The young gentlemen in the picture is my stepson. He and my fiancee live in New York City for the time being. Being a long distance away, I have to help him with his homework using Google Hangouts where I can share my desktop screen and work out examples with him together. A Data Analysis project came up, and I thought it would be a good idea to perform some analysis on the New York City Common Core Test Data. Anybody who knows me knows how important Education is to me. I was excited to find out that New York City has open test data on their Common Core tests results for students from Grade 3-8 in Math and Language Arts. You can download the full report from my Github repo, but I will give some highlights here. I performed this analysis using the R statistical language and the RStudio IDE. It is worthy to note that when performing analysis on data, you must be careful that the conclusions that you draw should not be biased and should be based on domain experience so that you are able to understand the bigger picture. Some of the conclusions that I draw are due to my domain knowledge that I have gained due to my numerous visits, as well as the conversations that I have had with New York City natives who are family and friends. One of my strongest skills is that I am a great listener. While listening, I am constantly processing the information and trying to paint the big picture. I should note that I have not visited Staten Island as I do not know anybody from there, so I do not have any domain experience on that borough. Let us start with a breakdown of a graph generated to analyze the results across boroughs:
To give a brief explanation about this graph, this is what is called a boxplot. Boxplots illustrate what is called the Inter-quartile range(or IQR) of the data. The IQR highlights the "inner" 50% of the measurements which is between 25th percent quartile and the 75 percent quartile. To bring this definition to context of our graph, let's look at the boxplot for Manhattan. We can see the lower bound of the box lies around 25%. This means that 25% of the schools in Manhattan had a student proficient(pass) rate of 25% or below. The black line in the box represents the median or 50 percent quartile. Therefore, In 50% of the schools in Manhattan, the number of students who met the common core standards was 50% or below. The upper bound represents the 75th percent quartile, thus for 75% of the schools in Manhattan, the number of students who met the common core standards was around 75% and below. Hopefully this explanation was good for you to understand what is going on in the graph. I would say the most important information is the black line since it illustrates the midpoint of your data. As you can see the midpoint across boroughs lies at various percentages for each borough. Based on the data, Queens has the highest percentage of students who are meeting the common core standards given that 25% of its schools have a student pass rate of 35% or below. Another way to look at this is that 75% of the schools in Queens have a student pass rate of over 35%. Staten Island came in a close 2nd to Queens. Manhattan has a wide IQR compared to all of the other boroughs. When I saw this, I thought about how the Manhattan borough is divided where you have the upper part of Manhattan (Harlem) which faces some of the same socioeconomic challenges as say the South Bronx may face, and the lower part of Manhattan where those economic challenges are not as prevalent. I wanted to visualize this by overlaying the data for each school onto a Google map, but I was having troubling getting the geographical data points for the schools (I'm not that good at working with ESRI and Shape files yet). I'm curious if my assumption is actually correct. Maybe in another blog post I suppose.
The borough I was most interested in was the Bronx since this is where my stepson goes to school. By looking at the graph, it is to no surprise that I was not elated by the results that I found. 75% of the elementary and middle schools have a pass rate of 56% or below. In other words, almost half of the students at 75% of the schools in the Bronx are not meeting the Common Core standards. To me, this screams "crisis" for the Bronx. Could you imagine telling the students at the 195 schools : "Look to your left, look to right... One of you are going to fail the Common Core Test this year.."? If you have been to the south Bronx you will know that there is extreme poverty, which is coupled with high crime and violence. If it is expected for any change in this cycle, these rates are going to have to improve dramatically. I'm not saying that I have a solution for it, but those are just the facts.
I was curious to see the correlation with the school funding and the pass rate, but I was unable to easily retrieve the school funding data. Hopefully this would be placed in a data set in the future for ease of use. I wanted to know if these schools are receiving the adequate funding necessary to improve their conditions. From listening to my family and friends, most of the failing schools are being closed rather than receiving extra funding. After reading Saving Captialism: For the Many, Not the Few, I found out that a majority of school's funding comes from the property taxes of the residences in the community. I'm unsure if this is case in New York City, but if this is so, then this spells "doom" for those struggling communities.
The graph above shows the trends of the proficient rates from 2006-2015. As you can see, there was a increasing trend from 2006-2009, then there was a sharp fall in 2010. This was followed by a steady increase and then a sharp fall again in 2013. There was a modest increase from 2013-2015. Let me explain the two dips in 2010 and 2013. The first abrupt fall was due to a change in the scoring metric which determined whether a student met the standards or not (the bar was raised). The second abrupt fall was change in the test to align more with the Common Core standards. The best year for NYC students was 2009 where the worst was 2013. It seems that there needs to be a process change at how changes are incorporated into the tests as this change generated a lot of noise in the trend given that the students were not prepared for such change.
The graph above shows how the students' passing rate trends by race. I was very intrigued at how tightly connected the trends of Black and Hispanic students were, and how tightly connected the trends for the White and Asian students were. Note that this is independent of the location. There were so many questions that were racing through my head when seeing this:
- What is the demographic breakdown of the schools that White/Asian students attend in high numbers?
- How are the Black/Hispanic students performing in majority White/Asian schools?
- How are the White/Asian students performing in majority Black/Hispanic schools?
- Do the White/Asian students interact/study with the Black/Hispanic students?
I have the data to answer the first three questions, but I'm more interested in the fourth question. I want to know are their programs setup where this type of knowledge sharing can transpire; where students from various schools can congregate and study, regardless of economic background or race. I am aware of the benefit of this "cross-pollination" of thought process and understanding as it was crucial in my studies in Computer Science. This may be the California hippie speaking as I'm aware of the challenges that presents itself in this kind of task (I'm from Atlanta. I'm far from naive), but I would be interested in seeing such experiment done on a small scale. I always find myself learning things from people from various walks of life, and I like to share my new-found knowledge with people in my community to give them different perspective on things. I like to think of myself as a spy whose task is to infiltrate knowledge bases from people across the world.
The next graph was generated at my stepson's request. He wanted to know how his elementary and middle schools were doing so I generated some data for Baychester Academy and Pellham Gardens Middle School. I only had data for the last three years for these schools so I plotted that information. I will breakdown the graph before giving my conclusions.
The data points represent data gathered based on Gender from 2013-2015. The black line represents the average(also known as mean), and the red line represents the median of the data points.
The data does not show a bright picture for either school. The average number of students that are meeting the standards is around 30% for Baychester Academy and between 10-15% for Pellham Gardens Middle School. This is extremely worrisome and means that I will have to make sure that I continue to work with him (maybe even more now) to assure that he is gaining the knowledge that he needs.
Now I'm not the one who uses tests as a "end all be all" measurement for one's intelligence or success. I'm personally not a fan of the amount of emphasis put on standardized testing in the public school system. I believe learning should be more free form and more emphasis on teaching students how to think and how to arrive to solutions. These skills I find more beneficial and more useful in a society. But I also understand the need for a standard metric to determine how well the students are doing across different schools and grades to see if progress is being made. I hope someone reads this and comes up with a solution to address the challenges found in this data analysis. While writing this, I was attending a TreeHacks hackathon competition and I was excited to see students come up with hacks related to Education and improving Math skills as well as Social Good. I'm hoping that someday, those ideas will be incorporated in our learning system to improve the education for all.
