Exploring methods and topics in Comp/Rhet dissertations

55 total topics, less 7 excluded topics

Here you'll find several higher-resolution data figures from my dissertation on the methods and topics of dissertations in Rhetoric and Composition. Follow the links below to view or download. The code that generated these images is available on GitHub.

Chapter 2: From Dissertations to Data: the origins and extent of this dissertation's exhibits

• Figure 2-1: What data and metadata ProQuest provided, and what I added
• Figure 2-2: World map of schools in the composition/rhetoric dissertation dataset
• Figure 2-3: Superimposed map of Carnegie-classified doctoral schools, schools in the composition/rhetoric dissertation dataset, and schools in the Consortium of Doctoral Programs of Rhetoric and Composition

Chapter 3: Mapping the Methods of Composition/Rhetoric Dissertations: a 'landscape plotted and pieced'

• Figure 3-1: Method tags aggregated by schools, all schools
• Figure 3-2: Frequency of assigned method tags
• Figure 3-3: Relative ranks of assigned method tags at Consortium and non-Consortium schools
• Figure 3-4a: Method tags aggregated by schools, Consortium schools only
• Figure 3-4b: Method tags aggregated by schools, non-Consortium schools only
• Figure 3-5: Method tags aggregated by schools, only non-Consortium schools with more than five dissertations in the 10-year dataset

Chapter 4: Tapping the Topics: what we study when we write in writing studies

• Figure 4-1: Two topics in relation to three dissertations, by way of introduction to topic modeling
• Figure 4-2: Hierarchical clustering of 150 topics, both without (4-2a) and with (4-2b) the final 55 topics identified.
• Figure 4-3: Screenshot of top topic browser, written in R
• Figure 4-4: Screenshot of top topic browser in detail mode
• Figure 4-5: Variation of topic proportions, top 10 topics per document
• Figure 4-6 (interactive!): Radial clusters - The 50 content-bearing topics arranged in a hierarchical tree, which positions topics more closely together the more similar they are to each other (i.e. the more highly correlated their word-frequency vectors are).
• Figure 4-7: Radial clusters with highlights showing the six major divisions of the tree
• Figure 4-8 (interactive!): Co-topic clustering - The same radial layout as in Figures 4-6 and 4-7, now showing which dissimilar topics tend to co-occur in the same dissertations.
• Figure 4-9: Co-topic clustering with highlights showing the same major divisions as in Figure 4-7.

Additional figures

The following figures update earlier versions from the dissertation, which I'm leaving above for archival purposes.

• Radial clusters, Consortium Programs only. This model uses the same topics from the analysis in chapter 4, above, but now filters out dissertations completed outside of the specific doctoral programs with representatives in the Consortium. (The previous models included dissertations from any department in a school with such a program.)
• Co-topic clustering, Consortium Programs only. This model used the same topics from the analysis in chapter 4, above, but now filters out dissertations completed outside of the specific doctoral programs with representatives in the Consortium. (The previous models included dissertations from any department in a school with such a program.)

Thanks for visiting!