The lab will walk you through analyzing an example data set, and you will then analyze two datasets from published research papers. You will gain basic familiarity with analyzing experiments using mixed‑model analyses of variance. Consistent with lecture, this assignment is not intended to provide complete knowledge of how to design or analyze experiments, which is far beyond the scope of one lecture or assignment. This assignment is instead focused on a pragmatic introduction to analyzing experiments based in designs you might later find useful. Please consider this assignment in the context of the material covered in lecture, as not all of it is repeated here.
You will complete a small statistics lab using R:
You will want to complete the assignment using R Markdown within RStudio Desktop:
https://rmarkdown.rstudio.com/
https://rstudio.com/products/rstudio/#rstudio-desktop
This will create a PDF that contains your interleaved comments, scripting commands, and output. This is a good practice for documenting your analysis, and easier than trying to copy and paste commands and graphs into a different document.
In addition to the lecture material and the contents of this assignment, you might benefit from working through all or part of Jacob Wobbrock's independent study in Practical Statistics for Human-Computer Interaction.
http://depts.washington.edu/acelab/proj/ps4hci/index.html
You are welcome to work with a partner throughout this assignment. Only one person in the partnership needs to submit an assignment. If you work with a partner, please include their name near the top of your report.
When analyzing data (e.g., to write a paper), it is often valuable to talk through your analyses with another person. This is useful for checking that what you did sounds correct and for thinking about how to proceed if stuck. If you talked through the analysis with others who were not your partner, please also indicate this in your submission.
You will work with three datasets: one artificial and two from published studies. The data is appropriate for these analyses, but explicitly and intentionally not cleaned up for the sake of this assignment. One important implication is that you need to be mindful of the types assigned to columns in provided data files. For example, it will be your responsibility to decide what type to assign each field (e.g., continuous, ordinal, nominal).
You will be graded on the correctness and the appropriateness of your responses to questions posed by the assignment.
Although correctness may seem self-evident, community norms can sometimes make this less obvious. For example, analyses in the papers associated with assignment data were performed before many forms of analysis were available in available tools. Analyses that are now available may also not yet be commonly-known within a community. You are expected to make appropriate choices in your analyses and to explain those choices as necessary. We explicitly do not expect your analyses or results to exactly match those in the previously-published papers.
Grading of appropriateness will be based in striking a balance between reporting sufficient detail while not reporting excessive detail. The goal is to gain experience reporting your results in approximately the same level of detail that should be included in reporting a research result.
The statistics lab is available for download:
Submit both your notebook in RMD format and a rendering in PDF format: