By *Jelle Hellings*.

Posted at 15 October 2024, 11:46.
Last updated at 15 October 2024, 23:36.

Last Wednesday, a publisher representative paid me an unannounced visited at my office to talk about the course textbooks (and supportive materials) they are offering for some of the courses I teach. This visit coincided with me already reflecting on the course textbooks I am using in my courses, specifically with respect to their *value*, as perceived by students, and the *costs* their usage imposes on students.

Textbooks can have *immediate value* (they cover the content of a course) and *long-term value* (they are useful after the course, e.g., as reference). As an example, I can take a look at the textbooks that were required for *my own* Bachelor and Master, which I finished a while back (hence, I believe I can fairly evaluate both their immediate and long-term value).

Unfortunately, most of the course textbooks I own ended up having low *long-term * value. The vast majority of my textbooks only saw usage in a single course and have *never been used* after finishing that course. A handful of books have seen use in multiple closely-related courses (e.g., an introductory course and an advanced-level follow-up course). Only *two* books I still regularly use (as reference material) and only a few more books have seen repeated use over the years.

As most of my textbooks only saw usage in a single course, one would expect at-least good *immediate* value (in that course). This is not the case, however: many of my textbooks *barely saw usage* in the courses that required them. First, the instructors of many courses provide excellent supporting course materials (e.g., slides, notes, and model solutions). In many cases, these materials were more than sufficient to succeed in their courses. Second, several courses used only a fraction of their textbooks. Finally, some courses even used *redundant* textbooks: textbooks that cover identical material as textbooks I already owned due to other courses.

For example, my Bachelor program had two related courses, one focused on *probability*, the other on *statistics*. These two courses had different required textbooks. The probability course required *Probability and Statistics for Computer Scientists* by Michael Baron and the statistics course required *Introduction to Probability and Mathematical Statistics* by Lee Bain and Max Engelhardt. These books are rather different in their presentation (the first one is more applied to Computer Science, the other more focused on the underlying mathematics). Still, these two books have a large overlap in content, as they cover very similar topics. When looking at the course objectives, either book could have easily been used in *both* courses. Hence, one of these books is clearly *redundant*. Furthermore, both courses only covered a small portion of their textbooks. Finally, I have never used either book *after* their respective courses: both books are currently in a moving box.

My experiences are my own and do not represent the experiences of others. Still, I regularly receive similar feedback from my own students: plenty of my students have expressed that they barely use their textbooks. For example, for the courses I teach, they instead rely on notes they make during the lectures, the slides and supporting notes I provide, my example problems and model solutions, and what they learn from the assignments.

The potentially-low value of textbooks could be an explanation as to why many students barley use required course textbooks. The role of the *cost* of textbooks in the decision-making process of students cannot be understated, however.

Textbooks for undergraduate and graduate courses, especially those on specialized topics, are *very expensive*: individual textbooks can easily cost 75-200$. As such, a year worth of textbooks can end up costing 750-2000$, a considerable amount of money. As argued in the previous section, the high price of textbooks does often not line up with their (perceived) value for students. At the same time, we should recognize that many students are struggling financially (e.g., due to the high costs of living and high tuitions), which has a well-documented adverse impact on the mental health and performance of students (see footnote). As such, the costs of course textbooks (and other study materials) are non-negligible when considering the barriers to quality education experienced by students from disadvantaged backgrounds.

To reduce the cost of textbooks, many students resort to delaying the purchase of textbooks (until they are convinced of their need or value) or skip buying textbooks altogether. As university courses typically cover *a lot of material* in very little time, these delayed-purchases will put students at a clear disadvantage in those courses that *genuinely* depend on their textbooks.

Instructors of university courses typically have a lot of freedom in the selection of course materials their courses require. Hence, it is good practice to regularly reflect on ones choices to see whether they are *still* in the best interests of students. This does not mean instructors should generally *avoid pricey* course textbooks, however: good textbooks can definitely be worth their price. For example, if I look at the collection of textbooks from my own studies, the two most-valuable textbooks I own are also among the priciest textbooks.

That being said, the usage of pricey course textbooks requires caution: instructors should assure that the *value* of these textbooks justifies their price. Indeed, courses with low-value yet pricey textbooks not only hurt themselves, but can also deter students from buying pricey textbooks in the future (thereby potentially impacting their performance in other courses).

Next, lets consider the kind of decisions instructors can make regarding their course textbooks to *minimize costs* and *maximize value* for students.

The first step of choosing the right textbooks is to figure out what the course requires from the textbook. Hence, an instructor can only select appropriate course textbooks *after* determining the environment in which the course operates (e.g., prerequisites and follow-up courses) and *after* determining the topics, the learning objectives, the expected learning outcomes, and the teaching methods of the course. Each of these aspects impact the course textbook (and other materials):

- What are the prerequisites and prior knowledge expected by the course?

*Impact*. Do the prerequisites and prior knowledge line up what the textbook assumes as prior knowledge? Does the textbook depend on notation and terminology the students are familiar with? - What topics will the course cover?

*Impact*. Does the textbook cover these topics at the level of detail required by the course? - What knowledge are students expected to have mastered (and to what degree) after completing the course?

*Impact*. Does the textbook cover this knowledge? Does the textbook support independent study of this knowledge, e.g., via self-tests or via online quizzes? - What skills are student expected to learn and practice during the course?

*Impact*. Does the course material help with practicing and learning these skills, e.g., via integrated suitable exercises, challenges, problems, and project ideas? - What teaching methods will be employed to teach the course?

*Impact*. Does the writing and structure of the textbook support these teaching methods? Does the textbook come with supporting material suitable for the teaching method (e.g., lecture videos, slide decks, quizzes, or solution manuals)? - Which other courses depend on this course? Are the prerequisites of these follow-up courses consistent with the topics, knowledge, and skills this course focusses on?

*Impact*. Can the textbook be used in these other courses? Can the textbook be used as a reference in these other courses? Does the notation and terminology of the textbook line up with what students will see in other courses?

Now assume we have determined *exactly* what we need from a course textbook. Hence, we are ready to choose a suitable textbook. There are plenty of options to consider. Next, I will consider these options by drawing from my own experiences teaching courses on introductory programming, algorithms and data structures, and databases.

Not all courses need a textbook. For some types of courses, an extensive in-depth textbook can even be a distraction that will impede student learning.

Take for example an *introduction to programming* course. Students will not learn programming by reading: they need experience writing programs, making mistakes, and fixing these mistakes. Hence, the course material and lectures should help students to get hands-on experience as fast as possible by introducing and demonstrating elementary programming concepts (e.g., variables, assignments, if-statements, loops, and functions) and by demonstrating the software environment in which they will write and test their programs.

On the one hand, an extensive course textbook *can help and support* students when they write and test their programs: the textbook can help the student understand what they see happening in their first programming experiments. On the other hand, programming is a rather foreign concept for many students, due to which some students have an initial fear of making mistakes. To avoid making mistakes, some students will *delay hands-on experience* by extensively over-studying their textbooks.

Hence, for introductory hands-on courses, the right course material might not be a course textbook. For these courses, slides (or elementary notes) together with an extensive set of training exercises, examples, model solutions, and assignments with detailed feedback could be a better choice.

Several organizations are considering the adoption of *open educational resources* to reduce the barriers to quality education experienced by students from disadvantaged backgrounds (see, e.g., this recommendation by UNESCO). Open-source or free textbooks are not an end in themselves, however: the choice of textbooks should always be in the best interest of the student, even if that means a non-free option.

Open-source materials are the most flexible to adjust to the specific needs of a course. For example, an instructor can remove the parts the course does not need (to keep the material focused and not overwhelm the students with irrelevant topics), can add missing content (if necessary), can adjust the material to ensure the notation and terminology is in line with what the students are familiar with, and can integrate suitable exercise sets that align with the course objectives.

Earlier, I argued that an extensive textbook for an *introduction to programming* course could be a distraction that impedes student learning. Indeed, we want students mainly to gain experience writing programs instead of studying a textbooks.

A decade ago, I was involved in an introductory course. We decided that a detailed textbook was not necessary, but a hands-on textbook to *support* students was welcome. The lectures of that course would serve to introduce all important concepts. Hence, we choose a hands-on textbook that would support the students to get used to the programming environment. For this purpose, we successfully used the open-source book

How to Think Like a Computer Scientist by Peter Wentworth, Jeffrey Elkner, Allen B. Downey, and Chris Meyers.

As the book covered many topics outside the scope of our introductory course, we did massively trim the book to only cover the relevant topics.

The open-source book

Open Data Structures by Pat Morin

covers all elementary data structures one would expect in an introductory data structure course (e.g., arrays, linked lists, heaps, binary search trees, red-black trees, hash tables, and a few more). Unfortunately, as the title suggests, the book mainly focusses on *data structures*, and does not cover many algorithms.

The free (but not open-source) book

Algorithms by Jeff Erickson

is an excellent resource that covers many algorithm topics, including the elementary search, sort, and graph algorithms one would expect from introductory courses and topics one would expect to see in advanced course.

Both books are excellent in their own right. Unfortunately, neither book covers all topics one would expect from an introductory course that covers *both* data structures and algorithms. Hence, for such a use case, neither book is a perfect choice.

Using an established traditional non-free textbooks certainly have benefits: good non-free textbooks come with a plethora of support material (e.g., slides, instructor guides, video lectures, or online learning environments). Furthermore, on specialized topics, a traditional textbook might be the best or only choice. To *maximize value* and reduce the cost-impact of non-free textbooks, instructors can implement several measures.

Students might have (digital) access to the textbook of your choice via the library. If this is the case, then that could make the book effectively free for the student.

If a course is related to other courses series (e.g., it is a follow-up or has follow-ups), then it is a good idea to coordinate the choice of textbooks across these related courses: a single textbook might serve the needs of all of these courses.

Such a single *multi-course* textbooks has obvious benefits: the student only has to buy a single book and one eliminates the risk of *redudant* books (reducing costs), students will end up using more parts of the book (increasing value), and students will only have to familiarize themselves with the structure and style of the book once (potentially improving learning). Furthermore, the usage of a single book across courses makes it easier for students to see connections between courses (instead of seeing each course as a stand-alone unit).

Within Computer Science, a typical multi-course textbook is *Introduction to Algorithms * by Thomas H. Cormen, Charles E. Leiserson, Ronald L. Rivest and Clifford Stein. The breadth of topics covered in this book makes it a good candidate for a series of courses on algorithms and data structures (e.g., an introductory course, an advanced algorithms course, and a specialized topics course). In addition, the book is an excellent reference for most common algorithms and data structures. That does not mean the book is without flaws, however: for a novice student, the book is certainly not an easy read and the problems and exercise sets are harder than those in dedicated introduction-only books.

As a second example of a multi-course textbook, I will step outside of Computer Science. Before I switched to Computer Science, I was briefly enrolled in a Physics program. This program extensively used *University Physics with Modern Physics* by Hugh D. Young and Roger A. Freedman. This behemoth of a book was among the most expensive books I had to buy. Still, this book was worth the money (especially if I had continued with Physics): the book covers a vast portion of the basics of a Physics undergrad program and, as such, the book was used by multiple physics courses in the first and second year of the program.

One obvious way students can save money is by buying second-hand books. The use of second-hand books should especially be considered or supported if the course textbook has limited long-term value, e.g., the book is not used in other courses and is not designed to be used as a reference.

Unfortunately, some publishers are hostile toward second-hand use. For example, some books that come with online support and learning materials lock these online components behind a single-use key. Hence, students that obtain second-hand copies will not have access to these online components. Instructors should be aware of the existence of these materials and inform students whether these materials are necessary (e.g., if the course depends on these online components).

Additionally, non-free books change over time: new editions can drastically change the covered content or the exercises and new prints typically make minor changes (e.g., to exercises). Hence, to fully support second-hand books, it is important to be aware of these changes and to accommodate students with multiple versions of the books. In plenty of cases, new editions are minor updates. In these cases, it is easy to support multiple editions. This is not always the case, however: sometimes new editions seem explicitly designed to kill the second-hand market:

A while ago, I was teaching a course for which the textbook got a new edition. This new edition of the book did not make any changes *except* that all exercises were replaced (and the general printing quality took a nosedive). Unfortunately, our course depended on these exercises: they were intensively used in the tutorials. Hence, switching to the new version would prevent second-hand use of the old book. In addition, we had several students that failed the course the previous year and re-took the course (and, hence, had the old edition of the book).

As forcing everyone to use the new book would unnecessarily burden students (especially those re-taking the course), we decided to switch to book-independent exercises as a short-term solution. These book-independent exercises ensured that any edition of the book could be used. Later, I opted to no longer use that textbook.

Open book exams are a double-edged sword. On the one hand, students typically like having open book exams: for many students, having the *option* to look up things in their book during the exam significantly lowers exam-related stress. On the other hand, by making exams open book, some students will end up buying a last-minute copy of the course textbooks. The reason for that is simple: if the option exists to bring a textbook, many students will feel pressure to do so *just to be sure*.

For some courses, an open book exam is the right choice. In my courses, books are not that useful during an exam, however. Indeed, whenever I allowed books in the past, plenty of students brought their books, but almost no one used their books. Days before these open book exams, multiple students did ask me whether they should buy a physical copy of the book (or whether they could bring copies of parts of the book). It is likely that some of these students ended up buying the book, brought the book to the exam, and never used the book.

There are several valuable alternatives to an open book exam that equally provide students with peace of mind. First, one can design exams with an included *reference sheet* that contain all the important facts a student might find useful to look up. One can even share this sheet with the students before the exam to make sure they know what to expect from the exam. Second, instead of providing a reference sheet, one can allow students to write their own *cheat sheets*. Allowing students to write their own cheat sheets not only brings peace-of-mind to students (as they can write down those things they easily forget in a high-stress environment), but also serves as a good exercise for student to properly review the course material (and, hence, decide what to write down), thereby supporting their learning process.

Using non-free books benefits the author of these books and if money is involved, so are unethical incentives. Hence, it is important to remember that whichever book is chosen, the choice should *always* be made in the best interest of the students. This double so when an instructor uses or requires textbooks written by themselves (or by personal or academic relatives): instructors should avoid any appearance of a conflict of interest *unless* no other options are available.

I do use my own textbooks in some of my courses. I do not require students to buy these books, however: they are freely available via the library.

Courses evolve over time, both due to instructor experiences and due to external forces (e.g., changing requirements by the overarching program, by society, or by industry). For example, an introductory programming course could very well be using Pascal three decades ago, using Java two decades ago, and use Python nowadays. Furthermore, textbooks change over time and new and better textbook options might become available. Hence, it is a good idea to regularly revisit the design of the courses one teaches and, while doing so, also revisit the used course textbook.

Relations between financial insecurity, mental health, and student performances can be found in various studies (e.g., Johnson, Maynard et al., Broton et al., and Becerra et al.). As I am not an expert in the field of student welfare, I cannot verify the quality of these studies, however. Still, even on its own, it sounds reasonable that stress factors like those caused by financial insecurity impact health and student performance.

Some people will argue that they were able to work alongside their studies to afford their studies and that current students should to the same. Unfortunately, tuition and the general cost of living have increased much faster than wages in most places, due to which summer jobs or part-time jobs come far short of these expenditures. Furthermore, for students that *do manage* to find well-paying jobs, the demands of their studies are often incompatible with long work hours. For example, for many engineering students, keeping up with their studies is already a more-than-full-time job.

I also note that *personal experiences* do not necessary reflect the experiences and struggles of most students. For example, those students I studied with over the years all finished their studies with distinctions and on time. My experiences are not reflective of the general student population, however: the *average student* in my program already had some delays (due to failing courses) after the first year. As I had no delays, the group of students I studied with shrank over time and had a massive survivorship bias toward excellence.

Similarly, it is likely that the experiences of *average* professors are unrepresentative for the average student. First, ending up as a professor already hints at an above-average interest in their field of study. Furthermore, it is well-documented that professors disproportionately have a socioeconomically privileged background (see, e.g., Morgan et al.).

## Join the discussion