25 What next?
25.1 Data wrangling and visualization
Complex analyses can be vital means of developing insights and testing ideas, but often simple reorganization and visualization of data are at least as useful. Particularly for the former, tidyverse tools are efficient and flexible (if sometimes reliant on syntax that is difficult to decipher). Even as you add additional analytical methods to your toolbox, you may want to revisit the notes on Data Wrangling for Archaeology and plotting (using both Base R and ggplot). There are many guides available to using R for data visualization, and a few specifically archaeological treatments (e.g., (Edward B. Banning 2020; Mike J. Baxter and Cool 2016); the latter is available here). See also Joe Roe’s slides on Data Visualization for Archaeologists in R.
25.2 Other analytical approaches
Correspondence analysis is, of course, only one of many methods that don’t make it into Drennan’s book. For a brief review of books on quantitative methods in archaeology, see Drennan’s section on ‘Suggested Reading’ (2009, 321–25). More recent texts not covered there include include the 2nd edition of Baxter’s Exploratory multivariate Analysis in Archaeology (2015), VanPool and Leonard’s Quantitative Analysis in Archaeology (2011), McCall’s Strategies for Quantitative Research: Archaeology by Numbers (2018), Carlson’s Quantitative methods in Archaeology Using R (2017), and the first half of Banning’s The Archaeologist’s Laboratory (2020, Ch.1–8). Baxter’s Notes on Quantitative Archaeology and R (2015), available here though not formally published, is at least as cogent and useful as many things that are.
There are several recent chapters and papers presenting introduction to particular topics, especially multivariate (Glascock and MacDonald 2023; López-García and Argote 2026), Bayesian (E. Otárola-Castillo and Torquato 2018; E. R. Otárola-Castillo et al. 2022), and modeling (Crema 2025) approaches. There are even more examples of their application, particularly in spheres where analytical approaches are fundamental to data interpretation (for example in use of chemical compositional data); listing those is beyond what I can do here. Attention to these topics results at least in part from the increase in easily-available computational power that we touched on when we covered resampling approaches. They also tend to reflect the reservations about null-hypothesis statistical testing (NHST) and the skepticism about fetishization of p-values that runs through Drennan (though they certainly do not originate with him (see Cowgill 1977), and are becoming common in the sciences more broadly (see Amrhein, Greenland, and McShane 2019; Wasserstein, Schirm, and Lazar 2019)).