[ArXiv] The Importance of Being First: Position Dependent Citation Rates on arXiv:astro-ph
The full article is found from [arXiv/astro-ph:0712.1037]. According to J. P. Dietrich, the positional citation effect (PCE) is significant so that preprints appeared at the top of daily lists tend to be more cited than other preprints. Although the study is not statistically rigorous to confirm that up to 6th article on the list is more likely cited, the number are drastically large enough to make people believe the author’s hypothesis.
Personally, this issue has been the biggest obstacle when the weekly [arXiv] is posted. Reading 40-60 abstracts daily always met time constraints. Sometimes, I take a tactic of reading titles and abstracts from the bottom not to skip bottom ones but it’s quite challenging to read abstracts without any break, then I lost my cutoff to continue my backward reading. After many months, actually, I thought about stop reviewing all abstracts from [astro-ph] but get one relevant preprint from the top a few times a month. Now, I felt like someone hit me behind.
As the author suggested, it’ll be better that [astro-ph] is subcategorized; for example, theory, observation, and data analysis; planets, stars, and extended sources (distance based subcategorizing would be needed); multi-wavelengths, gamma-ray, X-ray, UV, optical, IR, and radio; and other classes of subcategories. Like other fields, cross listing will help if the topic cross different subcategories.
I’m not sure how often people feel my [arXiv] is useful. Obviously it’s shorter than astro-ph but the subject is limited to (personally thought) astrostatistics. Giving 5-15 papers may be too many to read. Instead of [arXiv], reviews of short papers or tutorial type papers from statistical journals for astronomers are better suited. Please, send me your feedbacks.
vlk:
Clearly what we need is something like a stock ticker, but with astro-ph titles and abstracts, running in a loop in the background on your desktop or something.
Im afraid I find even your lists are too long, Hyunsook. If I may suggest, it will be more helpful to highlight one or two preprints at a time.
Let me also go on record at this time saying that I loathe and detest citation indices. (Well, not them so much as their use to measure the importance of some work. And I say so even though our wavdetect paper is one of the most cited of all Chandra related papers.) This study just reinforces my view that they are arbitrary, capricious, and unreliable data to measure anything useful, and have large systematic biases that are completely ignored — nay, not even suspected.
12-10-2007, 11:46 amhlee:
I guess it’s time to stop screening notable preprints. Unless [astro-ph] goes under changes, citing/reviewing a few preprint/month will suffer from this selection bias. I’ll come up with something different from the next year starting with unsubscribing to [astro-ph].
12-10-2007, 11:59 pmTomLoredo:
Hyunsook, let me say that I for one greatly appreciate the work you’re doing in distilling the long abstract lists for us. I’ve given up trying to go through the daily arXiv digests myself. Even if you are missing some due to PCE, you are still doing us all a service. (But I would certainly understand if you chose to give it up.)
I think a simple way arXiv could help us deal with this, rather than (or in addition to) categories, is to allow authors to *tag* their entries, and provide tag searchability. A search for terms in abstracts can partly fill this purpose, but not entirely. For example, a couple years ago I was doing some reference counting via ADS, tracking Bayesian work. There was a steep rise in papers using “Bayes”, “Bayesian”, etc., over the past 10 years, but the rate of increase slackened in the last year or two. Part of it turned out to be that in some fields (CMB and grav’l waves) Bayesian methods had become so entrenched that use of them was no longer mentioned in the abstract. Of course, an author could omit such info from tags, but I still think it would be a useful addition to arXiv.
12-11-2007, 7:37 pmhlee:
Tom, I observed important keywords are omitted because the prevalence/popularity of the methodology. Even abstracts do not contain the word Bayesian but the Bayes rule appears in the 2nd section. I recognize that many blogs offer both tag and category options and they are helpful in a difference sense. I agree with you that tagging will be useful as long as there’s a clear instruction of tagging for astronomers when they submit the paper. I saw a distinctive way of keywording in astronomy journals and those keywords could be handy for tagging as well as searching. Millions of tagging phrases make submitting/searching difficult when we have powerful search engines that scan the contents. I hope [astro-ph] adopt a smart way soon not to overwhelm subscribers.
By knowing there’s someone who sees my weekly [ArXiv] list, I should go on. But I cannot help PCE. Thank you for your positive comment.
12-12-2007, 7:51 pmhlee:
Another paper by the same author was up in May 2008. Check [astro-ph:0805.0307] Disentangling Visibility and Self-Promotion Bias in the arXiv:astro-ph Positional Citation Effect
05-05-2008, 9:52 pm