In this lab, I sought to better understand NYC Public School graduation and drop out rates, the way they change over time in different schools, and the connections between them. I used the dataset Graduation Outcomes- Classes of 2005-2010 (School Level) from NYC Open Data. This dataset records the various graduation outcomes from public schools that accepted incoming students between 2001 and 2006, expecting graduation within four years. The data lists the number of students per cohort for each school and the number of students in that cohort that graduated, received a Regents Diploma, received an Regents Diploma with Advanced Designation, received a Regents Diploma without Advanced Designation, received a Local Diploma, were still enrolled, and who dropped out. While the dataset includes many demographics for the cohort bodies (race, gender, learner type, etc.), I focused on the Total Cohort demographic, which lists all students enrolled that year. All visualizations were created using Tableau Public 9.
To prepare, I looked at three different visualizations related to high school graduation and drop out rates. The first is SchoolBook: Public High School Graduation Rates, published by WYNC. This interactive line graph shows the average graduation rate of NYC public schools from the Class of 2005 to 2014. Users can also select one school to compare against the citywide average. While this graph is useful, it has some limitations. It only allows the user to compare one individual school at a time against the average, it contains no information about the students who did not graduate on schedule, and it only lists schools that are currently open, meaning any data from schools that closed before 2014 is not visualized. The second is The Status of NYC Children: Drop Out Rates, published by the Citizen’s Committee for Children. This interactive visualization allows users to display drop out rates by School Year from 2005-2014 as a table, bar graph, or line chart. Users can also select what demographic they are interested in viewing, as well as location (ranging from all of NYC to School District). Like the previous visualization, this does not allow the user to directly compare graduation and drop out rates. Also, it doesn’t allow analysis of individual schools. The third is Bay County, MI: School Graduation and Drop Out Rates, published by Michigan Live. This visualization compares the 2010 graduation/drop out rates of all schools within the county, the state graduation/drop out rate, and the national graduation/drop out rate. While this visualization does let users compare graduation and drop out rates for each school, it only does so for one year.
Before creating the visualizations, I cleaned up the raw dataset in Google Refine. Initially, I removed the recorded percentages, the s-es used to indicate that no data was recorded, and the columns representing the two types of Regents Diplomas. I also created two columns, one for graduation outcome type and one for number of students who achieved that outcome by stripping the columns down to the numbers that added up to the total cohort and using the transpose feature to create a more workable data structure. Once in Tableau Public, I filtered the demographics to just the Total Cohort figures and filtered out the 2006 August cohort, composed of students who graduated in June and August of 2010, as no other cohort recorded August graduation outcomes. I also grouped the number of students with Regents Diplomas and Local Diplomas together into a “Graduated” category, as the total graduation rate was what I wanted to analyze.
I set out to create four visualizations. Two that showed the 2005-2010 graduation outcomes of the five schools with the highest 2001 Cohort drop out and graduation rates, and two that showed the 2005-2010 graduation outcomes of the five schools with the highest 2006 Cohort drop out and graduation rates.
To determine which schools to look at, I created visualizations that showed the 2001 and 2006 Cohort data for all of the schools. I produced a bar graph, with the Number of Students on the x-axis (Column) and the Cohort and School Name on the y-axis (Rows) Rather than look at the raw number of students who graduated, dropped out, or were still enrolled for each school I calculated the percentages for each status. I also showed the mark labels so the exact percentage was viewable. Unfortunately, I was unable to reorder the bar graph to show which schools had the highest graduation rates and drop out rates, so I had to scan through the visualization to determine which schools to focus on. These visualizations are viewable here.
Once I finalized which schools I was looking at, I created bar graphs, with Cohort on the x-axis (Column) and School Name and Number of Students on the y-axis (Rows). All available data from the 2001-2006 cohorts was represented for each school.
Out of the five schools with highest drop out rates in 2005 (2001 Cohort), only Liberty High School Academy for Newcomers, with a drop out of rate of 81.67%, was still open by 2010. As data is no longer available, we can assume that South Bronx High School closed after 2005, and William H. Taft High School, Theodore Roosevelt High School, and Park West High School closed after 2006. This is possibly due to a push by the Department of Education to close failing schools, fold them into new schools, or break down large underperforming schools into several smaller schools. Liberty High School exhibits a sharp decline in drop outs between 2005 and 2006, and while graduation rates stay consistently around 50% through 2010 there is also an increase in the number of students who stayed enrolled. The schools with the highest drop out rates in 2010 (2006 Cohort) had lower drop out rates than the schools with the highest drop out rates in 2005. However, graduation rates and rates of continued enrollment had stayed consistent since 2005, and did not fluctuate the way they did in Liberty High School. From these visualizations, we see that underperforming schools with high drop out rates are likely to close or to not significantly improve over time, with the exception being Liberty High School.
Schools with high graduation rates in 2005 (2001 Cohort) stayed consistently high through 2010 (2006 Cohort). The only outlier is George Washington Carver High School for the Sciences, which had a graduation rate of 100% in 2005 but dropped to 71.8% by the following year. The graduation rate continued to drop and then rose again by 2010. Even while graduation rates changed at George Washington Carver, the drop out rate never rose higher than 10.8% with the majority of non-graduating students remaining enrolled. Schools with high graduation rates in 2010 (2006 Cohort) had also been consistently high since 2005 (2001 Cohort). Again, drop out rates are low, with the majority of non-graduationg students remaining enrolled. From both of these visualizations, we see that schools with high graduation rates remain high performing over time, and that even when students do not complete their degree within four years that they are more likely to stay in school than drop out.
To better understand the drop out and graduation rates, future visualizations should focus on analyzing other available demographics, such as race or learner type, to see if drop out and graduation rates are consistent throughout those demographics or if they vary. It would also be helpful to further research Liberty High School Academy for Newcomers to discover what changed in the school that so dramatically changed their graduation outcomes and determine ways to apply those methods to other underperforming schools. In that same vein, using this data, other schools that changed significantly, both for the worse and the better should be identified and further researched to determine what exactly caused that change. It would also be helpful to look at graduation outcome data from 2011 to 2015 (2007-2011 Cohorts) to see if graduation outcome trends continue in the nineteen schools identified in these visualizations, or if they begin to vary. Throughout this research, special attention should be paid to important events in education such as the introduction of the Common Core, as major curriculum changes and education initiatives may impact graduation outcomes.