First off, I think that a data visualization seminar is an incredibly important and relevant course, and it is worth continuing, especially when considering the shift of print journalism to online media. Since there is so much value in data visualization, I think it would be more effective to expand the seminar rather than shrink it into a 3-week module in an existing course. As a fan of the visualizations produced by The Economist and the New York Times, I began the seminar too ambitiously, and thought I would be able to produce similar visualizations. My lack of coding knowledge was definitely a source of frustration. Thus, from my personal experience, I think that option 3, expanding the course with more content and coding, is the best method of redesigning the seminar. This would enable the students to already have a set of tools to create visualizations before working with a community partner. I felt in the first few weeks of the seminar that I couldn’t really begin a fruitful relationship with my community partner because I had very limited data viz skills at the time.
Though, community partners should definitely remain a component of the course, because it offers a real world opportunity to put the skills you have learned to work. At the same time, more emphasis on coding might seem daunting to non-techie people. However, it is probably better to struggle through class with others than struggle alone. Therefore, it might also be beneficial to pair up students and have them work together with a community partner. Finally, I think that the talk from Alvin Chang was helpful in that it enabled me to take a step back from the more technical aspects of data viz, and remember what is the objective or the issue is that I am trying to address through my visualization, and how can I tell an effective story with it.
Currently, all of the visualizations I have made are GitHub repositories that can be forked (shared) with others. The repositories are titled as follows: CAPTMap, CMTMap, CAPTProficiencyChart, CMTProficiencyChart, and CohortChart. After creating a GitHub account, these repositories can be easily accessed by searching for my username marissablock23.
The underlying data was provided by Achieve Hartford and the data is already public. I did not make many changes to the data that AchieveHartford provided me, so they should easily understand my spreadsheet, though I cleaned it up a bit. I saved the data which as Google Spreadsheets and then created the maps using Google Fusion Tables. The maps are also available publicly entitled CAPT_Proficient and CMT_Proficient.
Finally, in terms of updating the visualizations for the future, one does not need to have a significant amount of coding experience to update the maps and charts. However, enhancing the visualizations will require more time and effort. With limited coding experience, I could not include some of the more interactive features that I would have liked to.
Cohort data follows the same group of students from year to year in order to track their achievement gains and/or losses. This data tracks a group of students beginning in Grade 3 in 2007 until Grade 8 in 2012. Looking at the cohort of all Hartford students, the percentage at proficiency increases from Grade 5 to 6, but then gradually declines in Grade 7 and 8. In looking at all the schools, it seems as if the general trend is that percentage at proficiency peaks around Grade 6.
Another important takeaway from the graph is comparing magnet school performance to Hartford overall, and comparing Hartford residents of those magnet schools to the magnet school overall. In general, the magnet schools tend to be above the Hartford average. When comparing Hartford resident magnet school students to the rest of the magnet students at that school, a lower percentage of Hartford residents tend to meet proficiency. However, when comparing Hartford resident students of magnet schools to students of district schools, a higher percentage of magnet school students tend to meet proficiency.
The maps below offer a spatial distribution of the schools in Hartford by school zones. The schools are coded (red, yellow, or green) based on the percentage who met proficiency in 2013. The maps aim to answer the question, “Do certain zones have better school options than others?” The map of schools based on CMT proficiency seems to indicate that Zone 1 (top left) has the most school’s with higher achieving students. The CAPT is more difficult to interpret because it contains a small number of schools. The CAPT is taken during 10th grade, so the map only shows high schools. Based on the map, there does not appear to be very much middle ground between performance levels. Furthermore, in viewing the graph that tracks the progression of proficiency from 2007-2013, all the magnet high schools are above the Hartford average for every year between 2007-2013. All the district high schools are below the Hartford average, with the exception of Bulkeley in 2009 and 2010.
CMT Proficiency Map
CAPT Proficiency Map
As a result of school choice programs, Hartford students are not limited to simply attending their neighborhood school. Students can apply to attend any district school within the HPS system, or participate in interdistrict choice, which includes magnets, charters, or district schools in the Hartford metropolitan region. Since Achieve Hartford! focuses on education in the HPS system, the former choice program will be the most important.
In conjunction with Achieve Hartford’s report detailing the results of the Connecticut Mastery Test (CMT), the following visualizations enable viewers to grasp the most salient results by amalgamating the series of achievement scores into one place. Though most of the schools are HPS district schools, there are a few magnets and one charter (Achievement First), which are all located within the boundaries of Hartford. It is clear that achievement, based on the percentage of students who met proficiency (3 and above) on the CMT has steadily increased since 2007, though in the last two years (2012 and 2013) it has leveled off. Now, if a Hartford parent would like to send his or her child to a different school within Hartford, this chart will facilitate making an informed decision. The parent can choose any number of schools to compare with each other, or with the Hartford average. As seen in the chart, the magnet schools tend to perform above the Hartford average. Magnets typically have specialized curricula, and they are designed to draw students from across districts. High achievement may stem from the peer effect, so lower-income Hartford students are benefiting from learning side-by-side with higher-income suburban students.
The performance of district schools is more difficult to generalize and there is no clear pattern. On the whole, most district schools do not seem to exhibit a steady increase in scores, but rather fluctuate from year to year.
If parents are unfamiliar with the many school options within Hartford, a map is more relevant to helping them make an informed decision. This map incorporates the achievement data from the graph, but displays it spatially. Thus, parents can easily see the distribution of schools based on achievement throughout Hartford, and type in their address to identify schools nearby. When the data are displayed spatially, it seems that Zone 1 (top right) has the best school options, several of which are magnets.
The first map below shows the four school zones of the Hartford school district, which are shaded based on the average percentage of CMT proficiency of the schools in the that zone in 2013. The averages range from 50.4 to 74.09, so in the first map the red shaded areas are 50.4-62.23 and the green shaded areas are 62.24-74.09. This splits the zones into either good or bad, which is obvious in the map. Additionally, the use of red also further highlights the negative performance of the school zones. In shading an entire school zone in red, such as Zone 2 (top right), viewers might react negatively towards all the schools in that zone, even though one point (Capital Prep) is green, meaning the percentage at proficient is 70% and above.
The second map shows the exact same data, but rather than dividing school zones into good versus bad, zones are shaded as a gradient, so identifying the worse performing school zones does not quite jump out at the viewer as much as the last map. Again, the range for the gradient is 50.4-74.09, and as the shading gets darker the percentage at proficient increases. In this map, the viewer can determine that Zone 1 (top left) seems to be the darkest shaded, and this is confirmed by the yellow and green points, which are schools with percentage at proficient of 51% and above. In terms of the other zones, it is not quite clear which are better performing. For example, it does not seem that Zone 2 (top right) and Zone 3 (bottom left) are any worse than Zone 4 (bottom right), whereas in the first map, Zone 2 and 3 were shaded red, while Zone 4 was green.