GRADS (the GRaduate Assignments DataSet) is, to the best of our knowledge, the first dataset of readings assigned in graduate coursework across the discipline of political science in the United States (for prior studies limited to International Relations, see work by Jeffrey Colgan and the LSE IR Gender Project). Replication data files can be downloaded from Havard Dataverse. Our multilevel dataset contains data from:
- 88,673 non-unique readings (meaning that some readings appear in multiple syllabi)
- 137,305 non-unique authors
- 840 syllabi and 65 reading lists
- 606 unique course instructors
- 95 unique political science departments
- Courses taught between 2004 and 2017 (87% from 2013 to 2017, and 58% from 2015 or 2016)
The data thus provide a snapshot of what political science graduate students were reading in this time period. The listings can help faculty discover what their peers are assigning, assess trends in graduate coursework, and diversify coursework. Here is the distribution of our syllabi and reading lists by subfield:
| Characteristics of the GRADS Dataset | ||
|---|---|---|
| Subfield | N Syllabi | N Reading Lists |
| All documents | 840 | 65 |
| Comparative politics | 268 | 23 |
| International relations | 162 | 10 |
| American politics (including judicial/courts) | 163 | 18 |
| Methods | 179 | 2 |
| Theory | 94 | 7 |
| Political economy | 60 | 1 |
| Public policy/Public administration | 15 | 4 |
| Political psychology | 21 | 0 |
| Gender/Identity | 22 | 4 |
| Note: Counts by subfield sum to more than the total because some syllabi/reading lists are counted in multiple subfields. | ||
Assembling the dataset involved several phases of data collection.
- Through our own and others’ web searches in 2015 and 2016, we obtained 160 syllabi in Microsoft Word or PDF format. Professor David Samuels of the University of Minnesota generously shared a dataset of syllabi in comparative politics, which was gathered largely through web searches.
- In September 2016, with generous help from the American Political Science Association Research and Development Division, we gathered 301 syllabi via a national survey.
- The APSA Research and Development Division also shared 38 reading lists collected via a summer 2016 survey of graduate program directors.
- In the fall of 2016, we recruited Ph.D. student project affiliates in 27 of the top 50 political science Ph.D. programs. Through their efforts, we were able to collect 450 syllabi and 28 reading lists.
In assembling the final dataset, we excluded duplicates and earlier versions of syllabi. We then used a combination of manual and machine coding methods to extract citations from syllabi, and to parse those citations into constituent elements.
We coded author gender primarily using automated methods, relying on a list of the genders of given (i.e. first) names assembled from several lists published by the US and UK censuses and private firms relying on social media data. However, we also developed an extensive list of known scholars whose genders would be miscoded or unclassifiable using standardized tools. We were able to code the gender of all but 76 of the 137,305 authors, for a missingness rate of 0.055%. Another 54 works were coded as not having an author, and 285 as having a “corporate author” such as the United Nations Development Programme.
The full GRADS dataset contains data at the level of instructors, syllabi, and departments. We anticipate making the full dataset available in early 2019. In the full dataset, we will remove instructor and department data that either is sensitive or could be used to identify individual instructors, in accordance with IRB registration with our universities. In this online, searchable dataset, we only make available the data on the readings.
Heidi Hardt