Assignment 12: Redesigning the Seminar

Posted on

Among the three curricular models, I think the last option is the most potential one for redesigning the data visualization course. The idea of separating the course material into two halves is necessary since the students can learn more and be proficient with different techniques of data visualization before meeting with their partners in the second half of the semester. I suggest adding more homework assignment for each class, so students have chances to practice designing different kinds of visualization with simulated datasets. It will help to build up confidence as well as experience before students meet their partners. Another reason for this option is this course structure making students to work as a same pace. Judging from my personal experience, Minh Anh and I employ more techniques from the first half of the semester for our Lottery Winning project such as “creating thematic maps with Google Fusion table” etc. Therefore we had more work to do from the beginning and not much work in the end. Nevertheless, some students in our class have fewer things to do from the first half and have to work so hard in the end of the semester. Last but not least, a 1 – credit course means students have more time to enforce course materials through several classes meeting per week. For me the disadvantage for “0.5 credit seminar” is meeting once a week, which takes a quite to refresh my mind what we had learned from the last week seminar. The only con for the last option is the fulfillment of Numerical & Symbolic reasoning requirement. Even though data visualization, especially with scatter plot, requires some basic understanding of econometrics, the course material may not be enough to meet this distribution compared to the other mathematical classes.

Curricular models 1 and 2 are not as effective as model 1, but they also have some advantages and disadvantages that should be taken into consideration. Although the cons for option 2 are listed above, a current 0.5 credit seminar combined with 0.5 credit internship may be a great selection for seniors, people who are busy with applying for graduate school, writing thesis, looking for an internship or simply need more time to work independently outside class. Therefore it depends on people who are interested in taking the seminar next semester, instructor can be flexible with the time schedule for the class. Model 1 is great in term of concise course materials in three weeks. However it requires a specific subject for the seminar, and data visualization is only a subset topic for the class. As I read the description for Educational Studies 308, this class is designed for the studies of cities, suburbs, and schooling in the metropolitan Hartford area; one of the research skills is creating interactive data visualizations on school choice policies. Since students may have different interests or work with different type of community partner, it is challenging to finalize a single topic for this seminar.

Last but not least, it is effective if the community partners have their databases ready for use and finalize their goals of data visualization before working with Trinity students. On the first few meetings, community partners discuss about their priorities and how Trinity students can help them to create data visualization that benefits their goal. Each week students make different rough drafts of map, chart or graph based on the goals, and show community partner to receive constructive feedbacks. In the end of the semester, community partners will choose the best data visualizations that meet their goals for future uses.

Assignment 8: How to Lie With Maps

Posted on

The two maps below both display the percentage of college graduation in Connecticut, in which data is provided by Social Explorer. By using different colours and break intervals, these two maps provide two completely different portrayals of a same dataset.

In the first map, I use different shades of green to show the college graduation rate across Connecticut. Since there is no dramatic change in variation,  the similarity among various census tract polygons are highlighted. Moreover, by breaking the data into 9 buckets, I intentionally vary the length of each interval to manipulate the number of census tracts in each group. The top 2 and bottom 2 groups have the less data sample, while 5 middle buckets have the most data sample. Even though the legend shows wide range of rates, my dataset is mostly concentrated in the middle group. Therefore, it seems like the percentages of people who graduated from college are quite similar among all the census tracts in Connecticut.

Using the same data but with different choice of colours and intervals, I display my second map as a highly unequal distribution of college graduation rate. The data is divided into only two buckets, which are assigned with two highly contrast colours: extremely dark and light green. As illustrated on the map, it seems like more people graduated from college in Northern and South West of Connecticut.