Grinnell’s Data Analysis and Social Inquiry Lab (DASIL) helps students and faculty members integrate data analysis into research and classroom work.
Data Science Case Study – International Global Education: Project Assessment
DASIL in collaboration with IGE 2020/2021
This project was created to help the IGE and College leadership reach a decision regarding the viability of running off-campus study programs for the 2020-2021 academic year. The visualization below is the result of a principal component analysis (PCA).
Data Science Case Study – International Global Education, Project Assessment evaluates the viability of OCS (off-campus studies) programs at Grinnell College during the COVID pandemic. This project is the result of a collaboration between IGE (Institute of Global Engagement) and DASIL (Data Analysis and Social Inquiry Lab). On the y-axis, “openness” refers to indicators measuring how open the country is to foreign visitors in terms of crossing border requirements, mobility allowed within a country, and social gatherings. Openness also utilizes the available information on countries’ levels of openness through the Government Stringency Index. On the x-axis, “goodness” represents the capacity of each country/program to respond to a health crisis as a result of COVID. “Goodness” measures COVID protocols as well as a series of resources and objective indicators such as ICU units available, COVID cases and deaths per capita.
The data was collected using the resources available at World In DATA and other multiple sources including the following:
- Center for Systems Science and Engineering at Johns Hopkins University. (2020). COVID-19 Data Repository.
- Hale, Thomas, Noam Angrist, Emily Cameron-Blake, Laura Hallas, Beatriz Kira, Saptarshi Majumdar, Anna
Petherick, Toby Phillips, Helen Tatlow, Samuel Webster (2020). Oxford COVID-19 Government Response
Tracker, Blavatnik School of Government.
- Institute for Global Engagement. (2020). Partner Survey.
- Roser, Max, Hannah Ritchie, Esteban Ortiz-Ospina and Joe Hasell (2020) – “Coronavirus Pandemic (COVID-19)”.
- TerraDotta. (2020), Alert Traveler. Grinnell College Institute of Global Engagement.
- U.S. State Department (2020). Travel Advisories. Travel.state.gov.
Methodology – Since we are handling a lot of information that measures both “openess” and “goodness” simultaneously, a data reduction technique based on information compression is needed to integrate the multiple indicators of the data. This significantly reduces the numbers of measures and more intentionally describes the general situation of the OCS programs. We adopted PCA (principal component analysis) to reduce the multi-dimensional data set into two overall dimensions. The principal components are linear combinations of data features which represent the direction that contains the largest information (variance). Then, we selected the first two principal components and defined them as “openness” and “goodness” to measure each OCS program and used a data visualization for comparative temporal analysis.
Goodness & Openness
Using PCA, we obtained two major measurement indexes namely “openness” and “goodness” computed as follows:
- Openness= – 0.12*Border crossing – 0.56*Social gathering restriction – 0.60*Government stringency index + 0.55*Hospital beds per 1000.
- Openness is measured as a dimension capturing whether border crossings are enforced, whether social gatherings are allowed, Government stringency index, and number of hospital beds per 1000 inhabitants. High values of these measures are associated with more “Openness”.
- Goodness= – 0.23*Housing – 0.36*COVID protocol – 0.35*Safety protocol – 0.28*Lockdown plan – 0.59*Health system sufficiency – 0.49*Cases per million – 0.14*Deaths per million.
- Goodness is measured as one dimension resulting from new cases and deaths per 1 million. New cases and deaths are measure monthly as a 7 day average. It includes the OCS Partners COVID Protocol variables. High values of these measures are associated with more “goodness”.
Recommendation for Spring 2021 – “Go” or “No Go”: Based on the data visualization provided resulting from the PCA, this is the list of countries that we define as a “go” or “no go”.
Low Risk (“go” countries):
- Denmark (10 – 30)
- Italy (4)
- Russia (1)
- Hungary (8)
- Taiwan (3)
Moderate Risk (near the line of “go”):
- France (2)
- Germany (4)
- Japan (9)
- Ireland (1)
- The Netherlands (2)