For me, statistcal inference is a a tremendously powerful tool, but it requires you to respect all the rules lest the inference be lost. I was not trained as a statistician; for me (as you might expect), inference is an intimidating tool.

The idea of resampling is very appealing to me. I appreciate that you can answer (albeit somewhat less formally) many of the same questions as you can using inference. As long as you make a good-faith effort not to fool yourself, you can use resampling to get useful answers.


This is clearly not a list of academic citations.

Prof. Deborah Nolan

embed_user2016("Statistical-Thinking-in-a-Data-Science-Course") %>%
  use_start_time("39m21s") %>%
  div(align = "center")

One of the points that Prof. Nolan makes is that we can investigate the question of significance by using randomization, resampling, and rejection - rather than by using formal statistical inference.

She also refers to Tim’s Hesterberg’s paper on bootstraping and undergraduate statistics. Also refers to the Five Locks.

Broom package

embed_user2016("broom-Converting-statistical-models-to-tidy-data-frames") %>%
  div(align = "center")

GitHub repository

R for Data Science

embed_user2016("Towards-a-grammar-of-interactive-graphics") %>%
  use_start_time("15m15s") %>%
  div(align = "center")

R for Data Science book

Introduction to Statistical Learning

Link to the book