On 12/27/2021 9:15 AM, Jose Ramom Flores das Seixas via lyx-users wrote:
I'm no expert in empirical research methodology, but I've learned enough to know that one needs to minimize variables. It seems to me that, in this study, the variables included not only the different systems -- Word and LaTeX -- but also different *users* using the different programs. I have to believe that at least some of the differences in the results have to do with differences in user typing proficiency. Some typists make more spelling and grammatical mistakes than others regardless of what software they use. To validly test the *software*, I would think that one would need to have same user use both systems. That, of course, would be problematic as it would difficult to find a single use with equal proficiency with both Word and LaTeX.Translated with www.DeepL.com/Translator (free version)I just read into it a bit and see no reason to look into it further: * It only compares performance in short texts *only short texts*. However, TeX's strengths are in large papers. For example, one could have set the task of making changes in a dissertation: Move sections, make systematic formatting changes, create indexes....If one had measured the expertise in using a program not only in hours, but also the sensible use of the possibilities, other groupings might have resulted - even experienced Word users rarely use style sheets consistently, for example. * *Style importance.* Character styles for highlighting (as they are common e.g. in InDesign and as macros in TeX) do not exist at all... But it is very easy to construct tasks that favor one system or another. With TeX it depends also still on the editor and how humans know to use it. * The paper is already from 2014. *old paper! Lualatex. *Since then, most LaTeX users have probably gotten used to LuaTeX with Unicode and OpenType, so that many coding problems no longer apply (although they may not have played a major role here). * It would have been interesting to include LibreOffice and ConTeXt as well. Maybe also layout programs like InDesign, Affinity Publisher, Scribus.... (Is LibreOffice still more stable than Word for extensive work?). Hraban
When testing LaTeX, a lot depends on the editing front-end one uses. I have to believe that a person using LyX would be more efficient than one using TeXWorks, if for no other reason that LyX shields the user from so much LaTeX source code. But, unless I missed it, I didn't see any mention in the study of any of the LaTeX users editing with LyX.
As to formatting "errors," I wonder what qualified as an error. It appears to me that the test was to duplicate the visual formatting of the original documents. If the original document has a 14 point heading, is it an error if LaTeX produces a 14.3 point heading? Is it an error if the resulting document has 1.5 pica paragraph indents instead of 2.0 pica indents? Were formatting "errors" found in the tables if LaTeX automatically adjusted a table width rather than making it match the width of the original document? My obvious point is that, if formatting "errors" are determined on such a visual-matching level, then it could take quite a bit of time to tweak a LaTeX preamble to achieve the desired results. This isn't a knock on LaTeX; it is simply a recognition that one doesn't generally choose LaTeX for the purpose of precisely reproducing the visual formatting of another existing document. Rather, when I choose LaTeX, it is usually because I don't *care* about the level of formatting details apparently tested by this study. I leave that up to the document class; that's the beauty of LaTeX.
If I were given the task of duplicating the visual formatting of an existing document, I would always choose a WYSIWYG program over LaTeX, not because it's better, but because it would be more useful for a visually oriented task. The Word users who could see the formatting as they worked would inherently have an advantage over the LaTeX users who would be working somewhat blindly until they compiled their documents.
In short, I have serious questions about the validity of this test. Virgil
-- lyx-users mailing list lyx-users@lists.lyx.org http://lists.lyx.org/mailman/listinfo/lyx-users
