# Goodies

I am a firm believer in teaching base R before moving onto the tidyverse, and Norm Matloff’s post on Teaching R in a Kinder, Gentler, More Effective Manner is spot-on.

A fantastic cross-reference of operations in R, Python, and Matlab/Octave by Vidar Bronken Gunderson. [archived]

Common statistical tests are linear models by Jonas Kristoffer LindelĂ¸v

Rob Hyndman’s Twenty rules for good graphics

A good explanation of why you should use natural logs on betterexplained.com

Writing R Functions. Some sound advice.

10 R packages I wish I knew about earlier - blog post on Yhat.

New York Times article on the growing use of R, from 7 January 2009. Thanks to Scott Gifford for finding this!

Paulina Zheng’s useful guide for how to handle your confidence level when you conduct many tests, that is, the multiple comparisons problem.

Max Pekarsky on The Overflow, on writing code other people can read.

## Why p-values should be avoided (or at least used with great caution)

FiveThirtyEight on why scientists have difficulty explaining p-values.

A follow-up by Nate Silver on the problems with p-values.

Nature on the problems of p-values.

AGU's EOS forum (24 November 2009, v. 90, no. 47) on statistical significance, geological significance, and the problem with p-values.

Science News on the misinterpretations of p-values and their possible remedy in Bayesian statistics (*Available only to subscribers of Science News*).

## The XKCD Corner

XKCD on Tufte. Hover your mouse over the comic for Tufte.

XKCD about the misinterpretation of p-values in the popular press.

XKCD on curve fitting and the messages they send.

XKCD with a modified Bayes’ Theorem.