Biostatistics Textbook With Jamovi
A freely available textbook for statistics in the life and environmental sciences
I recently completed an introductory statistics textbook that uses jamovi software for all examples and practical exercises.
The book is freely available online (and always will be) on my website and on Bookdown, and a hard copy can be purchased from Routledge or Amazon. Datasets for exercises are freely available for download from the Open Science Framework (https://osf.io/dxwyv), and I have uploaded them to GitHub. The book uses R Shiny apps to illustrate some of the challenging ideas through interactive visualisations. My hope is to have a free audiobook available around October 2024 (roughly the same time as a print copy). The book is suitable for a one-semester introductory statistics (or biostatistics) class, and examples are derived from the biological and environmental sciences. Nine chapters of practical exercises using jamovi are interspersed with chapters introducing key statistical concepts. If you are a teacher wanting to use jamovi in your class, then I am happy to also share some additional materials that I have made that complement the text.
This post is primarily a reflection on writing the book, not a discussion about its contents. What follows is a summary of memories from my first few years of teaching introductory statistics, and a bit on academic life more generally. This includes some rather unpleasant memories of working through the COVID-19 pandemic. Finishing this textbook has become something of a milestone for me in a slow but steady recovery from the exhaustion and stress of starting an academic job during COVID-19. This kind of experience was not unique to me, but I think it is worth writing down.
Stepping into university teaching during COVID-19
The whole writing process has taken over two years, and it wasn’t my intention to write a textbook when I was first getting started. My original objective was to update the learning content for the introductory statistics class that I started coordinating in the 2021 spring semester. In late 2020, my role at the University of Stirling changed from being a research fellow (i.e., entirely research focused) to being a lecturer (which includes research, teaching, and administrative roles). Like most of the world, Scotland had been in a COVID-19 lockdown for quite some time. I had been working remotely since early 2020, and for public safety reasons, all teaching at the university had been abruptly moved online. I was just finding my footing as a lecturer the spring 2021 semester, and most of my time was spent keeping the introductory statistics class running and producing learning content that was previously delivered face-to-face. The class included about 150 students and was taught using SPSS statistical software. Lab practicals were especially challenging for both staff and students. Students had to get SPSS running on their own machines and work through exercises in isolation over a 3-hour MS Teams meeting. Teachers and teaching assistants would wait for virtual hands to be raised, then privately message or call students who needed help; it was new to all of us and a huge change from the traditional face-to-face computer labs. Lectures had to be pre-recorded and posted online, which required time for preparing, filming, and editing. A good 15 minute video could easily take half a day to make, so creating 2-3 hours of weekly lecture content took a lot of time. In addition to teaching, research and administrative work also had to be moved entirely online. For me, and many of my colleagues, the job never stopped; there was too much to do. For all seven days of the week, I would get up as early as I could, then go to bed when I was simply too exhausted to continue.
Unfortunately, I made a bad habit out of this routine. It lasted for years. Work continued even after the semester of teaching ended; a lot of grant writing, coding, paper writing, and administrative tasks that had been put aside to keep classes running now needed to be done. When autumn rolled around, some limited face-to-face teaching was to be introduced, but we still needed an online option for students, meaning that the learning content would take a hybrid approach (I was coordinating an Evolution and Genetics class in the autumn). In 2021, I took three days of annual leave: two days in the autumn, and Christmas Eve. I was working through the weekends. I rarely left the house.
When the 2022 spring semester started, the hybrid approach to teaching was still being used. Statistics lab practicals could now be attended face-to-face (with appropriate physical distancing in the computer lab) or online, with lectures and tests still being entirely online. Rather than just scrambling to get existing learning content online, now I was working to update it and improve on the previous year. I wrote new notes, lectures, practical exercises, and assessments, but there was a lot more that I wanted to do to improve the class. I realised that SPSS was holding back student learning. The esoteric protocols required for navigating SPSS, combined with the somewhat unusual labels for statistics and tests (e.g., ‘Sig.’ instead of ‘p-value’), made learning difficult. The fundamental statistical concepts that students were learning in the notes and lectures did not match up well to the software we were using to do the practicals, and students were quite sensibly focusing more on learning SPSS protocols than statistical concepts. Worse, there were new complications concerning how students accessed SPSS on their personal computers, and a lot of students reported difficulties using SPSS during their tests. This made a lot of students very stressed and anxious, and that was more than enough reason for me to try something new.
Switching from SPSS to jamovi
Given the challenges with student access to SPSS, I was keen to find something free and open-source, which could be easily downloaded and used by all of my students. I know that many of my colleagues at other institutions go straight to teaching R programming in introductory statistics. I love the R programming language, and I have been using it to do statistical analyses, code biological models, create webpages, build interactive apps, and write manuscripts for over a decade now. But I was reluctant to use R with students being introduced to statistics for the first time. For starters, I worried that I would run into some of the same issues that I did with SPSS. I wanted students to be able to focus primarily on the challenge of learning statistical concepts and understanding statistical tools, and less on learning new software. For its many, many advantages, learning R is its own challenge, and one that a lot of students find intimidating, especially if they have no prior experience with programming. At my university, we also have a separate, more advanced, statistics class that introduces R to third and fourth year students. Not every introductory statistics student takes this advanced class, but it is available as an option for students who are keen. In truth, while I encourage my introductory statistics students to strongly consider learning R (and even gently introduce it in the last two weeks of my introductory class), I don’t think that it is necessary for understanding or doing basic statistics. Instead, I wanted something user-friendly for students with no previous background in statistics or coding – something that would be a stepping stone to more advanced tools, but not distract from the challenges of learning statistical concepts for the first time.
I started searching for options. I decided that I needed something reliable, user-friendly, free, and that would work across as many platforms as possible. I was pessimistic at first, but then I saw that there were some viable options. I looked at PSPP, but the interface looked too much like SPSS (this is by design, of course) for my liking. When I discovered JASP, I was relieved. It had all of the features that I absolutely needed; it was free, open-source, user-friendly, and worked across Windows, macOS, Linux, and Chromebook. If nothing else was available, JASP would definitely work! I continued looking around, then came across jamovi.
Jamovi looked similar to JASP, but with a slightly more user-friendly interface. And while many users probably appreciate the many Bayesian statistics options promoted by JASP, these would have to be ignored for my introductory statistics class. Jamovi was not focused on promoting a specific statistical philosophy. Like JASP, it was free and open-source, and would work across Windows, macOS, Linux, and Chromebook. Jamovi could even be run directly from a browser using jamovi cloud, which was a nice touch. I could be confident that all of my students would have access to jamovi even after finishing their studies and leaving university. The spreadsheet feature of jamovi also seemed more advanced than JASP, and the add-on module options in jamovi seemed especially promising for teaching. Base jamovi was quite minimalist, but with the option to expand the number of possible analyses with these modules. This minimalist baseline with add-on module options was a big pedagogical benefit in my mind. Students just getting started wouldn’t be overwhelmed by the interface, but as they gained confidence, the add-ons could be downloaded to do more advanced analyses or visualisations. I could even write my own modules and contribute them to the jamovi project (I’m hoping to write a module for randomisation and bootstrapping to go with the final chapter of my book). Even better, jamovi allowed for the R syntax underlying its analyses to be visible above the output (it’s hidden by default). Students intending to make the eventual transition to R could turn on this syntax mode to see the code underlying the jamovi point and click tools. And the jmv package used for analyses was fully accessible in R, meaning that I could run jamovi analyses and get jamovi output from within R. This was very useful! Because my tests were still entirely online, I needed to give each student a custom dataset, then write an R script to get the correct answers and deliver specific feedback for each student. Jamovi gave me everything I needed, and it seemed like the my best option for teaching.
A textbook and lab practicals for jamovi
I then needed a lot of new learning content, especially new practical exercises for jamovi. The existing exercises were all in SPSS. The lectures often mentioned SPSS, as did the notes, and the tests were written with SPSS in mind. A lot would need to be changed, and it wasn’t just me who needed to update the learning content. Two other faculty members taught on the introductory statistics class; to my relief, they were both incredibly supportive of the switch and agreed to make updates to their own lecture content. I was excited at this point and ready to rebuild all of the learning content. In the summer of 2022, I scrapped the old notes and decided to start from scratch. I needed a textbook, but I wasn’t sure what was available for introductory statistics with jamovi. As with the statistical software, I wanted the textbook to be free and open source. I didn’t want my students to have to pay for a textbook, or to have access through the library, which would become unavailable once they left university. This was going to be more difficult. I came across Navarro and Foxcroft’s online textbook, and I really liked it. The writing was clear and engaging, the scope of the statistical content was just about right, and the focus was on jamovi. The only problem was that it was heavily focused on psychology. My students, with some exceptions, were in the biological and environmental sciences. I needed a book more focused on using examples in these subjects, and ideally with practical exercises that students could work through each week of the semester. After some searching, I concluded that there was no such book for introductory statistics using jamovi. If I wanted one, I was going to have to write it myself. My initial goal was more modest. I would just write some detailed lecture notes on key statistical concepts, and rewrite the lab practicals from scratch. I would post these online for my students, with links to other useful materials (I still do this, with Navarro and Foxcroft’s text as recommended reading along with some other good textbooks). I started writing sometime around late July 2022 (still, unfortunately, unable to break the habit of seven day work weeks). I was excited.
At the time, I had no intention for the learning content to reach beyond the University of Stirling. I wanted the notes to form a coherent statistics workbook and showcase the interesting, diverse, and globally-focused research that my colleagues were doing in the department while also being grounded in Scotland (the final book retains Scottish words such as ‘loch’ and ‘outwith’, and includes exercises inspired by data collected in Scotland). I asked my colleagues for examples from their own projects, which nicely spanned the biological and environmental sciences from various systems around the world. I also had plenty of data from my own doctoral research, which focused on nonpollinating fig wasp communities in Baja, Mexico.
I wanted the workbook to be accessible, and for obvious reasons, I needed students to be able to engage with it entirely online given the new era of remote and hybrid teaching. Bookdown was the best choice (Xie 2016, 2023), and something that I was keen to learn anyway. I had already been writing manuscripts in Rmarkdown for quite some time, and bookdown was not a huge leap. I spent a lot of time during the late summer of 2022 writing, mostly during evenings and weekends (I had other things to do during the workday), but I only made it through a draft of about a third of the workbook before the autumn semester started and my time was consumed by more pressing matters. Nevertheless, it was a good start. Having taught the class for two years, I realised that the first week really should be devoted to teaching three topics: (1) review of basic mathematics, (2) how to organise datasets and files on a computer, and (3) how to work with spreadsheets. In the past, we assumed that students had sufficient background in these topics, but this wasn’t always the case, meaning that many students felt a bit lost early on in the class. These topics became the first three chapters of the textbook, which is the content for the first week of my class. After a few more weeks of content were written, I put the writing on hold for the autumn semester.
When the autumn semester ended, I had reason to panic. I had resolved to rewrite the class from scratch, but I didn’t have even half of the learning material complete. I got back to work in December, writing statistics learning content whenever I could find the time. I still didn’t manage to take any time off, save for 24 and 25 December, which fell on a Saturday and Sunday that year. Outwith that one weekend, I was always working, and I continued working seven days a week until a very brief pause that summer. I got a bit more written alongside other tasks that needed to be completed in teaching, research, and administration during that time. The spring 2023 semester arrived in January, and I had enough material banked away to be comfortable, at least for a few weeks. The online workbook looked good, at least the bits of it that were written. This was entirely due to Bookdown being such a useful R package. It looked good viewed from a browser on my computer, but it also looked good on a mobile phone, which is how I figured a lot of my students would want to read it. The navigation between sections was easy, there was a search function built-in, and there was a pull-down menu that allowed readers to change the font type, font size, and the colour scheme. This seemed really beneficial for accessibility; students could make the font as large as they needed to and generally adjust the settings to read comfortably. The figures had alternative text. The format seemed almost perfect, but I also thought that an audio version would be useful. I figured that reading the workbook aloud would help me spot a lot of typos and bad writing (I was right about this), even if only a few students had any interest in the audio version (I was wrong about this – a lot of students used it). I got to work reading each section aloud. The quality was not the greatest at the time, but I could at least post the audio files for the students to download.
Week by week, I continued to write. I would usually start writing on Friday early evening and finish a week’s worth of content by the end of the evening on Sunday. It wasn’t a healthy work cycle, but there was something satisfying about starting with a blank slate every Friday and having a complete draft by the end of the weekend. I would often think throughout the week about the best way to approach the next topic on the agenda, such as hypothesis testing, ANOVA, or regression. I wanted to lead with concrete examples as much as possible, and with datasets that I thought would be accessible to my students. I wanted the writing to be interesting and engaging. At the start of each weekend, I would search through the peer-reviewed literature on the relevant statistical topic and look through old statistics textbooks that I had accumulated throughout the years to try and fill out my understanding of the topic as much as possible. It was always satisfying to re-learn these topics, and to discover new nuances and slightly different ways of thinking about them. By the end of the semester, sometime around April 2023, I had somehow accidentally written a textbook. Was that really necessary?
In any case, the semester had gone very well. Jamovi was an absolute success. It was so much of an improvement for student learning that whenever my students from previous years would get in touch to ask for statistics help (for other classes or their dissertation research), I would tell them to switch over to jamovi immediately. At this point, we had new learning content and a solid foundation for future semesters. There were adjustments to be made, of course, but nothing went terribly wrong.
It was around this time, perhaps a few weeks before the end of the semester, that I seriously considered making the textbook available to people outside my own institution. It wasn’t ready yet. The content was too specific to my own university and class, so some more work would need to be done. But given how much work had been done already, I figured it would be worth it on the off chance that someone else would be in the same position that I was a year earlier. Jamovi was still relatively new, and I thought that other university teachers might be more inclined to start using it if there was a viable textbook available.
Publishing a hard copy
My original plan toward the end of the spring 2023 semester was to make some hefty edits to what I already had and publish the book on bookdown.org and my own website. This seemed well worth the effort given how much work I had already done, and I felt that I really should contribute something to the jamovi community given how useful jamovi had been for my teaching. I wanted to make sure that the book would be freely available online and as an audiobook for any educator considering making the switch to jamovi. In a 2017 interview, Jonathon Love, co-founder and developer of jamovi, remarked on the need for content creators to provide supporting materials for the project:
“People are reluctant to adopt a new platform when not all the supporting materials, videos, textbooks, etc. have been created yet. At the same time, the content creators are reluctant to provide supporting materials, because people seem reluctant to adopt it. In this way, markets tend to resist change, and overturning the status-quo often poses a frustrating challenge.”
I noticed that a lot of bookdown books had print versions published by CRC Press. I had bought and read hard copies of CRC Press books on R programming and data analysis before, and I saw that a lot of CRC Press books were freely available in bookdown form, but I never quite understood the details of how this worked. That several very good books were available for free online and as print versions intrigued me, and I thought that it was worth getting in touch with the publisher on the off chance that they would be interested in what I had written and could disseminate the book to a broader audience. I sent an email to the publisher in late April 2023. I was very pleasantly surprised when I received a positive response inviting me to send a more formal proposal to go out for review.
This was new territory for me; I had published plenty in peer-reviewed journals, but never a book. My first experience was excellent all the way through. I wrote up a proposal and sent what I had for the book. I received three constructive and positive reviews, and this led to an agreement for drafting a full book by February 2024. This gave me about half a year to make edits and additions, and to remove some content that did not really fit with the overall goal of a textbook for jamovi. I was (and still am) enthusiastic about a publication model that allowed me to keep a free copy of the book online. The publisher even allowed me to retain all rights to the audiobook, so I could make a free audiobook version too. As far as I can tell, this is a win-win for everyone involved. CRC Press would be able to sell copies of my book, and I would benefit by having a version of the book to disseminate in print. I would also benefit by having a copy editor read through the book and suggest edits, which would fix any inconsistencies and errors, and this would feed back to having a better textbook to work from in future years of teaching.
February 2024 came around, and I submitted a full draft to CRC Press. I was actively teaching statistics at the time and using some of the new material I had written. Teaching each week made it easier to spot errors and inconsistencies, and to make small improvements to the writing based on informal feedback from students, teaching assistants, and academic colleagues. I wrote these down; the next phase was working through copy-edits where I could make further changes. This was in June 2024, and included new formatting edits and indexing. I was surprised at how many typos there were still left for me to correct.
During mid June, I attended a workshop on teaching statistics in Manchester, England, which was immediately followed by the UK Conference on Teaching Statistics (UKCOTS). I had never been to a conference focused on teaching statistics, and I wanted to share my new book with colleagues, hear some new perspectives and insights, and get some ideas for future teaching. The workshop and conference did not disappoint! I had an incredible time, saw some outstanding talks, and finished my time in Manchester with a wealth of new plans for improving my own teaching for spring 2025. I also met other statistics teachers who use jamovi. Someone introduced me to the shinylive R package. I wish I had written down their name (if you are reading this, thank you again!) because shinylive gave me the perfect solution to a problem that I had been ignoring for a long time. The shiny apps that I had written to complement the book were being hosted by the Posit shinyapps server. This was perfectly fine for a class of 150 students, but if more people needed to use the apps, then the cost of hosting them online might become more than I could afford. The shinylive package completely removed the need for a hosted server by having everything run from within a browser. I could just post the apps on GitHub in the same repository as the book. This was a huge relief. In fact, it was such a huge relief that my brain began to search for something new to worry about on the way home from Manchester. I found a new worry in the rights to a couple previously published images (Figure 30.1 and Figure 35.1) that I had used for two papers in the American Naturalist (Duthie et al. 2015; Duthie and Nason 2016). Fortunately, a quick email exchange on the train home confirmed that I did not actually need permission to use these. Everything was fine.
Soon after returning from Manchester, I received the typeset proofs and again had a chance to make some final edits. I re-read the book (now nicely typeset), and I was again surprised, and at this point a bit horrified, by some inconsistencies and errors that still needed to be fixed at this stage. Again, the publisher was excellent and reassuring throughout the entire process. There were no problems making the necessary edits. Nevertheless, I found it very easy to second guess myself at this stage, and I was nervous that time was running out to find any errors. I went through four rounds of proofs, with my edits being a bit more pedantic each time until I really was satisfied that the job was complete (I was quite tired of reading my own writing by this point in time). The final version was sent to print on 9 August 2024.
Next steps and conclusion
It is a huge relief to have finished the book. I now look forward to seeing it in print in October 2024. My goal is to complete the audiobook in the next couple of months. I will also need to integrate some of the improvements that I have made to the book into my own teaching materials for spring 2025.
Completing this book coincides with what I hope will be the end of a long period of instability and exhaustion, which began with the start of COVID-19 and a new academic role. There is a lot that I have left out of this story. The book has been mostly a side project of mine, and it usually, rightfully, took a back seat to more pressing matters of teaching, research, and administration. Nevertheless, I now find myself, at least temporarily, with time to reflect. After years of continuous stress that seemed like it would never end, there is some recovery. Not every evening and weekend needs to be spent working anymore. I look forward to working hard without harming my health.
I hope that people find the book useful. I will continue to make new updates here.
References
Duthie, A. B., Abbott, K. C., & Nason, J. D. (2015). Trade-offs and coexistence in fluctuating environments: evidence for a key dispersal-fecundity trade-off in five nonpollinating fig wasps. American Naturalist, 186(1), 151–158. https://doi.org/10.1086/681621
Duthie, A. B., & Nason, J. D. (2016). Plant connectivity underlies plant-pollinator-exploiter distributions in Ficus petiolaris and associated pollinating and non-pollinating fig wasps. Oikos, 125(11), 1597–1606. https://doi.org/10.1111/oik.02629
Navarro DJ and Foxcroft DR (2022). learning statistics with jamovi: a tutorial for psychology students and other beginners. (Version 0.75). DOI: 10.24384/hgc3-7p15
Xie, Y. (2016). Bookdown: Authoring books and technical documents with R markdown. Chapman & Hall/CRC, Boca Raton, USA. https://bookdown.org/yihui/bookdown
Xie, Y. (2023). Bookdown: Authoring books and technical documents with R markdown. https://github.com/rstudio/bookdown