Skip to content

LyX SuX

I’ve been meaning to move away from Word, and toward LaTeX, for a while now. A few weeks ago, I was just starting a new manuscript, so I thought I might as well take the plunge. Someone (I don’t remember who) told me about LyX, a ‘front end’ for LaTeX that, in theory, should make using LaTeX easier.

Why move away from Word? Because it’s a pain in the ass to use the automatic table of contents, figure and table numbering, bibliography, general section/subsection/subsubsection, and equation formatting features in Word to produce an acceptable academic (i.e., boring, not filled with useless chartjunk) manuscript, and I’m writing my dissertation. Which is to say, I will be needing a TOC, a bunch of numbered tables and figures, and various (sub)section/cross-referencing, and bibliography formatting done, and I want it to be easy and convenient. On the down side, almost everyone uses Word, so intentionally moving away could cause some (likely minor) problems, at least when it comes time to collaborate.

Anyway, the obvious choice once you decide to move away from Word is to move toward LaTeX. It’s widespread, the documentation is excellent, and the TOC, figure and table numbering, etc…, are orders of magnitude better than those in Word. It’s the obvious choice. I guess you could use OpenOffice, but that only solves problems with Word I haven’t mentioned here.

I was a bit worried about using a ‘raw’ LaTeX distribution, though I can’t now recall exactly why. I suppose I wanted to maximize both user-friendliness (i.e., WYSIWYG-iness) and all the formatting issues simultaneously. So I downloaded LyX.

It was easy enough to download and install the basic program, but it was less than simple to get the aspell spellchecker or, more importantly, various ‘extra’ format article/citation/reference formats to work.

The spell checker didn’t end up taking too long to figure out. It just required some extra downloading and copying of various files. Easy, but it’s stupid that the spell checker doesn’t install with the main program. Installing new LaTeX classes is not so easy with LyX, though. Technically, it’s easy to get the LaTeX distribution to recognize a new class, but it’s much less easy to get LyX to do so, as it requires an additional ‘.layout’ file that tells LyX how to present the .cls/.sty in the editor and how to process the ‘.lyx’/.tex file and make a pretty pdf.

If you’re not familiar with LaTeX, of course, not much of that makes sense to you. The point is that, with a plain old LaTeX distribution, it’s quite simple to copy three or four files to the right directly and then tell the TeX program to look around for new stuff. It’s substantially less simple to do that, plus figure out, with poor help files and worse online documentation, how to tell LyX what to do with the copied files. The point distilled still more: LaTeX takes n steps to use; LyX takes n+m, where m >= n.

I meant to write this post immediately after getting so frustrated with LyX that I gave it up (forever), but it’s been a week or so now, and I no longer feel the urge to explain in more detail what’s so bad about LyX. I gave it up and have been teaching myself how to use plain old LaTeX. It’s better in pretty much every respect.

To use LyX is to add a completely unecessary middleman,though I’ll admit that it has a few nice features: the equation editor is immensely better than that in Word (faint praise, I know) and the bibliography/cross-referencing works really well. But these two features don’t even begin to make up for the fact that introducing a new document class requires as much extra work as it does, especially when simply downloading and copying four files to the proper directories gets standard LaTeX to use a new class flawlessly.

A big part of the point of using LaTeX is to avoid worrying about typesetting/formatting issues and focus on the content of your writing. Trying to get the JASA LaTeX class (and associated files) to work with LyX required many hours of worrying about formatting and typesetting, and I never got it to work right. In a sense it was worse than that makes it sound, as the formatting I was worrying about wasn’t even entirely the formatting of the final document, but also the formatting of the document in LyX itself.

As mentioned, I gave up on LyX. I exported the draft I was working on as a .tex file, edited it in TeXShop (since LyX had, conveniently, thrown in some farewell garble, as it were), and everything’s been working fine ever since. LaTeX takes a bit of time up front to learn the basics, but the basics are really quite simple, and the results are wonderful. LyX takes more time up front, and the results are frustrating and unsatisfying.

On the off chance that [search engine] crawls by here and some wayward soul thinking of using LaTeX (I’m so very tired of typing it that way, but how supercilious can I be if just type ‘latex’ and you think I’m writing about rubber trees and allergies? I mean, one of the benefits of using LaTeX is feeling superior, right?) happens by after a [search engine] search, and they are dissuaded from using LyX, then maybe, just maybe, this week’s feeble attempt to keep my blog from finally dying will have been worth it.

Poetry by Briggs

It drives me nuts when people write that the result of a statistical test is ‘highly significant’ or ‘marginally significant.’ William M. Briggs writes wryly today: “Smaller p-values are usually accompanied with the claim that the results “stronger” or “more significant”. False, of course, but since everybody says so you will be in good company.”

The whole post is, as usual, worth reading.

[M]athlete’s foot[note]

In the new issue of Harper’s, there is a footnote in the main cover story that nicely sums up what a barrier to communication innumeracy can be:

And to say there is room for academic improvement at the school [The High School for Health Careers and Sciences in Manhattan] is a vast understatement. Only 58 percent of its students graduate in four years. Of all graduates, 41 percent leave with a full “Regents diploma,” which is conferred if a student scores 65 percent or higher on five subject-proficiency exams. A mere 3 percent of Health Career’s students graduate with an “advanced” diploma, which can be earned if they take three additional Regents exams and an increased credit load. The state average is 36 percent.

First of all, the state average what is 36 percent? Graduation rate? This would put ‘Health Career’ above average, but that’s neither here nor there, as it’s plausible (likely, even) that the average is so far below ceiling that the entirety of above average schools have ample room for improvement.

Maybe it’s ‘Regents diploma’ rate, instead, though I have no idea (and the article certainly hasn’t made it clear) whether ‘Regents diplomas’ are unique to this high school or a staple of New York state secondary education. Which is to say that I have no idea if there is a state average for “Regents diplomas” at all.

Or it could be “advanced” diplomas we’re talking about.

Consider, too, how the the percentages are presented. 58 percent of students graduate. Of these, 41 percent (i.e., 23 percent of the school’s students) get the “Regents diploma.” 3 percent of the students (i.e., 5.8 percent of those who graduate) get an “advanced” diploma. For good measure, throw in a 65 percent passing grade on a test, which adds no useful information at all, and maybe makes the passage even more confusing.

Of course, the fact that the reader has no idea what the last sentence of the footnote refers to and the fact that the writer casually switches from conditional to non-conditional percentages isn’t necessarily due to innumeracy. It could just be sloppy writing. At the very least, it reflects a cavalier attitude toward marshaling numbers in support of an argument. But is that any better than being innumerate? At best, such an attitude enables continued innumeracy in the reader. At worst, it fosters it.

Exercise: Show how the referents of the subject ‘it’ and object ‘it’ are resolved, thereby showing why, in context, the meaning of the clause ‘it fosters it’ is unambiguous.

Litmus test

Okay, so I’m a bit more than 3 years late watching the ‘reimagined’ Battlestar Galactica series, but I just finished watching the sixth episode of the first season - ‘Litmus’ - in which an independent tribunal is convened to investigate a cylon suicide bomber. The chief source of drama in the episode is the ongoing affair (against explicit orders to the contrary) between ‘Chief‘ and Number Eight, er, Lieutenant Valerii.

In a nice touch of ‘world building’, Chief invokes the 23rd Article of Colonization (i.e., the 5th Amendment), which allows him to withhold self-incriminating testimony.

Well, 2 + 3 = 5.

Coincidence? I think not.

80% new, 10% promised

It looks like I’ll be (at least) a week late with the promised new content. Naturally enough, since conferences are about 75% socializing and/or schmoozing and about 75% attending talks to try to locate the cutting edge of (math) psych research, I came home sleep deprived (on a statistically related note, I was recently told that tenure decisions are weighted according to the following scheme: 50% research, 50% teaching, 50% service, which clears up an issue I addressed briefly a couple posts ago).

In addition, all of the people I shared a room with had also just finished attending the Cognitive Science Society conference, so they were already sleep deprived, and one of them was sick with a cold or the flu or something closely resembling these maladies. I got sick, too, and am just now getting back on top of things.

The short story is that the conference was very good. I have extra motivation to get work done on some old projects, and I have plenty of ideas for new projects. More soon.

Hiatus

I’m off to the 41st Annual Meeting of the Society for Mathematical Psychology tomorrow. I’ll be giving a talk on three different models of decision bounds in multidimensional signal detection theory (a.k.a., General Recognition Theory, a name I have come to prefer to multidimensional SDT; more on this later).

I’ve decided to travel without my computer, so there won’t be any new posts for a few days. Of course, there usually aren’t posts for weeks (even months) at a time, so a few days without posts would hardly be noticed by a regular reader. Given the typical length of time between posts, the idea of a regular reader of this blog conjures up a strange (and kind of creepy) image.

I thought about taking my computer with me so that I could blog about the conference as it happens. Instead, I’ll avoid the hassle of porting a laptop through ’security’ at the airport and write about the meeting when I get back. I write this as something of a promissory note, then. Expect plenty of new material starting next week.

Peer review

A few months ago, I agreed to review a manuscript for the journal Language Learning. I printed and read the manuscript within a day or two of receiving it, and then put it aside. Although I would occasionally remember that I needed to re-read it and write a review, I didn’t do anything about it until I received an email from the editor reminding me (politely and professionally) of our agreement.

I put it aside primarily because I wanted to ponder the manuscript a bit. It wasn’t particularly good, but I felt that it had some value. I remember at the time feeling like it was a close call between ‘reject’ and ‘revise (heavily) and resubmit’, though by the time I got around to re-reading it and writing the review, I decided on ‘reject’. Nonetheless, it deserved a thoughtful critique, so I did as much as I could.

While the paper could have benefited from a longer and more detailed review, the paper was just enough of a mess that I was worried that I could inadvertantly spend as much time reviewing it as the authors had spent writing it. I pointed out the two biggest problems with the work and discussed them in some detail, and I briefly mentioned four other fairly significant problems. In addition, I had my own work to get back to.

Writing the review had served as a welcome break from this work, given extra value by the fact that I was finished in short order, and that I felt productive and responsible while doing it. Of course, the satisfaction of finishing the review was itself short lived, as I was immediately back to writing my own manuscripts (which will, I hope, fare better than most).

In the end, I’m glad I agreed to write the review. It’s obvious that peer review is an important part of empirical science. I’ve certainly benefited from receiving good (i.e., useful, not necessarily positive) reviews. The experience of having my first, and thus far, only first author paper published in JASA was very nearly pleasurable; the paper is much better for the comments provided by the reviewers and the action editor, and the whole process was remarkably efficient. (I will refrain, for now, from discussing the experiences that make the efficiency of JASA remarkable.)

My understanding is that serving as a reviewer is not given much official value, though. For example, I don’t think being a reviewer counts for much in a tenure dossier (which, like it or not, is one of the goals of the game I’m in). Maybe I’m wrong; there is plenty I don’t know about how the acquisition of tenure works. Given how important it is (or can be) to communication of scientific work, though, it would make sense to provide strong incentives (or at least a reward) to get people to do it.

Maybe it’s unnecessary to do so, though. I clearly didn’t need too much enticement to review an article this time, and I imagine I won’t next time, either. Perhaps we can get away with relying on people’s sense of professional responsibility.

Some declarative sentences

Some of my .m files (i.e., Matlab code) are available on my models page now. More will be available soon, as will information about any and all available code. Information and code for various other aspects of my research will be made available soon, too. The research page has links to the topic-specific pages, most of which have little content so far.