walking around saying they are bad at math. Of course, there are many teachers that are great at math. But isn’t it a bit ridiculous that we can’t say that about all teachers that teach math? To borrow Lewis’s logic: Teachers who think they are bad at math are not going to be able to teach children to be good at math.

Sure they can, you say. We can just teach those teachers how to teach math. They don’t have to be good at math, just teaching math! Learning how to teach multiplication or the area of a triangle or solving for x can be broken into steps and teachers can learn those steps without having to feel great at math.

Well, sure, but that has played out pretty poorly: millions of students learning steps to solve for areas or the value of x and not really understanding why or how, millions of students growing up to be adults who call themselves bad at math.

But we don’t have enough people who are good at math wanting to teach kids! Those people go on to make the big bucks as bankers or engineers or programmers.

Nonsense. Being good at math doesn’t require an innate ability only bestowed on the few. Being good at math requires thinking flexibly about numbers, using what you do know to figure out what you don’t, and perseverance. 

By this definition, all math teachers can get good at math, but we need to be purposeful about getting them there. In teacher education, we spend too much time developing a teacher’s ability to teach math, and not enough time developing a teacher’s ability to think about numbers.

That means we don’t review formulas with teachers or make sure they can pass a multiple choice test. It means professional development sessions where teachers are pushed to solve problems beyond the scope of what they’ve seen in a text book, where teachers wrestle with a single problem for thirty minutes, where teachers share strategies for solving math problems (rather than just strategies for teaching math solutions).

If a teacher can’t be expected to persevere through an unfamiliar math problem by using logic and problem-solving, we’re kidding ourselves to expect she will be able to teach her kids logic and problem-solving.

So for those math teachers that cannot yet do that, they need to practice it until they can. Rather than get distracted by this new reform or that, this is what we should be spending our time on.

As Lewis says:

“As the teachers are, so they will teach. Your ‘reform’ may incommode and overwork them, but it will not radically alter the total effect of their teaching. Planning has no magic whereby it can elicit figs from thistles or choke-pears from vines.”

School's out for summer, and so my time with the Magnolia kids is done for the year. Due to traveling and workworkwork, I wasn't able to go as often as I liked and I didn't get to do a post-test for the kids individually. But we had a lot of fun (the kids often begged for just one more problem after their 20 minutes was up) and I was proud of the kids’ progress in being able to clearly articulate their reasoning and to break down what first seemed like scary problems into easier, more manageable steps so we could find the solution instead of just throwing up our hands saying “I don’t know.” Kids talked and reasoned like true mathematicians. 

Unfortunately, the videos turned out horribly, which is pretty pathetic for a 20-something year old, I know, but I was always more focused on the kiddos so when the camera fell or when kids leaned in front of it, I just kept teaching. 

I was able to salvage a couple clips of student misconceptions about place value, however, and in my mind, there's nothing more fascinating than witnessing a new faulty way students are thinking about math. Knowing how kids can get tripped up makes me a better math teacher.

One of the most interesting moments occurred when working with students multiplying decimals (keep in mind these are 2nd and 3rd graders). In the moments before the video starts, I had shown them a string of problems:

6 x 100

6 x 10

6 x 1

6 x 0.1

6 x 0.01

4 x 0.2

0.4 x 0.2

We had reasoned through them by simply reading the problems out loud. It wasn't too hard for students to figure out that six times one tenth was simply six tenths and likewise with hundredths and thousandths. It was almost like asking what's six times one meter. Six meters.

Then when they saw 4 x 0.2, it was the same thing. Four times two tenths is eight tenths like four times two meters is eight meters.

Then when we got to 0.4 x 0.2, someone had made a trick: just count the zeros. So it would be 0.08 because four times two was 8 and there were two zeros within 0.4 and 0.2 and so there would be an 8 in a decimal including two zeros, making it 0.08

I wasn't expecting that at all. I had expected that they would have recognized that it was simply one power of ten lower than the previous problem. (My previous problem sequence had set them up for that realization, I had thought). But it was silly to be surprised: she was simply applying a trick she had learned about zeros and multiplication. She clearly had some misunderstandings about it (which is why when a student doesn’t understand why a trick works, she shouldn’t yet learn it. Yes, that means she’ll sacrifice some answers on a test that she would otherwise get correct in the name of true understanding. Live with it).

Instead of walking her through why her trick wasn't quite right, I gave them another problem (0.04 times 0.05) that didn't fit that pattern as a way of saying "okay let's test that trick." 

You’ll see in the video, a student gives an answer of 0.20 and then explains there are two zeros she has to put somewhere (which was confusing because I was figuring she would say there were four zeros in the problem). She couldn’t really explain in a mathematical way, so I then asked the students if we could convert the problem into fractions, if that would make it easier. After some thought, a student gave the answer of twenty one hundred thousandths, and I intentionally wrote it in fractional form. 

I was trying to get the students to see that ‘count the zeros’ trick does, in a sense, work when you put in fractional form (i.e. the numerator is four times five or twenty and the denominator is 100 times 100 and in order to calculate that you can ‘count the zeros’ for a total of 10,000). The video picks up again with the student explaining her answer but she is referencing the decimal notation not the fraction notation. Here, the four zeros in the decimal notation do represent the negative power of 10 that should be applied to four times five (This of course only works when one actually writes the 0 in the ones place). 

After this, I drew their attention back to the fractions and asked if anyone could make a connection between the four zeros in the decimals and what I wrote in fractional form. No one could. As 2nd and 3rd graders, I understood (not all of them had even multiplied a fraction by a fraction prior to this lesson) and decided to end the lesson there, ecstatic to learn of this new “trick.”

After this, I’m never going to teach the actual multiplying decimals trick without first giving the students this “counting the zeros” trick and asking if it works. Because it works some of the time, it really forces kids to grapple with place value. Why does it work with problems like 0.02 times 0.04 but not 0.06 times 0.04? Can we come up with a better trick that works all the time? In what ways is this trick similar to the one that actually works? In what ways is this trick different?

After interviewing the kids, the next step was to organize the data so that I could group them into small groups.

For the interview, I had made a Google Form that allowed me both to select the strategy used and to type in the student’s explanation word for word. Then I took the strategy data from the Google Form and color coded it to more easily view and sort.

I sorted kids on understanding, placing the kids who understood the least up top and who used the most efficient strategies on the bottom. This was not at all “ranking” the kids from worst to best, because that was impossible given the qualitative and limited aspect, but rather, it was a very general grouping by similar understanding/strategy use. (As in, this isn’t to say Student 15 is 4 students more advanced than Student 11, but they both often used counting strategy and couldn’t solve 28 + 35, and confirmed by their explanations, I concluded they were in a similar place of math reasoning).

I looked at data from the CGI questions and MRI questions at the same time to do this ordering but it’s simpler to discuss them separately. Below is the data from the CGI problems (the problems were all word problems, but for the sake of space, the columns show the algorithms the word problems represented):

Group A: These were the kids who weren’t able to fluently count to 120. Regardless of how they performed otherwise, I wanted to group them together and prioritize their counting. Notice that 3 of them couldn’t solve any problem, but one of them solved 4 of the 6 problems correctly (I’m guessing only because they were with smaller numbers, which she was fluent in counting, but once she gets her counting solid, I’m assuming she’ll take a big leap in her math understanding).

Goal for Group A: Count fluently. Mentally add/subtract numbers close to each other, having an understanding of the counting sequence (i.e. quickly know 50-47 is 3).

Group B: These were the kids that were confused about how to represent one or more of the word problems (determined by the explanations I recorded). 

Goal for Group B: Fluently solve simple multiplication/division word problems w/ counting and derived facts.

Group C: These were the kids that clearly understood what the questions were asking. Some made minor mistakes in their counting or derived facts. When it came to the MRI Questions (graphic below), their explanations showed lots of confusion concerning place value as it relates to addition/subtraction.

Goal for Group C: Mentally add/subtract 3-digits, using understanding of place value.

Group D: These were the kids that were able to solve multiple MRI problems with strategies that were efficient (the efficient strategies are in blue and purple - with one not being better than the other; I wanted the visual to show if students used different strategies or all the same strategies), or in their explanations convinced me they understood place value as it relates to addition/subtraction. However, all kids struggled with either multiplication or division of multi-digit numbers that were multiples of 10.

Goal for Group D: Mentally multiply/divide multi-digits, using understanding of place value and powers of 10.

These seem like reasonable goals for the few weeks I have with the kids, but will be revisiting them after each group lesson.

Over the course of last week, I interviewed all 24 students (first through third graders) in the class. It took on average 25 minutes per kid and was generally what I had expected: kids used a wide range of strategies, with some kids needing direct modeling while others were fluent in derived facts or automaticity (for examples, see previous post). However, there were a few things that stood out with this group in comparison to other kids I’ve interviewed that have been taught in a more traditional math setting (one teacher in front of the room teaching 25 kids the same thing at once).

A reminder of the questions:

Questions they could use manipulatives:

I have 10 dollars. If I spend 4 dollars on lunch, how much money would I have left?

Isabelle has 3 apples. How many more apples would she need to have 12 apples altogether?

There were 11 people on the bus. Some got off. Now there are 6 people on the bus. How many people got off the bus?

Corey has 4 boxes of books with 6 books in each box. How many books does Corey have?

Henry has 28 books. If he wants to put his books in 4 boxes with the same number of books in each box, how many books should he put in each box?

Alyssa had 28 strawberries. She picked 35 more. How many strawberries does she have now?

Questions I would not allow any use of materials (only mental math):

1000 - 998 =

99 + 17 =

15 + ___ = 200

20 x 15 = 300, 21 x 15 =

60 x 40 =

7000 / 70 =

15 x 12 =

There are 295 students. Each school bus holds 25 students. How many buses are needed to fit all the students?

Takeaways:

1. Kids were persistent and patient. If a student didn’t know how to do 28 + 35 without a pencil and paper, I let them use the stamp game. Some students counted out 28 tiles, then 35 tiles, then counted all 63 tiles together, starting from 1, which is the least efficient strategy possible (and not the best use of the stamp game), and so generally a teacher doesn’t want to see this. However, for a student who struggles with understanding place value and has not yet understood that 28 + 35 is the same as 20 + 30 + 8 + 5, it is encouraging to see them have enough patience to count first to 28, then to 35, then to 63. The nature of Montessori education and kids having space and time to work at their own pace definitely allows to build the patience muscle in kids.

2. Kids have a solid understanding of making sense of word problems and knowing how to model them directly. When I’ve interviewed kids before, usually a higher percentage of them jump straight to simply picking addition or subtraction and the two numbers on the page. Only a couple did that (and those were the ones who had no math confidence). Again, this is a strength of Montessori education, where kids are constantly using manipulatives to represent math.

3. A few students couldn’t count fluently to 120. By first grade, most all kids know how to do this, so it often is just assumed when a kid walks into your class that they know how to count. I’ve never thought of asking a kid over 6 to count, but at NCTM, one presenter said it’s often a problem that goes undiagnosed in the most struggling kids. Looking back, I wish I had done this more often. This time, for those kids that struggled the most, I had them count to 120, and sure enough, they couldn’t so that will now be the number one priority now for them (I also have no experience with teaching kids to count so will be researching this a lot).

4. 60 x 40 was the hardest problem, harder than 12 x 15. With 60 x 40, the big numbers intimated a lot of them, with many saying “I don’t know how to multiply big numbers like that.” When I pushed them to reason through it, here were some of the answers I got (these are only answers from students who answered the 6 x 4 word problem correctly to show that it’s place value in multiplication tripping them up, not multiplication itself).

224. Basically the same as 6 x 4. Well, I basically did the same thing as 60 times 4. I took 60 one time and 60 another time and got 120. I added one more and got 180 and added one more and got 224.

240. 6 x 4 = 24, 0 x 0 = 0 so that would be 240.

240. I just like to do times in my head and then add zero. 6 x 4 is 24. 240. I added a zero because they had a zero at the end. [Me: Why just one zero?] If it was 600 x 400 then I would have to add two zeros.

90. 6 taken 4 times is 19 then 60 taken 40 times would be 90. Why? You just keep asking difficult questions like why? That's what I'm here to do? So really, I'm teaching you stuff?  :)

Lots of kids commented on how difficult it was to explain what they did to get their answer, but kids improved at this even within the time of the interview. With some, it took a lot of repetition “I know you don’t think you know how you did it, but I would really like you to try,” or to a kid who would say “I don’t think it makes sense, I just did it,” I would encourage “Actually, I think what you did does make sense, but I want to make sure what I think you did was actually what you did. Could you help me understand?”

Even if I had the best data possible, I think that working with them one on one and pushing them to explain their reasoning and giving them practice doing so in a very safe space was invaluable.

Letting them sit and think and sit and think some more required a TON of patience on my part, and I’m guessing it still will when we do group work together. But I’m hoping their Montessori patience will rub off of me...

Coming off NCTM, I’m inspired to put what I had learned into practice. So for the next four weeks, I will be working twice a week with a Lower Elementary (first through third grade) class at Magnolia Montessori For All in math small groups and documenting it here.

Before I started teaching, I needed to decide (1) what to teach and (2) how to group the 24 kiddos. 

In order to do that, I interviewed the kids to understand their mathematical thinking. That’s a pretty general term, so I am using two of my favorite frameworks to give me better data than “seems pretty good at math” or “shoot, can’t even count.” I combined questions both from the book Cognitively Guided Instruction (CGI) and Marilyn Burns’s Math Reasoning Inventory. 

CGI’s goal is to nurture kids’ intuitive sense-making of math, so that they come to understand math as intuitive rather than as a set of rules they either get or don’t get. Breaking that down, it means:

1. Kids can solve math problems with direct modeling before being formally taught math

Kids are able to model out everyday situations with rocks, chips, fingers before they’ve ever heard the word multiplication or know what + represents.

For example, about half of the children in a study of kindergartners were able to solve this multistep problem, which they had never seen before, using direct modeling: 19 children are taking a mini-bus to the zoo. They will have to sit either 2 or 3 to a seat. The bus has 7 seats. How many children will have to sit three to a seat, and how many can sit two to a seat?

2. Kids solve problems with different strategies. As a teacher it is important to know which strategies kids are using. 

For any given problem, there are multiple correct strategies to use. While all may lead to the correct answer, there are some more efficient and the use of them usually means a higher fluency and understanding of math (our goal for our kids).

For example, consider this: I have 10 dollars. If I spend 4 dollars on lunch, how much money would I have left? Here are 4 different ways kids in the same class solved it.

Direct Modeling: Kid takes out 10 tiles. Then kid drags 4 tiles away. Kid counts how many tiles are left and says 6.

Counting Strategy: Kid counts 4 down from 10. 9,8,7,6 and gets 6 for his answer.

Derived Fact Strategy: “Well I know that 10-5 is 5 and so 10-4 would be 6.”

Automatic Recall: “10-4 is 6. That’s something I just know.”

3. Teachers should guide kids along the spectrum to the most efficient strategies, while maintaining intuition as the highest priority.

There are ways teachers can teach that allow for kids to move from direct modeling to counting to derived fact without losing intuition. CGI is not a curriculum but rather a way of teaching that encourages this. It involves selecting the correct problems/progressions of problems, recognizing strategies kids are using, encouraging kids to communicate strategies clearly to peers, creating accurate visual representations of strategies (sometimes very difficult when a kid is not very clearly explaining himself), and selecting appropriate follow up questions to elicit additional student thinking. That sounds like a whole lot of words that don’t necessarily mean anything without an example, but I’ll be posting videos of my work with the kids that I hope will illuminate it.

In order to figure out where kids were in the spectrum of strategies, I asked them a series of word problems. I made sure they were all of different types (CGI gives a framework of problem type classification), because oftentimes kids can solve some addition word problems but not others. For example, most kids can easily solve what CGI refers to as Join Result Unknown, i.e. Sally has 5 apples and she gets 2 more apples. Now how many does she have? But the Join Change Unknown problems are a little more difficult, i.e. Sally has 5 apples and she gets some more from a friend. Now she has 12. How many did her friend give her? 

These are the CGI problems I selected:

Separate Result Unknown: I have 10 dollars. If I spend 4 dollars on lunch, how much money would I have left?

Join Change Unknown: Isabelle has 3 apples. How many more apples would she need to have 12 apples altogether?

Separate Change Unknown: There were 11 people on the bus. Some got off. Now there are 6 people on the bus. How many people got off the bus?

Multiplication: Corey has 4 boxes of books with 6 books in each box. How many books does Corey have?

Measurement Division: Henry has 28 books. If he wants to put his books in 4 boxes with the same number of books in each box, how many books should he put in each box?

Join Result Unknown (place value emphasis): Alyssa had 28 strawberries. She picked 35 more. How many strawberries does she have now?

Then, I added some problems from Marilyn Burns’s Math Reasoning Inventory (MRI). These problems are designed to uncover student reasoning with a particular emphasis on place value. These problems are specifically crafted to see the efficiency level of student strategies. A lot of students may be able to perform standard addition/subtraction/muliplication/division algorithms, but they may to know what they actually are doing when they perform the algorithm. These questions can reveal how students really think through problems when not given the opportunity to go straight to performing the algorithm, therefore displaying their true mathematical reasoning and understanding.

For example, when asked to solve 1000 - 998 with paper and pencil (or in a Montessori curriculum, with the stamp game), two kids may both perform the standard algorithm and get it correct. However when I tell them they can’t use paper, one student may look at it and freak out and tell me that’s super hard to do in their head because they have to regroup 3 times. When he tries, he gets  888, which makes no logical sense, but he doesn’t recognize that, because he is not focused on math making sense but on the algorithm. Another kid may take a look and immediately recognize that 2 more than 998 is 1000 so the answer must be 2. The former kid may be able to subtract on paper, but he lacks the necessary understanding to say that he has mastered the concept of subtraction. Teachers need to know the difference if the goal is true understanding.

Here are the MRI problems I selected:

1000 - 998 = 

99 + 17 = 

15 + ___ = 200

20 x 15 = 300, 21 x 15 = 

60 x 40 = 

7000 / 70 = 

15 x 12 = 

There are 295 students. Each school bus holds 25 students. How many buses are needed to fit all the students?