Rich School , Poor School

Is there any question that every child has the right to the same education? I would say that education is a human right. When we talk about human rights we talk in terms of every child having equal access to a quality education. People from all sides of the argument can rally around that last statement. One side of the debate can argue that charter schools are the answer. When families have choices and competition exists, public schools will rise to the occasion. Another side will say that teacher quality is at an all time low and that is especially  relevant in low achieving schools. Obviously there are more than three sides but another legion will claim that a consistent, rigorous set of standards such as CCSS will equalize the education that every student receives.

Just so we’re clear, the people above fall into two categories: either they have have no concept of what education is like for some of our poorer citizens or they are outright liars with some kind of profiteering agenda.

According to Do Something, students who live below the poverty line have higher absentee rates, higher dropout rates and lower rates of post secondary education. In a post I wrote several months ago I compared PSSA scores to school district aid ratios. That research showed that the richer the school district is, the better chance that students will reach proficiency. That doesn’t really sound equal.

I can’t speak of all states but in Pennsylvania the funding inequities are tremendous. I know that I’ll hear that poor schools receive way more state funds than rich schools.No kidding. Spend more than a millisecond trying to figure out why that is. Poor schools, especially poor rural schools, have a very limited tax base. A limited tax base means poor local funding and I can assure you that the state does not make up that difference. That of course leads to districts surviving on a shoestring. Cutting positions that lead to larger class sizes. Cutting positions that make a difference to student achievement: literacy coaches, math coaches, reading specialists, counselors and paraprofessionals. According to a study by the Pennsylvania Education Law Center, the highest poverty school districts spend $75,000 less per classroom than the wealthiest schools. As in many other areas of our country, the privileged rich continue to flourish as the poor continue to languish.

As a human right, education should be of the same quality no matter your demographics. Students should have access to the same resources no matter their economic status. Education should aim to raise the achievement of every child not just the rich, powerful and politically connected.

What we have in Pennsylvania and across the nation is not an achievement gap, what we have is a funding gap. Rather than billionaires spending there money on influencing education policy, what if they used some of that money to equalize the funding in our impoverished urban and rural school districts.

Can Money Buy Proficiency?

A week or so ago I published a blog post boldly stating that you could probably sort districts by their aid ratio from high to low and get a pretty accurate projection of how their PSSA reading and Math scores would look.   One of my loyal readers, Rogue Anthropologist, inquired about whether any such research was available.  Being interested in statistics and more so probabilities, I set out to determine how accurate my blurted out hypothesis had been.

What I found was not an exact, one to one correspondence but it is pretty telling. Using 2012 PSSA scores in reading and Math and district aid ratios (AR) for 2012 I was able to get a pretty good picture of district scores in relation to the economic status of the community.  I do want to go on record as saying this isn’t a statistical analysis.  It is a collection of facts based on the data.  I also want to clarify that when I talk about making Adequate Yearly Progress (AYP) I am talking about what we call in education “making the number.”  In other words I did not take into account whether districts made AYP using confidence interval, Safe Harbor, Growth Model or a combination of any of the above.  I simply used 78% for math and 81% in reading – “the numbers.”  Also, I will note that I used only district totals not individual schools.  For example, Central Dauphin School District is lumped together although it is made up of many different schools including two high schools.  As well, the Philadelphia and Pittsburgh school systems were grouped as one district.

I took all of the school districts in the state and order them from lowest AR (.1500) to the highest (.8814).  I then divided the entire lot into quartiles.  There are 500 school districts in the state of Pennsylvania.  I eliminated one, Bryn Athyn, because they don’t actually educate any students in their district.  Look it up!  It’s pretty interesting.  So, of the 499, since I was looking primarily at the top and bottom quartiles, there were 249 in the middle two quartiles and 125 in the top and bottom.

In the area of Math statewide 62% of the school districts made AYP.  That is important to know as a yardstick for the rest of the data.  School districts in the top quartile according to their AR made AYP 90% of the time.  Conversely, schools in the bottom quartile only made AYP 26% of the time.  That means that a rich school district is about three and a half times more likely to make AYP than a poor one.  To take it a step further, the schools in the top two quartiles – I gave them number 250 – made AYP 80% of the time while the schools in the bottom half made AYP only 42% of the time.

On the Reading PSSA only 28% of the school districts statewide made “the number.”  That seems to have really skewed the numbers based on AR.  In reading only 2% of the schools in the bottom quartile made AYP.  Of the top quartile school districts 70% made AYP.  Some of this in my professional opinion can be attributed to poor households being more apt to be language-poor and those students tend to come to school with fewer background experiences.  When you look at this information as a 50/50 split, the bottom half of school districts only made AYP 8% of the time and the top half made AYP 48% of the time.

There are of course exceptions to every rule.  Windber Area School District and Cambria Heights School District made AYP in both subjects despite ARs of .7324 and .7241 respectively.  Inversely, 14 school districts did not make AYP in either subject even though their ARs ranked in the top quartile.  Pequea Valley School District had the lowest AR of any school not to make either number; .2554.

In addition to economic indicators, I was also interested in determining other factors rather than the quality of teaching that may determine performance on the PSSA.  Research has been done about how the education of the mother impacts achievement.  When I look at that statistic I tend to think that it leads back to the affluence of the family.  Students in privileged circumstances would more likely have two highly educated parents than students living in poverty.

One thought that I have heard, and this is probably do to the poor performance of Pennsylvania’s urban areas, is that the percentage of minorities in a district has an impact on achievement and PSSA. The data on the percentage of minorities in a district is interesting but not very conclusive.  In the area of Math school district that have 10% or greater minorities were proficient 50% of the time (remember that the entire state only had an AYP rate of 62%).  On the opposite end, only 57% of the schools with 2.5% or fewer minorities achieved the 78% threshold.  In Reading the discrepancy is greater but weighted towards the districts with a higher number of minorities, with 33% of school districts with 10% or more minorities making the 81% necessary for AYP and the school districts with 2.5% or fewer minorities making AYP only 13% of the time.  Bear in mind that statewide only 28% of school districts made AYP in the area of Reading.

At the behest of my superintendent I also looked at scores based on the size of the school district.  This is a tough measure when used for measuring the entire district.  The reason being is that the statistic does not take into account the size of individual schools.  For instance the Pittsburgh Public School System, the second largest in the state, has nine high schools to serve 6000 students and the School District of Philadelphia, the largest school district in the state, has 55 high schools to serve roughly 38,000 students.  With those kinds of numbers it is possible to have some high school s doing very well and some high schools doing very poorly.  Anyway, here’s what the number say:  When broken down into quartiles, the top quartile – the largest schools – were proficient 45% of the time in Reading and 71% of the time in Math.  Thirteen percent of schools in the fourth quartile were proficient in Reading and 50% in Math; much to the chagrin of my superintendent.  I’ll have some good news though for him as our district broke many of the rules.

At this point it looks like rich, white – but not too white, large schools have a decided advantage in my personal; race to the top.  Let’s look a little closer though.  Where exactly are the schools that are making it and does that have any impact on their goal of Proficiency?  Well, I’m glad you asked.  There might just be a link.

The National Center for Educational Statistics breaks down districts into categories based on where they are located, how close they are located to a city and the size of the city within the nearest proximity.  Following is the data that I collected on those categories:

  Example Number #Proficient-R %Proficient-R #Proficient-M %Proficient-M
Small City Reading 13 2 15 3 23
Mid-sized City Erie 2 0 0 0 0
Large City Pittsburgh 2 0 0 0 0
All City   17 2 12 3 19
Suburb of large city McKeesport 165 78 47 107 66
Suburb of mid-size city Harbor Creek 21 8 38 18 86
Suburb of small city Pottstown 19 12 63 7 37
All Suburb   205 98 48 132 64
Distant Town Huntingdon 36 0 0 11 42
Fringe Town Brownsville 74 16 22 47 50
Remote Town Dubois Area 10 1 10 6 60
All Towns   120 17 15 64 58
Distant Rural Twin Valley 85 9 11 47 55
Fringe Rural Yough 77 19 25 48 62
Remote Rural Galeton 12 0 0 5 42
All Rural   174 28 16 100 57

As you can see from the data, it is definitely a benefit to live in a suburb especially the suburb of a large or mid-size city.  Those two categories along with the All Suburb categories were the only categories to have a higher percentage of district proficient than the state averages listed previously.  The worst place to live and therefore go to school is obviously in the city with the lowest percentages of all groups.For more information on what each of these categories mean see the NCES website.

With a little dime store analysis you can determine why the suburbs do so well.  Or at least generate an additional hypothesis.

For the conclusion of my hypothesis, without formal statistical analyses, I would say that I was pretty close to accurate.  More affluent school districts are almost three and half times more probable to be proficient on the Math PSSA than their poorer counterparts and more than 35 times more probable to be proficient on the Reading PSSA.  Put another way:  if you live in the top quarter of the wealthiest district in the state your child’s school will have a 9 in 10 probability of being proficient in Math and a 7 in 10 probability of being proficient in Reading.  In opposition, districts that comprise the poorest quarter of all Pennsylvania districts will have a 1 in 4 probability of being proficient in Math and a 1 in 50 probability of being proficient in Reading.  As AYP expectations go up in 2013, to 91% in Reading and 89% in Math, look for those numbers to change.

False Proxy + False Proxy = Your Life

Inspiration to write can come from a lot of places. For me it comes quite often from Seth Godin‘s blog and a friend who goads me into connecting his work to education.

Today Mr. Godin blogged about false proxy traps. You can check out his blog for details. In a nutshell a false proxy is when a someone measures a component of something that is difficult to measure in order to justify the entire product. Good example: measuring the quality of a police force by how many people are put in jail. This measure would not take into account that a good police force may limit crime by there mere presence or that they are exemplary at solving problems. Crazy example: measuring the power of the Republican Party by watching Fox News exclusively.

Everyone may not agree but the forced high stakes testing required by NCLB is just such a trap. The idea of the testing program is to determine the quality of a school and its staff. Make no mistake about it. These tests, differently labeled in each state, were never meant to test the knowledge of students. The false proxy comes in when we try to take one test, administer it to thousands of students, and then compare them across a wide breadth of cultures, economies, and immeasurable demographics. My guess is that a district’s aggregate PSSA score can just as accurately determine the median income of the school’s coverage area as it can the success of the school. They could also pretty accurately determine the number of parents who attend parent conferences. The first thought would be easy to prove. Take every school and list them from high to low based on aid ratio (market value/personal income) and then make another list and sort them from low to high on district average PSSA score. I’d be willing to bet there is a high degree of comparability. It’s all public knowledge; give it a whirl!

So, I think we have shown pretty accurately that the PSSA is a false proxy for determining the quality of a school. Don’t get me wrong; some teacher’s should find a new career path. But I can compare scores of teacher’s that I work with who have abilities that are across the board in terms of quality instruction and the one’s that have limited skills have students who do just as well as the distinguished teacher’s students.

Second false proxy: The new Pennsylvania teacher evaluation model. This is even simpler. Charlotte Danielson developed this model to assist in improving the quality of teaching. Never, and the company developing the evaluation tool for Pennsylvania has admitted this, did she intend for the rubric to be diminished to a number. Statistically speaking, you can’t take a measure that is qualitative and quantify it. That is, however, what the Pennsylvania Department of Education intends to do. A tool built to determine the strengths and weaknesses of a teacher and guide him or her to being a distinguished educator will be used to measure his or her effectiveness.

Not only will it water all of this high quality information down to a single number but that number will count as 50% of a teacher’s – and eventually an administrator’s – annual evaluation. Throw in that another 15-30% of the annual evaluation will be determined by PSSA scores and you have a conglomeration of false proxies and statistical fallacies. Goog luck! Two years of low scores and poor observations or probably two years of average observations and average PSSA scores and you may be looking for a job – and me too!