Wrong Data, Wrong Methods: An Exercise in Linear Regression

As part of a course in regression using R, I analyzed log data about museum visitors’ interactions with a multi-touch tabletop (called Oztoc). Oztoc is located at the New York Hall of Science and supports visitors’ inquiry about biology and electrical engineering. To analyze visitors’ interactions,  I included variables about visitors’ activity on the tabletop (such as type of object used, x,y-coordinates of the object, etc.) and tested each variable for significance using nested models.

This was my first real exposure to making sense of a large data set using regression techniques available in R. Visualizing data helped me to identify how assumptions about linear regression were met (or not met, in this case). After this project, I felt more confident when critically evaluating how variables in a data set may contribute to a model that actually tells us something about collaboration and learning with a multi-touch tabletop exhibit.

Note: this project is not published but is written in an APA-style paper. Log data was converted from JSON to CSV output.