Last week I attended a training course at The University of Sheffield,
"Help ! I need to use R", which aims to "enable researchers to
make an assessment of whether they want to use R". This post is a summary
of what I learned on this course.
What is R?
The Comprehensive
R Archive Network defines R as being "a language and
environment for statistical computing and graphics". It does not define it
as being a software, meaning that it is not directly comparable with statistical
software like SPSS, Minitab, Stata or SAS. Instead, it is a highly tuned
programming language that has the ability to do statistical calculations.
Different R packages are collections of coded procedures usually created by a
third party.
Why use R?
- It is free when other
methods of doing statistics can require a license
- It can create journal
quality graphics (e.g. the package ggplot2 is one of the most advanced and
flexible package for the production of graphs)
- Because it is what it
expected of you. Maybe it is common in the type of journals you are
interested in publishing in or maybe your supervisor prefers the use of R
How to Download R
If you are using a managed computer at The University of Sheffield, the
Software Download Centre can allow you to download R. On your
own computer, one place to download R is from Bristol
University. You can then download extra packages which contain
functions to do statistics.
Useful Packages
- MASS- package associated
with Modern Applied Statistics
- Psych- functions are
primarily for multivariate analysis and scale construction using factor
analysis, principal component analysis, cluster analysis and reliability
analysis
- Car- companion to applied
regression
- Multcomp- multi-comparison
in ANOVA
- Advanced plotting packages-
Lattice, ggplot2
Some places to get help
- In R type: help()
- Helpful book- R in Action by
Robert Kabacoff
- Use the Quick R website
- Google it! e.g. "r
ggplot2 box plot"
