Word or LaTeX typesetting: which one is more productive? Finally, scientifically assessed

One of the main goals for anyone in the Academia, for any research group, is to publish the results of their research. Therefore, enormous amounts of time and effort are devoted to producing manuscripts which will be evaluated for publication in scientific journals or international conferences. Being this goal so central to the career of many people, it is more than reasonable trying to employ the right tools to achieve the best results in writing, measured in terms on quality vs. time devoted to writing.

Figure 1. Personal view of the author: Word is preferable for not-too-complex documents. Defining what does “complex” exactly mean is one of the goals of this post  | Credit: JL Blanco
Figure 1. Personal view of the author: Word is preferable for not-too-complex documents. Defining what “complex” exactly means is one of the goals of this post | Credit: JL Blanco

We can take for granted that the reader knows Microsoft Word, but perhaps some of you are not familiar with that LaTeX thing. LaTeX (yes, it is always capitalized like that) is a sort of “programming language”, a markup language actually, pretty much like HTML. So, if one want to write a bold sentence, instead of selecting with the mouse and clicking, one must type a specific command around the sentence. The same applies to any other document element: images, tables, section headers, etc. there are commands that must be memorized for every element. It may seem cumbersome. Indeed, it is. It is really cumbersome, at the beginning. I stillr remember how much I protested when a Professor forced me to elaborate an assignment in LaTeX during the PhD courses. But once one overcomes the (steep) learning curve of LaTeX, the realization of its advantages means that there is no turning back:

  • Mathematical equations of unbeatable quality.

  • Easy generation of all kind of indices, bibliography, cross references, etc.

  • It allows you to focus on the text content, instead of its appearance. In theory, a text prepared according to the guidelines of a specific journal may be made to fit another’s style by just using one external style file. In practice, sometimes one finds some incompatibilities, but in all cases it is orders of magnitude faster, cleaner and easier than manually changing all the formatting details in Word.

At least, these are the advantages that we, part of the researchers, see in LaTeX: the call to pick among Word or LaTeX has become one of those endless, “religious” discussions. One support one or the other just like people support different football teams.

But finally here comes a space for objectivity in this debate. A group of researchers on Experimental Psychology from the University of Giessen (Germany) has addressed the problem by measuring the pros and cons of each typesetting system 1. Next we summarize the experimental setup and their findings.

The tests

The experiment consisted in asking a group of volunteers to typeset a number of texts using either Word or LaTeX, provided they had 30 minutes to end the task. In order to characterize the performance for all kinds of texts, the researchers proposed three different texts: (1) a long, continuous text, (2) a text with some tables and (3) one plenty of equations (as one typically found in Mathematics, Physics, Engineering…).

Regarding the volunteers, they were 40 users selected such that there were 10 experts and 10 novices in each typesetting system (Word and LaTeX). The intention was determining whether the efficiency improved or not as users gain experience in each system.

Afterward, the researchers measured the amount of typeset text and the number of mistakes. Text length was such that only about 90% of experts could type it entirely. Ah! The motivation of volunteers was boosted by means of prizes of up to 150€ for the best one in each category.

The results

We start with the manuscripts comprising just long blocks of continuous text, in which is clear that Word is unbeatable: even the (relatively) novice Word users perform really far better than LaTeX users in both, the quantity and the quality of the typeset text. Regarding the existence of orthographic errors, we must highlight that LaTeX users were allowed to use their favorite IDE with any desired plugin or additional tool, thus their errors cannot be attributed to the intentional privation of any of those helps.

Figure 2. Results for long, continuous texts | Credit: Knauff & Nejasmic (2014)
Figure 2. Results for long, continuous texts | Credit: Knauff & Nejasmic (2014)

Regarding table editing, Word wins again… and this is not big news to anyone who ever tried to enter a table in a LaTeX document. How can something so easy in a WYSIWYG editor and so incredibly complex in LaTex? How many visits to TeX – LaTeX Stack Exchange does it take to create a decent LaTeX table?

As can be seen in Figure 3, the study found that it is not only less efficient to typeset tables in LaTeX, but that the number of mistakes skyrocketed as well. However, the large standard deviation in LaTeX mistakes reveals that it is possible to make it right… but there are so few people capable of such an achievement!

Figure 3. Results for texts with tables | Credit: Knauff & Nejasmic (2014)
Figure 3. Results for texts with tables | Credit: Knauff & Nejasmic (2014)

And finally we get to LaTeX comfort zone: texts with plenty of equations. In this last experiment, it is demonstrated that even untrained LaTeX users make fewer mistakes and are more productive than expert Word users.

Figure 4. Results for texts with equations | Credit: Knauff & Nejasmic (2014)
Figure 4. Results for texts with equations | Credit: Knauff & Nejasmic (2014)

Apart of measuring the efficiency of typesetting in such an objective way, the researchers also asked the volunteers to fulfil a usability questionnaire (results shown in Figure 5), from which we must highlight the emotional aspect: LaTeX users claim to enjoy much more than Word users while writing (5.2±1.4 vs 3.6±1.7), while also reporting feeling less frustration and tiredness.

Figure 5. Results of the usability questionnaire | Credit: Knauff & Nejasmic (2014)
Figure 5. Results of the usability questionnaire | Credit: Knauff & Nejasmic (2014)

These findings are coherent with the well-known popularity of LaTeX among researchers of the more technical areas, where writing equations is typically a significant effort if attempted to do in Word. I would like to remark (making clear that this is a subjective opinion) that aesthetics is another important factor: Word equations are far less “elegant” than those obtained from LaTeX, which “look” totally professional.

Quantitatively, we can point out another work 2 which asked the Editors of several Scientific Journals for the prevalence of each format in received manuscripts. They found that LaTeX rules in the areas of Mathematics (97%), Statistics (89%) and Physics (74%), then followed by the group of Computer Science (46%) and Astronomy-Astrophysics (35%), with the rest of areas totally dominated by Microsoft Word.

References

  1. Knauff, M., & Nejasmic, J. (2014). An Efficiency Comparison of Document Preparation Systems Used in Academic Research and Development. PloS one, 9(12), e115069.
  2. Brischoux, F., & Legagneux, P. (2009). Don’t Format Manuscripts. The Scientist, 23(7), 24.

17 Comments

Post a comment

Daniel Manzano

I really think that both programs are to different. As it is stated in the post, for writing long texts Word is unbeatable, but for equations it is the other way around. The discussion comes when you have a long text, few equations, and a lot os references and figures. In this case Latex is better, because it manages the bibliography in a better way, but the time needed to learn how to use Latex is also much longer and it is not clear to me if it is worthy.

What is absolutely true is this sentence: “the call to pick among Word or LaTeX has become one of those endless, “religious” discussions. One support one or the other just like people support different football teams.”

The same happens with Linux VS Mac VS Windows…

Ines @inesgnInes @inesgn

This is an interesting study but I think the Word/LaTeX dilemma will soon be overcome in many fields. For example, for data analyses made in the free R environment one should seriously consider using knitr and R Markdown. It has both the benefits of dynamic reports (if you change a single data point all tables and figures are authomatically updated for you) and easiness of use. No need to know LaTeX to get comparable results. And of course it’s infinitely many times better than using Word, which is a text editor and nothing more.

AnonymousAnonymous

This study was very effectively rebutted by Daniel Lemire some time ago. It does not test tasks that are representative of what scientists do, and it does not address the difference in output quality of word vs. LaTeX.

Erik EngheimErik Engheim

I don’t see how you can test anything of interest in 30 minutes especially for something like LaTeX. In 30 minutes you are not even remotely going to reap benefits from using LaTeX over Word. If I decided to use LaTeX right now I’d probably spend at least 30 minutes just setting up templates, add ons etc.

LaTeX shines when you work on large documents over several days. The ability to split into multiple files and version control it with great version control systems like git is a real benefit. Also as you write more you can benefit from easily copy pasting and modifying styles or things you need.

Also I don’t get how LaTeX user could bomb so badly on tables. If you really need to have control of this you can use a GUI tool which generates the LaTeX for you. Lots of editors has such capabilities. But for a 30 minute assignment most might not consider it worth it to use any special tools to make a table.

DOGMA1138DOGMA1138

Well office 2013 and onward comes with a built-in TeX interpreter, previous version has plenty of addins that did that same so it’s not really a question any more.

Word offers the easy high level editorial capabilities and low level TeX syntax editing when you need it.

Ross SniderRoss Snider

Typesetting is the final act in publishing. Writing is where the bulk of the effort lies. LaTeX addresses 10% of the effort effectively (typesetting) and is cumbersome and gets in the way for 90% of the effort (writing). This is especially true for editing and collaboration where Word’s track changes is required.

Jon ForrestJon Forrest

“untrained LaTeX users make less mistakes and are more productive than expert Word users” ->
“untrained LaTeX users make fewer mistakes and are more productive than expert Word users”

andycandyc

I wonder what exactly qualifies as a “Word Expert” here? Many people who use Word regularly aren’t typing in a lot of formulae, so are likely to do it in a clumsy way via the mouse and find it unnecessarily difficult. Indeed most of the students I knew who were using Word for documents containing math were surprisingly unaware of Word’s Math Autocorrect feature, which lets you type in a very LaTeX like format. If you use it often, papers containing math are basically on par with type in ordinary text documents.

Tim Scarfe

I didn’t read the paper but it seems to me like there are some obvious methodological flaws. As pointed out later on; to be an expert in LaTeX – one is also likely to be highly trained in Mathematics. Tables are indeed the most annoying thing in LaTeX but it’s just a markup language! The real problem here is that there is no better table handling in some of the popular editors like TexStudio. Also — in case it wasn’t obvious; it does make sense to separate the structure and content of manuscripts from their presentational style in much the same way web authors do with HTML and CSS. Most academics want to get their work published in a context where its formatting will be standardised. LaTeX and git is a really powerful way of collaborating and managing revisions on manuscripts — would you trust Word to do that? I didn’t think so.

Mark EllisonMark Ellison

It also misses out another important point, vital in maths and computer science, but also increasingly in other fields: how easy is it to use macros and/or construct parts of your document by script? I had lots of tables for my PhD (computer science + linguistics), so I constructed macros to help me write them in a meaningful way. I’ve also used PERL (in the past) and PYTHON (more recently) to transform data straight into LaTeX tables.

Owe JessenOwe Jessen

I’m a bit surprised, as Word has a formula editor since Office 2007 which is nearly identical to LaTeX, that the latter has such a profound advantage.

M SingerM Singer

When last I visited this comparison, the issue we had with Word had nothing to do with user mistakes or relative ease of use. The core problem we encountered was more sinister and debilitating. As our manuscripts grew, the likelihood increased that Word would drop portions of the text, lose or change formatting, crash, or in some cases refuse to open a working document. As you can imagine, a lost manuscript was a substantial defeat for Word. Some chose to break their manuscripts into chapter documents to reduce their risk of outright loss, but this tended to make the inter-document references more complex. Our conclusion, was that if you wanted to graduate on time, better to depend on LaTeX.

WojtekWojtek

I wrote my MSc and Phd (Physics) in LaTeX (ca. 90′) . Also convinced my girlfriend to do the same.

What did I gain? Peace of mind. I just did not care anymore about formatting, numbering, fonts, etc. because I did not know LaTeX enough to modify them. The numbering was always right, even if inserting a picture was a nightmare.

I now work in a company which uses Word and when thing go well, they go well. When things go wrong, they go terribly wrong. But what the heck, the tracking mode is fantastic. (please do not tell me about diff for LaTeX – I am an uber-geek and administred unix for many years, I would not even remotely approach tracking with LaTeX.

But this is not very much relevant anymore, the future is with Web publishing in a collaborative mode.

paolo gaipaolo gai

I would also suggest to add lyx (www.lyx.org) to this trial… it is more intuitive than latex but still has most of its advantages!

hmdzhmdz

Does Word papers result in overall accuracy?

I have written many ISI papers in MS Word. Yes the writing part and formatting were easy, but during revisions I always came up with hundreds of mistakes and headaches:

– Wrong equation numbering
– Missing headers
– addition of new sections –> required entire document modification!
– addition of new images would deform most pages, –>required entire document modification!
– adding footnotes was odd
– writing simple inline math required many clicks
– adhering to ISI formatting standards was difficult
– choosing proper fonts was subjective

Each technology targets specific audiences. Latex targets academic and researchers while Word targets the rest of the society.

DrCommandoDrCommando

If you write a lot, LaTex is definitely worth the effort. Clicking in Word requires to know the menu. Certainly knowing the menu is considerably less effort than knowing the commands in LaTex. Hence, it is clearly a question of proficiency. If you suck in LaTex, LaTex is horrible.

After I reached a certain skill level in LaTex I started to use it for plotting and drawing figures as well. MATLAB and other math plot programs just cant hold up to the standards LaTex produces – Word does play in this league at all.

In fact LaTex is the most complete solution.

9 Trackbacks

[…] ¿Merece la pena el esfuerzo de aprender LaTeX o con lo que sé de Word me apaño para lo que tengo que hacer? Esta es la pregunta que se hace cualquiera que necesite escribir documentos y dotarlos de un aspecto profesional. José Luis Blanco puede que aporte algo de luz con Word or LaTeX typesetting: which one is more productive? Finally, scientifically assessed […]

Leave a reply

Your email is never shared. Required fields are marked.

Required
Required
Required

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>