Elementary Statistics With R

Instructors with

• some user experience with R
• who have studied the Project Mosaic approach to teaching elementary statistics with R

Useful eCOTS 2014 sessions:

• Effective Teaching using R, RStudio, and the MOSAIC Package
• Planting Seeds of Reproducibility in the Introductory Statistics Course with R Markdown

install.packages("mosaic")

Gentle down the coding!

• minimize number of functions to know
• provide uniform user interface for those functions
• don't force students to write their own functions
• pay little attention to types of R objects
• avoid teaching flow-control structures

install.packages("devtools")
library(devtools)
install_github(repo="tigerstats",
          username="homerhanumat")

• Used alongside of package mosaic
• Customize output to GC course objectives
• Alternative function help (Course Notes, blog posts)
• Simulation with manipulate/Shiny apps
• tigerstats source code is here.

Uniform function template for both descriptive and inferential procedures:

\[ procedure(Formula,data=Mydata,\ldots) \]

Formulas are of four types, based on nature of variables in one's “Research Question.”

Output is customized to GC course objectives. Paradigmatic examples:

• chisqtestGC()
• ttestGC()
• lmGC()

• Course Notes integrated with R-code-instruction
• knitr package converts R Markdown to HTML, pdfLaTeX
  • R Markdown Homework assignments
  • Lecture slides start as R Presentation documents
  • Student data analysis reports start in R Markdown, convert to other formats
• Further tools for automated production of materials:
  • Pandoc for Course Notes, Octopress for blogging with R
  • New and promising: GitBook with Jason Bryer's Rgitbook package

RStudio's manipulate package enables interactivity in the Plots pane. (More info here.)

• Can be programmed to work with arbitrary datasets;

CIMean(~height,data=imagpop)

The shiny package produces web applications. (More info here.)

• More flexible user interface;
• Permits detailed explanation of learning activity;
• A bit more cumbersome for user to load datasets.

• Web-based (no R required), RStudio hosts: http://homer.shinyapps.io/ShallowReg
• Web-based, set up your own Shiny server: http://rstudio.georgetowncollege.edu:3838/ShallowReg
• (with shiny) fetch from Github repository, run locally:

require(shiny)
runGitHub(repo="tigerstats",
    username="homerhanumat",
    subdir="inst/FindRegLine")

• (with shiny) run locally from installed package:

runApp(system.file("SlowGoodness",
        package="tigerstats"))

Further Resources:

• Course website: http://statistics.georgetowncollege.edu
  • most course materials available here
• Blog: http://statistics.rainandrhino.org/
  • includes Tech Notes for instructors

Remember:

• You don't need to use our package: modify, and make your own!
• tigerstats = training wheels; students who move on in statistics should learn standard R functions