Counterfactual Reasoning, Black Swans, and Common Planning Conundrums

There are many nuts-and-bolts topics discussed (which is probably why you are here) in the name of counterfactual thinking and black swans.  All interesting, but you may by here for only one, so scroll down to what interest you.  In order, the topics are: Introduction, Configuration, Schedule, Placement, Last Thoughts.  Enjoy.

Counterfactual Reasoning and Black Swan Planning

A few weeks ago I wrote about counterfactual reasoning–that it is easier to prove something doesn’t work than it is that is does work.  For example, it is hard to prove that coming to school leads to learning, but we know truants fall behind.  This is important to remember as we plan–we can’t confirm much about the future, but we can plan with that blind spot in mind.

The idea becomes important when looking at the surprises we seem to encounter after we do a lot of planning for the next year (placement, schedule, resource allocation, etc.).  The theory of this is the “black swan“; that unexpected events have a major effect on plans.  Because of this, the theory goes, we should plan with the unexpected in mind.  The other half of the theory is that it is difficult to know what to plan for.  In short, just be ready for something.

Note: Much of this is inspired by  Nassim Nicholas Taleb’s Black Swan: The Impact of the Highly Improbable.  Much of this essay looks at the unexpected events that cause those disruptions every educator can relate to–nothing Earth shattering, but annoying nevertheless.

As educators, we tend to lock everything down–protocols, lessons, spaces, schedules, class lists–because we like to think we can predict everything.  For example, our school locked students into the three homerooms last June, but we did not take into account several unexpected developments.

The largest was we had asked students their World Language preference (French or Spanish).  Our students move through the school by classroom–Art, PE, Music and World Language.  During the summer we realized we could not honor most preferences because placement was based on a different criteria.  We punted.  But we also did not expect several students from one classroom to move away or that so many new families would move into our district.  We placed with insufficient information.  Then student preferences for WL created more turmoil, as parents demanded their kid get this or that language (they, of course, waiting for three weeks into September before doing so).  It also took a few weeks to figure out the WL/Tier II needs of new and leaving students.  Who knew?

As the old Yiddish proverb goes, “Man plans and God laughs.”

One of the mistakes people make is in trying to anticipate the black swan.  If we only anticipate more contingencies we think we can control it all.  No.  We certainly can straighten out many issues.  And we still have to make plans–schedules, room assignments, placement, etc.  How, though, can we plan while respecting the unknown?

Here are some examples of common issues facing schools.

Configuration

A few weeks ago I shared a configuration idea for our middle school.  For forever they have been multi-age, but they are contemplating going straight-grade “just for next year” because class sizes are getting too large and students are falling through the cracks.  There are several challenges facing the team, but the main one will be the influx or exodus of students.  I argue maintaining multi-age classrooms is the best safeguard against student numbers changing.

They are adding a fifth teacher to a four person team.  The Core is responsible for Language Arts, Math, Science and Social Science.  Currently, two teachers are licensed to teach STEM and two Humanities.  To alleviate class sizes they plan to hire a Science teacher.  In theory, this lowers the class average from 28 students to 23.  Of course, because the new teacher can only teach Science it does nothing to help the Humanities classes .  In theory, one of the STEM teachers will shift some of her load towards Math, but Math being taught straight-grade and other elements conspire against such neat divisions.

Using the our middle school’s straight-grade plan as an example, three teachers for the 7th and two for the 8th creates an automatic imbalance.  Of course, these can be mitigated (supports, aides, schedule).  But, should five students leave the 7th, or five join the 8th (or both), there is suddenly a difficult situation with no easy fix.  We can’t know if that scenario will play out, or be worse.  The plan, though, is crafted with little wiggle room should any black swan appear.

Using counterfactual thinking, instead create a multi-age made up of two teams.  Each team has a Science and Humanities teacher.  A single Math teacher instructs all students–a big load, but she is relieved of other duties and the two teams can focus on social-emotional middle school programming.

Imagine, now, five students leave the seventh–the release would be divided between two teams, offering a slight relief of two or three on each team.  In the case of the five joining the 8th, a much more devastating event, that burden would be shared and only result in two or three each.  In fact, if both scenarios play out, much of the effect is a wash.  Multi-age benefits can mitigate the other effects of population bubbles.  For example, each year teachers move (with their classrooms) and they need to relearn new curriculum and start new team protocols, etc.  Multi-age allows multi-dimensional flex.

But the question to ask is why that plan will not work?  I have shown the problems of straight grade (there are more, unwritten here), but that does not mean multi-age is better.  The black swan, counterfactual advice is to take each option and poke at it–with some of the more outlandish variables–until it reveals its flaws.  Then, put down other ideas, too, and poke them.  One area to look at is which “losses” are most tolerable and which would be devastating.  The plan with the most resiliency is probably your best bet.

Schedule

The difficulties of crafting a schedule are long and well known (we even know that we don’t know enough).  This is especially true in schools with many grades and diverse needs (we are a pre-K through 8).

I actually learned a lot recently on why crafting a schedule is so hard through an article on chaos theory (an article for later).  In short, such endeavors are dynamic and resetting course causes exponential problems down the path.   What we do know is that schedule craft works up to 80% of the elements are placed, but then conflicts trouble the last 20% of needs (that’s the dynamic nature of the process).  To fix it we begin to make compromises (which creates more conflict, exponentially).  Each year, someone (or many) is (are) unhappy.

Then, each year, we have a black swan surprise.  Last year, it was WL and everything that followed it.  Other grades had other emergencies.  Each fix seemed to bring more problems.  We live with it–we have no choice–but that 20% consumed a lot of resources and continues to create ripples (e.g., it is unclear if student preference for WL will be honored in 7th grade now that they have a year of one language under their belt.).

My solution is blocking out time instead of scheduling in time.  For example, one grade level figures four uninterrupted hours of seven school hours are needed for Core–90 minutes of Literacy, 60 minutes for Math, 60 for Science/Social Science and 30 for TA.  We demonstrate what’s important to an organization by the resources we put towards it (time, money) and four hours uninterrupted seems a reasonable commitment.  That leaves three hours for lunch, recess, Unified Arts (UA), band, mentor, counseling and any other non-Core activity.

How is a block resilient?  Let’s look at how our current system is not.  Predicting a seven hour day is hard; there are many variables.  It is also easy to be fooled by small numbers–three kids pulled out here, four there.  Currently, music lessons, mentor, speech and physical therapy, and other student needs are pulled from Core.  There are many reasons for this–part time employees and mentors have locked-in schedules, no one wants to deny a child their one Art class a week, and a few kids here or there are seen as non-disruptive.  But in my class of 20, one absence and three music lessons are 20% of the class.  Plus, recovery now becomes a negotiation not with two other Core teachers, but with three times that–pull from Art? PE? (And, because of the dynamics of the interwoven schedule, also lose 20% of their students and need recovery for them).  Again, here is the problem I am poking a hole in with our current schedule and how it was created.

By dividing the day into two blocks (4 hour, 3 hour) that do not overlap, students and teachers only have to negotiate within those blocks, with only those adults.  It is more likely I can flex with my Core teammate’s schedule because a) we have an aligned schedule, b) fewer pulls (classes) means fewer and smaller disruptions, c) fewer agents means less communication needed.  On Core, we can debate the virtues of Math vs. Science and time required to do it right, but with blocks we do not have to debate the death by a thousand cuts we currently face.  Four hours uninterrupted shows we value Literacy, Math, Science and Hive.  Three hours should be enough for others to negotiate what is best for students, among actors who have more in common (UA).  Or, to demonstrate that we want more for our students than the resources we have (or are willing) to give.  Again, this may be flawed.  Poke it.

Student Placement

Has there not been an issue with placement?  Do kids every come as advertised, stay friends (or enemies) with the kids from the previous year, or continue to need the same services?  Too often, we’ve even had to change the teacher assigned to the class (illness, pregnancy, number shifts).  For all the benefits we argue placement provides, those are the holes I poke in our locking kids in as we do.

Three years ago, the middle school wisely created two teams with a blank slate for placement.  First, Special Ed (SE) placed students on the two teams with a variety of rationales–service providers, friend/tormentor locations and the like.  This was open to debate/poking by the group, but we deferred to the experts.  Second, a small number of kids who do not receive services were separated or paired because of behavior or emotional needs.  Some kids need friendly peers, while others are a bit too friendly during class time.  All told, about 20% of placement was locked in like this.  That is an example of necessary planning with what is  known (at that time).  What’s important to note is what they did not do–lock in the remaining 80%.

Then, the sending teachers made two “balanced” teams.  More poking.  But the most important part was that no one was locked in until mid-September.  Each week the middle school would observe interactions, listen to students and families, and shift kids around.  Then, they locked in all kids, creating permanent teams and focused on creating identities for those teams and bonds within.

They waited because kids change–they move, they mature and their allegiances shift.  Because services did not have much flexibility (the nature of Spec Ed and part-timers), and some kids need reassurances (i.e., not being placed with a bully or knowing an adult face before the first day of school), those 20% never shifted.  But, as things shook out, a flexibility was built in for the other 80% so that movement was possible.  By the third week, the two teams were as solid as one is going to get.  It was a successful year.

Compare that flexibility with what happened when we locked in the grade last year that suffered at the hands of WL.  For what was gained (a faint sense of knowing their homeroom) much was lost.  While we should be aware of the anxiety we create with some unknowns, it can be fleeting compared to the year-long stress of a bad fit created by the black swan.  Again, the above might not survive a good poke, but our current spring lock-in is troublesome in the long term.

A Note About Counterfactual Poking

It is easy to spot flaws.  The question is a) if that flaw is true in any viable model, b) if that flaw has a major effect.

For example, one argument I have heard against postponing placement until a few weeks into the year is anxiety and parental concern.  Yet, when we locked in the placement of kids, one or two parents would inevitably call with concerns (or be more forceful about it).  Sometimes, that brought a dam break of other requests as parents micromanaged.  In short, parents and kids worry regardless.  The former is easily dealt with because the concerning elements are often soothed with more information, and solutions can be shared (including locking in needs).  The latter upends a carefully planned order, with results (according to chaos theory) that can prove exponential.

So, caution is necessary when weighing options.  People need to be objective that all plans have downsides–sometimes the same downsides–but some recover better.

Conclusion

The idea is that the we don’t know what we don’t know, but we still have to plan.  We want to poke holes in things early, and be honest about defects, before we choose the model that is best for students.

The Follow Up: Interventions Often Need Interventions

9780812973815

There is no silver bullet.  Change requires tweaking and digging until you get to the bone.  Be ready; it’s hard work.

Last week I wrote some idea of looking at data, in part use counterfactual questions.  In a counterfactual assessment you prove what is wrong, because it is easier than proving what is right.  For example, I can’t prove that coming to school daily works, but I can easily show that students who are tardy and miss a lot do poorly.

So, what do we know does not work?  What stops learning?  These questions came while reading Nassim Nicholas Taleb’s The Black Swan: The Impact of the Highly Improbable.  Check out that post (Counterfactual Questioning of Data and Needs) here.

The working method behind counterfactual questioning is to continue to shed what does not work.  As a teacher or leader, I am sure you already know the kids the system is not working for.  And I am sure you know what areas are not working for them.

Why, then, do you continue with your system?

I will bet that the finger points to the kid, the family, or some other factor (poverty, tragedy, punkishness….)–but NOT the system in general.

When we adopt a pedagogy, we do so because we believe it will work.  It’s proven.  There’s data behind it–perhaps, even, from your school!  But, looking at your data now, you know it is not working for all students.

And you adapt.  And that does not work.  Or it doesn’t work for another group.    Or, the system is blamed and abandoned TOO SOON.  What you are doing, or some version of it, probably has many components that do work, but you have to adapt it to the students sitting in front of you.

I used the example of reading in that post, so let me focus on it again.  By putting SSR (Sustained Silent Reading) in the school we can focus more specifically on what does not work until it does (we control the environment, thus making it a nice little laboratory). Is it seating (move them), book selection (look at level and keep trying new titles), fluency ro eye tracking (Google Text to Speech)…. And as we take away what does not work we are left with nothing but reading.

Sherlock Holmes said, “Once you eliminate the impossible, whatever remains, no matter how improbable, must be the truth.”  This is kind of like that–the second half of the process.  Once you stop trying to make your square peg fit the round hole, you’ll find a whole bunch of things that don’t work.  Great.  And as you try more, seemingly improbable solutions, you will find more that don’t work.  Until you hit on one that does work.  Pay dirt.

Start with your solid pedagogy.  You have to.  But, then, measure.  If it isn’t working, figure out why.  Then try something.  You might start with the old toolbox–why reinvent the wheel–but quickly throw the net wider.  As you do, you’ll get a sense of the student and why they aren’t succeeding.  Be Sherlock Holmes and find that improbable solution.

511ngcy9isl-_sx395_bo1204203200_

All of that said, let me recommend Nancie Atwell’s In the Middle: A Lifetime of Learning About Writing, Reading and Adolescents.  In short, she promotes reading and writing workshops: Students work, and she offers mini-lesson and conferences.  Unlike current workshop queen Lucy Caulkins, Atwell’s program adapts, constantly.  The program fits the student.

Don’t worry if reading or writing or middle school is not your focus–the methods she employs works for any subject.  And she is honest about her journey finding these methods–she’s being doing for thirty years and is still refining.  Try to introduce a bit of it into your instruction.

 

 

Counterfactual Questioning of Data and Needs

9780812973815I have been reading Nassim Nicholas Taleb’s The Black Swan: The Impact of the Highly Improbable after reading a Wikipedia article about Stephen Bannon and his influences.  This was one, and it made the news clearer in many ways.

It also showed me a different way of using the data we are now getting almost too much of.  Here are a few interesting take-aways as how I might look at data new–and use it moving forward.

1. Look at the outliers.  When we focus on what works, or decide that it does not work, we often don’t really look at who falls short.  We look at the bulk–the middle kids–and evaluate the program based on them.  Then, we look at how we can “fix” the others to fit into the system which (we think) works.  Or blame them for their shortcomings.

It becomes about fitting a square peg into a round hole.

For some, we intervene with Tier II or even go EST (Educational Support Team), if they don’t already have IEPs or a 504.  For others, we explain it away (often with history–they’ve always done that–or other–family issues, anxiety, a bad day….) which is another version of blame-the-kid.  But if 20% of the group is not succeeding, how can we call a program a success?  (FYI, I’ve seen programs at 40% success declared working, because you have to understand…)  And if the response is that those 20% get something different, a) does that work, b) is that the best use of resources, and c) is that best for all kids?

One concern about any program outside of the normal program is measuring effectiveness of that intervention.  Often, if the intervention works–great work–but if it fails the child we find excuses with the child.  Honestly, once a kid falls to Tier II or Tier III they stay there.  (There is little incentive to change that, but that’s addressed later).  So the first step needs to be coupled with….

2. Tailor the group to the outliers.  While a program should not revolve around a single student, are there changes that would help other students as well?  For example, in the talks about anxiety our school did the presenters sold the idea that such interventions help other students.  True.  Compared to our childhood, we no longer time tests while we do allow movement breaks, reteach, and incorporate a dozen other techniques that began with outlier kids and are now mainstream.  But what is required is….

3. Thinking of needs outside the standard toolbox.  My students, for example, started the year talking about needing “hands on” learning (whatever that means).  Then, it was movement breaks.  Then, social time.  All of these were valid, but they did not really have an idea of what they needed.  Instead, our team read between the lines and gave them what they actually needed.  But even then, our responses were pretty standard (more projects, less seat work).  And, we are having a difficult time with a different outlier group (we are working on working independently, working with peers, time management and other, new issues).

Recently, we have looked at other needs.  We have a few kids questioning their identity (sexual orientation, interests, even their given name).  We have several kids with family issues–really worried about their family, illnesses, moving, relationships and the like.  We can see with after school groups that a desire for identity and connection exists.  A social survey with gave students showed a desire for help with peers and impulse control.  What to do with that?  We are currently looking beyond the wisely constructed groupings to regroup with needs, and soft-sell the focus bringing them together now.

But we aren’t counselors and this is not a therapy session.  Our goal is academics.  Plus, there are other needs and other solutions.  For example, in 2014 I found that what moved scores was participation in a competitive activity (AAU, Mini Metro, Far Post, ballet…).  There are a number of needs, many you don’t see because our students are faces in a specific location and a data point (go out to recess or the common spaces and see them differently).  So, we find that need by….

4. Ask counterfactual questions.  In a counterfactual assessment you prove what is wrong, because it is easier than proving what is right.  For example, I can’t prove that coming to school daily works, but I can easily show that students who are tardy and miss a lot do poorly.

So, what do we know does not work?  What stops learning.  Time is a general issue.  Interruptions are more pressing.  Homework does not really help those we want it to help.  But what the homework issue shows us is that time is needed–to read, for example.  When we do get kids to read at home, they get better at reading.  Now, we don’t know if that is due to attitude or practice or what.  But by putting SSR (Sustained Silent Reading) in the school we can focus on that.  For those who is does not work, why?  Seating (move them), book selection (look at level and keep trying new titles), fluency (Google Text to Speech)….  And as we take away what does not work we are left with reading.  But you need to ask….

5. Instead of asking what they need, ask what they don’t need and shed it.  Does a student who reads all of the time need SSR?  They might like it, or it may serve a calming function, but it does not add much to the academic purpose of SSR for them.  Found time.  They might, though, need that calming function–can it be given more effectively (offer meditation or counseling)?  Or, when their non-needs are whittled away, what’s left?  You can ask this of the student who gets instrument lessons outside of school (why do they still get school lessons?), the kid who plays Mini-Metro during the basketball unit (is relearning how to dribble worth their time?), and so on.  And we know who is lacking–we can tick it off the top of our head.  We have a list ready of what is not working.

A simple example is leadership.  We say we need it.  In our school, we have a group of intelligent athletes who think leadership is getting their way, instead of making their peers look better and rise up together.  They pick the team, have the ball, take the shot, and are ready with the excuse of why X didn’t get the ball (“They aren’t very good.”)  They are teachers waiting to be shown how.  If they teach X, that kid X will get the ball.  A simple pivot around the data sets that up–Kid A is measured on basketball skills and Kid B is measured on his teaching others.

6. And before you talk about time, think about how much time we spend accommodating, chasing down a few kids, and adapting work.  Differentiation is hard and time consuming and still we have failure–because we differentiate the wrong thing.  Use data to find the right thing, but cutting back everything that does not work.

Of course, this is still turning around in my head.  But it might cause you to think about things differently as you approach configuration, schedule and PBL in the next month in anticipate of the next school year.