Thinking with Data in the Second Course

• Activity Number: 97
• Type: Invited
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Education
• Author(s): Nicholas J. Horton*+ and Ben S. Baumer and Hadley Wickham
• Companies: Amherst College and Smith College and RStudio
• Keywords: statistical computing ; statistical education ; reproducible analysis ; databases ; model eliciting activity ; big data

Abstract

Students in their second statistics course need exposure to bigger datasets and more complex questions to be able to make sense of the increasingly data-centric world that they will inhabit. This talk will describe how the airline delays dataset (150 million records on all commercial flights in the US from 1987 to 2012) can be integrated into the second statistics course. This large dataset can be introduced early in the semester through a model eliciting activity due to Garfield and colleagues that leads students to describe distributions and undertake informal inference when comparing the on-time performance of two airlines servicing a pair of airports. Later in the course, students are able to assess the performance of their comparison rule by repeatedly sampling from the underlying population of flights, as well as visualizing this large population using straightforward commands in R to access the database using simple SQL commands. The techniques are facilitated by use of reproducible analysis tools (R Markdown). In addition to providing insight into flight delays, the activity helps expose students to the power of statistics to make decisions in the face of uncertainty.