Check out some of our favorite sites

Melting heart of statistics in Salvadore Dali’s surrealist style, by DALL-E

We have learned so much from blogs, bookdowns, and videos openly shared by others. Here, we compiled a list of some of our favorite resources.

This list is far from exhaustive; there are too many influential sources to recall and credit as deserved. Nonetheless, we felt it would be helpful to pull together some of those that we routinely find ourselves revisiting and recommending to others.

statistics, especially Bayesian approaches, and more

Andrew Gelman’s blog is perhaps our favorite place to go for musings on all things statistics, especially Bayesian inference, frequentist evaluation, causal inference, multilevel modeling, uncertainty, and reproducibility. We also appreciate the frequent attention paid to philosophy of science issues throughout many posts. Bonus: if you browse long enough, you might even stumble upon an answer to the question that perennially motivates a career crisis for us: The social sciences are useless, so why do we even study them?.

Speaking of Andrew Gelman - do you want to learn regression or Bayesian modeling? Gelman and colleagues’ books and accompanying websites are excellent resources on these topics. However, our go-to resource for learning statistics from a Bayesian framework is Richard McElreath’s Rethinking Statistics (2nd Ed.). All code from the book and a corresponding R package (rethinking) are openly shared online, and McElreath’s brilliant video lectures on YouTube, which he routinely updates, simply are without equal. Seriously - stop reading our site now and go work your way through those lectures and associated code.

As much as we love McElreath’s book and lectures for learning a Bayesian approach to data analysis, we usually rely on Paul Bürkner’s “Bayesian regression models using Stan,” or brms(), R package for everyday Bayesian modeling tasks. In addition to the brms() package, we recommend checking out Bürkner’s research projects and publications. In particular, we especially appreciate his work on modeling and visualizing uncertainty in (Bayesian) model comparison and his efforts to incorporate cumulative link functions for modeling monotonic effects of ordinal predictors in Bayesian regression models. Meanwhile, we are particularly fond of Bürkner’s brms vignettes, especially those on multivariate (i.e., multiple response variables) models and monotonic effects. You can also watch his talk on multilevel modeling in brms for free courtesy of @GenerableHQ.

So, what if you want to learn Bayesian analysis from McElreath’s book and lectures, yet you also want to follow along with his examples using the brms() package? Enter Solomon Kurz, the data science hero we need but do not deserve. Kurz has dedicated an unfathomable amount of energy to developing Bayesian translations of some of our favorite statistics books into R language and the brms() package, then sharing these valiant efforts in open-access bookdown projects. Specifically, here are Kurz’s translations of the first edition and second edition of McElreath’s book. Additionally, many years agom Jake and I initially learned longitudinal data analysis in the SAS program using Singer and Willett’s (2003) now-classic book; Kurz actually offers a free tidyverse and brms translation of their book as well! Likewise, he has similarly translated two other favorites of ours: Kruschke’s Doing Bayesian Data Analysis and Hayes’ Introduction to Mediation, Moderation, and Conditional Process Analysis. If you make it through all that free content, then we suggest heading over to his informative blog on data analysis in R for more useful content.

There are so many others worth mentioning in this section but, in respect of your time, we will simply share the rest in list format:

• Matthew Kay’s site for inspiration on communicating and visualizing uncertainty with tidybayes and ggdist. In particular, we recommend Kay’s various package vignettes, such as this one on using tidy data with Bayesian models and this one on slab + interval stats and geoms with ggdist. Also, you can watch Kay’s talk on visualizing uncertainty courtesy of @GenerableHQ.
  
• Andrew Heiss’s blog, especially entries on different types of posterior predictions, marginal effects terminology, and conditional and marginal effects in multilevel models
• Aki Vehtari’s work, especially that on model selection, model checking and cross-validation, and posterior stacking. Also, be sure to check out his various videos, like this one on Bayesian workflow or his open access course videos on Bayesian data analysis.
• Michael Betancourt’s writing and videos on probabilistic modeling in Stan, and especially those on adopting a principled Bayesian workflow.
• The 100% CI is a newer favorite. We are especially fond of Julia Rohrer’s (one of the site’s authors) various writing related to causal inference, like these on directed acyclic graphs, collider bias, and what can go wrong with nonrepresentative samples.
• The Data Colada blog offers great analysis on various topics. For instance, they write about replication, conduct preregistered replications, and even offer an R package (groundhog) for improving reproducibililty of code. They also have interesting things to say about meta-analysis and problems with common techniques such as funnel plots and trim and fill “corrections”.
• Dan Quintana, co-host of the Everything Hertz podcast, also has a blog containing great resources for meta-analysis and more in R.
  
• Daniel Lakens’ blog, the 20% statistician, is rife with useful posts about two of our favorite topics - statistics and philosophy of science. Some of our favorites include posts about p-values, power and observed power, the smallest effect size of interest, or whether philosophy of science matters in practice (we think it does a lot in criminology). Several of his lecture videos are also available online, including this one that corresponds to another favorite post discussing why more of us should seriously consider whether we are really ready to test a hypothesis in our research. Lakens has combined and expanded on a lot of these materials in his open access book, Improving Your Statistical Inferences and his free coursera course of the same name.
  
• Of all the podcasts we enjoy, the Quantitude podcast deserves a special shout-out for its highly informative yet quirky and accessible introduction to all sorts of important issues in statistics. Now, if only the hosts would join the Bayesian dark side.
• Jacob Kaplan has multiple resources related to R that are geared specifically to Criminologists. Perhaps most notably is a free online book Crime by the Numbers: A Criminologist’s Guide to R. Kaplan has great programming skills and, as such, many of his examples rely on base R functionality more so than ours do.

learning data science with R

While there is so much to learn from resources found at the links shared above, those are not the first places we send our students and friends that are interested in learning data science skills in R, yet that lack any R programming experience. We find the following links to be especially useful for these purposes - so much so that they deserve their own dedicated section! Of course, we also hope our course materials might be useful in this regard to some of you.

Danielle Navarro’s blog is as delightful as it is informative. The blog is a model for reproducible R programming and showcases Danielle’s inspiring art throughout the site (see more of Danielle’s computational art here). The newest version of the blog often focuses on technical posts addressing issues beyond our range of program expertise. However, we love various topical essays from earlier blog renditions, like this one on preregistration, and we strongly recommend Danielle’s Data Science with R course and associated video tutorials for R beginners or those interested in leveling up their R skills.

Allison Hill’s blog is a great place to find resources for learning to do as well as to teach R-related topics. For instance, we especially appreciate her post on teaching R Markdown, which was influential in creating our own courses, as well as her post on transitioning to Quarto, which was helpful to us in creating our blog.

The psyTeachR website is a phenomenal repository of resources for learning and teaching reproducible research using R. They offer a wealth of open access courses and tutorials for undergraduate and graduate levels.

The easystats collection of R packages is a great place for those new to R to learn basic skills related to modeling, visualization, and reporting within a relatively unified framework. Those new Unlike curated nature of R and package conflicts Those transitioning from curated statistical packages to R’s open-source environment, with its a la carte format and associated challenges (e.g., package conflicts), may find the easystats suite of packages especially helpful.

Of course, we also love creating data visualizations with ggplot2 in R, so we owe a huge shout out to Hadley Wickham for all his R programming efforts and clear tutorials and foundational texts, which you can learn more about and freely access via his website.

creating a website

If you want to learn to create your own website or blog using R, we recommend checking out Andrew Rapp’s ultimate guide to creating a Quarto blog, Beatriz Mills’ guide to creating a blog with Quarto in 10 steps, Samantha Csik’s tutorial on adding a blog to your existing Quarto website, and Ezekiel Ekunola’s instructions for setting up a personalized domain address for your site.

In addition to R-based visualizations, we relied heavily upon Allison Horst’s website, where she freely shares amazing data science, statistics, and R themed artwork, to help improve our site’s aesthetics. We also had fun generating custom images and art using the artificial intelligence system DALL-E.