Every year teachers see presentations of standardized test data. Most are vague and initially incomprehensible, even if one is familiar with statistics. After finding an open data set from the NYC DOE on standardized math test scores, I chose to hone my focus on 7th grade math test scores. These tests are scored on a scale of 1-4; 1 and 2 being below average and 3-4 being average and above average respectively. I selected this case study, because it is both a significant testing year for NYC students at large and impacted my user feedback group as they are current middle school teachers.
Initially, the data set was quite large (over 500,000 rows) and it was challenging to know where to begin. My initial work demonstrates this problem – overwhelmed by column names with multiple uses (borough name, city, and DBN (District Borough Number) were all together originally), it challenged me to see where and how the data set needed to be sectioned. In order to map the data, a second data set was added with School Name, Latitude and Longitude coordinates.
Figure 1: Initial Mapping of the Data Set – 2013, Organized by District
My initial attempt with the data was a challenge technologically. The data had a spatial connection, but I wasn’t sure at first how to show this. I chose to use ArcGis Pro since it has layering features which enhance the data. After showing my graph to a couple of educators, they noted that the color scheme was confusing (and the breaks arbitrary) and since it was organized by district, the dots for each school location felt extraneous. Both agreed that having something interactive would enhance the map considerably.
On a second pass, it felt improved and more intelligible for everyone. The dots gone and the colors more crisp helped tell the story more clearly. However, my final result felt a bit over-smoothed – was all of Staten Island truly on the same level? Knowing the borough & district distinctions, while helpful, may have over/under estimated smaller grouped units – such as individual schools.
With this in mind, I decided to look at it again from a school by school perspective, but with an interactive map, rather than a choropleth map (seen above). This time my goal was to compare school by school and also to see what gaps were in the data set.
Figure 2: Revised Visualization of NYC DOE 7th Grade Data – 2013 Standardized Score Average by District
I knew that the data set demonstrated a spatial relationship – from an analysis perspective – hotspots might occur and it’s easier to look at social relationships that often impact schools when the data is arranged on a map. For a lay person, also having a map view is helpful to find their school that way. When we exchanged feedback regarding this project, the first thing asked was, “Can I find my kid’s school on the map?” She also inquired about seeing data for surrounding schools to compare with her daughter’s school.
For this project, I chose colors from the DOE’s color palette (orange (#F89838), two shades of blue (#6599CD & #003366) and two shades of green (#336600 & #80B44D)) when I could. One map required a few extra colors and I added in a dark brown and black, as those seemed to not detract from the other colors in view. The orange was used to signal troubled areas with blue and green signalling more positive change.
Click on the image below for an interactive experience.
Figure 3: Dashboard Visulization of School Data from 2013-2023;
After mapping this data, I realized that certain data was ill-reported or even not reported at all. To show this, I decided to make a histogram plot of the lowest 20 schools on data that was reported on the data set. This also could have been constructed with a choropleth map to see missing data portions of the map. I used Tableau so that it was interactive. Click on the image below for an interactive experience.
Figure 4: Dashboard of Individual Schools that Reported Low Numbers/Missing Data
Finally, I wanted to see how the school numbers were reported each year. I constructed a map so that it would be easier to compare schools that reported data to this data set. I used a light/dark color scheme so that you could easily compare every two years of data. I also again tried to be consistent with using DOE approved colors.
If I were to continue this project, I would add more social demographic information and also add more layers within the system. I would also try and communicate with the DOE for a more complete data set. Click on the image below for an interactive experience.
Figure 5: A Map Showing How Data Was Inconsistently Reported Each Year – Click for an Interactive View
Overall, a takeaway for me would be figuring out who the audience is for your visualization. A choropleth map feels more helpful at a glance vs. an interactive map feels like something more usable for more detailed information. If I were to use this for social change, having a platform that felt user friendly instead of overwhelming would be my first move. This semester also taught me about intentionality with choices and to aim for reducing the confusion on the part of my audience.