What Is Happening on Twitter? A Framework for Student Research Projects With Tweets

Figures & data

Table 1 Classifying project skills

Skill	Category
Communicate results in speaking and in writing	Enduring understanding
Formulate a research question	Enduring understanding
Develop data science strategies to address a research question	Enduring understanding
Use text analysis tools to analyze tweets	Enduring understanding
Translate analysis results into scientific conclusions	Enduring understanding
Describe assumptions and limitations of statistical analyses	Enduring understanding
Use Github to share code and documentation	Important to know and do
Use git for version control	Important to know and do
Use data visualization to clarify and inform quantitative analyses	Important to know and do
Incorporate supplementary data sources into an analysis	Important to know and do
Acquire data from internet sources	Important to know and do
Structure research project files as an R package	Worth being familiar with
Use cluster computing as needed	Worth being familiar with

Table 2 Rubric for summative assessment of learning objectives.

Learning objective	Assessment item	2 points	1 point	0 points
Write R code to perform text analysis of large volumes of tweets.	R code performs intended analyses	Code contains few or no bugs	Code contains one or more errors	Code contains many errors
Write R code to perform text analysis of large volumes of tweets.	Uses literate programming tools, such as Sweave or knitr	Report is written using literate programming tools. It compiles easily when run by instructor. Time-consuming calculations are cached.	Report is written using literate programming tools, but compilation takes too long or fails.	Report is not written with literate programming tools.
Write R code to perform text analysis of large volumes of tweets.	Uses git for version control	Log reveals regular commits with informative commit messages	Log reveals intermittent commits and uninformative commit messages	Does not use git.
Write R code to perform text analysis of large volumes of tweets.	Shares code and data via Github	Instructor easily clones repository from Github. Contains share-able data and instructions for getting other data to reproduce analysis.	One or more needed files is missing from repository.	Does not use Github.
Communicate results in a written report and poster presentation.	Organizes poster to highlight main points	When prompted, can describe main points in less than 1 min.	Less fluid presentation with periods of silence or confusion.	Disorganized presentation.
Communicate results in a written report and poster presentation.	Accurately presents study and findings during poster session	Fluently describes background, study goals, study design, approach, data, findings, and conclusions	At least one section is incomplete or is verbal explanation is incomplete.	At least one section is missing.
Communicate results in a written report and poster presentation.	Report structure mirrors a research manuscript	Contains abstract, introduction, methods, results, and discussion	At least one section is incomplete.	At least one section is missing.
Translate statistical findings into scientific conclusions.	Places statistical results in their scientific context	Demonstrates understanding of scientific context and integrates findings into it.	Incomplete scientific understanding or incomplete integration of findings.	Major gaps in scientific understanding or integration of findings.
Translate statistical findings into scientific conclusions.	Accurately portrays study limitations	Accurately describes, in writing and in speaking, limitations of the study	Incomplete or partially inaccurate description of limitations	Does not describe limitations.
Translate statistical findings into scientific conclusions.	Demonstrates familiarity with relevant literature	Fluent in both relevant data science literature and scientific literature.	Incomplete knowledge and understanding of relevant literature	Major gaps in knowledge and understanding
Develop data science strategies to address a scientific research question.	Presents an original research question	Presents, in writing and in speaking, a novel research question. Explains why it’s novel, too.	Partially lacking in elements of question’s background or novelty.	Does not present an original question.
Develop data science strategies to address a scientific research question.	Effectively uses data visualizations	Visualizations highlight main points of report.	Incomplete or omitted visualizations.	Does not use visualizations.
Develop data science strategies to address a scientific research question.	Presents accurate scientific conclusions	Effectively translates analysis results into their scientific context.	Minor inaccuracy in translation of findings into scientific context.	Major errors in translation of results.

Fig. 1 Top terms for LDA model from May 24, 2020. Results contain expletives and may be not suitable for work (NSFW).

Fig. 2 Top terms for LDA model from May 25, 2020 (Memorial Day).

Fig. 3 Top terms for LDA model from May 26, 2020.