1: % $Id: Astro-Ph-Position.tex,v 1.11 2007/12/04 13:58:44 jdietric Exp $
2: % $Date: 2007/12/04 13:58:44 $
3: % $Author: jdietric $
4: % $Revision: 1.11 $
5:
6: \documentclass{emulateapj}
7:
8: \usepackage{graphicx}
9: \usepackage{natbib}
10: \usepackage{apjfonts}
11: \usepackage{amsmath}
12:
13: \bibliographystyle{apj}
14: \bibpunct{(}{)}{;}{a}{}{,}
15:
16: \begin{document}
17: \title{The Importance of Being First: Position Dependent Citation
18: Rates on arXiv:astro-ph}
19:
20: \author{J.\,P. Dietrich}
21: \affil{ESO, Karl-Schwarzschild-Str. 2, 85748 Garching b. M\"unchen,
22: Germany}
23: \email{jdietric@eso.org}
24:
25:
26: \begin{abstract}We study the dependence of citation counts of e-prints
27: published on the arXiv:astro-ph server on their position in the
28: daily astro-ph listing.
29: %
30: Using the SPIRES literature database we reconstruct the astro-ph
31: listings from July 2002 to December 2005 and determine citation
32: counts for e-prints from their ADS entry. We use Zipf plots to
33: analyze the citation distributions for each astro-ph position.
34: %
35: We find that e-prints appearing at or near the top of the astro-ph
36: mailings receive significantly more citations than those further
37: down the list. This difference is significant at the $7\sigma$ level
38: and on average amounts to two times more citations for papers at the
39: top than those further down the listing. We propose three possible
40: non-exclusive explanations for this positional citation effect and
41: try to test them.
42: %
43: We conclude that self-promotion by authors plays a role in the
44: observed effect but cannot exclude that increased visibility at the
45: top of the daily listings contributes to higher citation counts as
46: well. We can rule out that the positional dependence of citations is
47: caused by the coincidence of the submission deadline with the
48: working hours of a geographically constrained set of intrinsically
49: higher cited authors. We discuss several ways of mitigating the
50: observed effect, including splitting astro-ph into several subject
51: classes, randomizing the order of e-prints, and a novel approach to
52: sorting entries by relevance to individual readers.
53: \end{abstract}
54: \keywords{sociology of astronomy -- astronomical
55: data bases: miscellaneous}
56:
57: \section{Introduction}
58: \label{sec:introduction}
59:
60: A number of studies looking at the influence of e-printing on citation
61: counts across disciplines \citep[e.g.,][]{2001Nature.411.521L} and in
62: Astronomy and Physics in particular
63: \citep[e.g.,][]{2004BAAS...36.1654S,2005BAAS...37..555M,2006JEPub...9....2H}
64: found that papers freely available online, particularly through the
65: arXiv e-print server\footnote{\texttt{http://www.arxiv.org/}}, are
66: cited more often than those not. This difference was studied in more
67: detail by \citet{2005IPM....41.1395K}, who proposed three
68: (non-exclusive) effects potentially responsible for higher citations
69: rates of journal articles also published as astro-ph e-prints. These
70: are defined as \citep[see][]{2005IPM....41.1395K}:
71: \begin{itemize}
72: \item The Open Access (OA) postulate -- Because the access to articles
73: is unrestricted by any payment mechanism authors are able to read
74: them more easily, and thus they cite them more frequently;
75: \item The Early Access (EA) postulate -- Because the article appears
76: sooner it gains both primacy and additional time in press, and is
77: thus cited more;
78: \item The Self-selection Bias (SB) postulate -- Authors preferentially
79: tend to promote (in this case by posting to the internet) the most
80: important, and thus most citable articles.
81: \end{itemize}
82:
83: \citet{2005IPM....41.1395K} found that open access to older issues of
84: astronomical journals through the Astrophysics Data System
85: \citep[ADS,][]{2000A&AS..143...41K} did not lead to increased citation
86: counts, plausibly denying a significant effect of the OA postulate.
87: The EA postulate was supported by a significant increase in citation
88: rate for recent papers. Concerning the SB postulate,
89: \citet{2005IPM....41.1395K} note that significantly fewer papers not
90: posted to astro-ph are among top 200 cited articles in the year
91: 2003 Astrophysical Journal article than expected from the combined
92: OA+EA effect alone. It is thus established that EA and SB play a
93: significant role in gathering citations.
94:
95: We observed that on many days the arXiv astro-ph listing is headed by
96: one or more articles submitted a few seconds after the passing of the
97: deadline for submissions for the next astro-ph mailing. One example --
98: among many others -- is the astro-ph listing of Sep 24, 2007 headed by
99: three articles received within 20\,s after this list was started.
100: Considering that the number of new articles posted to astro-ph on a
101: typical day is about $35$, this temporal clustering close to the
102: submission deadline is conspicuous and suggests a peculiar form of
103: self-promotion. Apparently, a subset of authors expects higher
104: visibility and thus also higher citation rates for articles listed at
105: or close to the top of an astro-ph mailing.
106:
107: In the following we study the dependence of citation counts on
108: articles' positions in the astro-ph listings. In Sect.~\ref{sec:data}
109: we describe how we reconstructed the daily astro-ph mailings for a
110: span of 2.5 years and count citations to e-prints. We analyze the
111: dependence of citations rates on astro-ph listing position in
112: Sect.~\ref{sec:analysis} and discuss three possible explanations for
113: the effect we find in our dataset. We present our conclusions in
114: Sect.~\ref{sec:disc-concl}.
115:
116: \section{Data}
117: \label{sec:data}
118:
119: Using the SPIRES High-Energy Physics Literature
120: Database\footnote{\texttt{http://www.slac.stanford.edu/spires/hep/}}
121: we reconstructed the daily astro-ph mailings in the period from July
122: 2002 to December 2005. The starting date of our analysis is fixed by
123: the date from which on astro-ph is automatically added to SPIRES HEP
124: as soon as it is posted. Restricting SPIRES HEP queries by the date a
125: record was added allows us to request astro-ph listings for single
126: days. Older astro-ph articles were added to SPIRES HEP later in 2002.
127: They were easily recognized by their arXiv identifier and discarded
128: from our dataset. Remaining articles added out of order and occasional
129: small gaps were also found based on their arXiv identifiers. These
130: articles were associated with the correct posting date based on the
131: consecutive arXiv identifier if the association to a date was
132: unambiguous. Otherwise the article and the dates in question were
133: flagged as unreliable and excluded from the analysis. Numerous spot
134: checks confirmed that we were in this way able to reliably reproduce
135: daily astro-ph listings.
136:
137: We chose not to include e-prints appearing on astro-ph after December
138: 2005 so that all articles had at least one year to gather citations
139: before we filled our citation database in December 2006. Our database
140: contains astro-ph listings for 839 days judged as reliable, with a
141: total of 27\,109 articles on these days.
142:
143: We used NASA's Astrophysics Data System (ADS) Bibliographic
144: Services\footnote{\texttt{http://adsabs.harvard.edu/index.html}} to
145: determine if and where an e-print was published in printed form and to
146: determine the number of citations that an article has.
147: \citet{2006JEPub...9....2H} estimate that the concordance achieved by
148: the ADS between astro-ph e-prints and journal articles is 98\%. The
149: remaining incompleteness should not be a significant source of error,
150: especially considering that ADS also lists citations to arXiv
151: e-prints.
152:
153: Astronomy is in the fortunate situation that only a few journals
154: dominate the publication of refereed articles. In agreement with
155: \citet{2005IPM....41.1395K} we consider these 7 core journals to be
156: \emph{The Astrophysical Journal (Letters)} and its \emph{Supplement
157: Series}, \emph{Astronomy \& Astrophysics}, \emph{Monthly Notices of
158: the Royal Astronomical Society}, \emph{The Astronomical Journal},
159: and \emph{Publications of the Astronomical Society of the Pacific}. We
160: set a flag in our database for those e-prints that are also published
161: in one of the core journals.
162:
163: \section{Analysis}
164: \label{sec:analysis}
165:
166: Random and systematical variations cause considerable fluctuations of
167: the length of the astro-ph listings. For our analysis we exclude
168: unusually short listings ($<15$ entries), which typically occur in the
169: period between Christmas and New Year, and extremely long postings
170: ($>70$), of which we have 5 in our database, from our analysis. This
171: selection leaves 816 days to be considered.
172:
173: \begin{figure}
174: \plotone{f1.eps}
175: \caption{Zipf plot for different astro-ph positions. The $x$-axis
176: shows the logarithm of the normalized rank of astro-ph postings
177: after sorting them by citations. The $y$-axis shows the logarithm
178: of the number of citations. The different line colors/styles
179: encode the Zipf law for different astro-ph positions as given in
180: the upper right corner of the figure. The solid black line
181: indicates the slope of the power law. The high citation count of
182: the top-ranked paper in the bin of astro-ph positions 21-25 is due
183: to the WMAP first-year paper by \citet{2003ApJS..148..175S}, which
184: has a submission time 1\,s before the deadline and appeared as
185: 23rd and last paper on Feb.~12, 2003. With 4118 citations reported
186: by ADS at the time of writing, this is the most cited article in
187: all of astronomy.}
188: \label{fig:zipf}
189: \end{figure}
190: \citet{1998EPJB....4..131R} found that the citation distribution is a
191: power law over a large range of citation numbers. The mean and median
192: number of citations that papers of a given sample get are not good
193: estimators as the mean is strongly affected by a few highly cited
194: papers in the tail of the distribution while the median looks only at
195: the large number of very poorly cited papers. A better way to analyze
196: such data is in a Zipf plot. A Zipf plot shows the $r^\mathrm{th}$
197: most cited paper out of an ensemble of size $M$ versus its rank $r$.
198: Figure~\ref{fig:zipf} shows the Zipf plot for some positions in the
199: astro-ph listing. We use the normalized rank $r/M$ instead of the rank
200: $r$ to account for the different lengths of astro-ph mailings. We bin
201: higher astro-ph positions to beat down the noise that comes from the
202: smaller number of articles at higher positions, i.e., from the smaller
203: number of astro-ph mailings that are longer than average.
204:
205: The constant slope of the Zipf plots over a wide range of normalized
206: ranks confirms the power law nature of the citation distribution. One
207: clearly sees that the loci of the curves for positions 1 and 3 are
208: higher than those of e-prints further down the astro-ph listing with
209: an apparent continuous progression down for lower astro-ph positions.
210: The different loci of the curves correspond to the different
211: normalizations of the power law, i.e., the different number of
212: citations that e-prints at different positions in the astro-ph listing
213: get. We call this the \emph{positional citation effect} (PCE) and
214: adopt the following procedure to quantify the magnitude and
215: significance of the PCE.
216:
217: We make a Zipf plot for the ensemble of all astro-ph e-prints and fit
218: a line $f(\ln(r/M)) = \beta \ln(r/M) + a$ to it in the range in which
219: the distribution follows a power law. In agreement with
220: \citet{1998EPJB....4..131R} we find $\beta = -0.48$ with a very small
221: error of $0.02$. Keeping $\beta$ fixed at this value we fit
222: $f(\ln(r/M))$ to the Zipf plots for each (binned) astro-ph position in
223: the range $-4.5 < \ln(r/M) < -2$. The value of $a$ is directly related
224: to the normalization of the power law $N(r/M) \propto
225: (r/M)^{\left(1+\frac{1}{\beta}\right)}$ and determines the loci of the
226: citation distributions in Fig.~\ref{fig:zipf}.
227:
228: \begin{figure}
229: \plotone{f2.eps}
230: \caption{Normalization of the citation distribution power law. Blue
231: boxes give the value of the power law normalization for all
232: articles on astro-ph and its error. The widths of the boxes
233: corresponds to the size of the bins. Red boxes show the same
234: quantity but with the analysis restricted to the 7 core journals
235: in Astronomy. The dotted horizontal lines are the normalization
236: factors of the respective samples of all papers on astro-ph
237: positions between $10$ and $40$.}
238: \label{fig:zipf_coeff}
239: \end{figure}
240:
241: Figure~\ref{fig:zipf_coeff} shows the results of this analysis for two
242: data sets. The blue boxes give the normalization for the ensemble of
243: all astro-ph e-prints, while the red boxes corresponds to those
244: e-prints that also appeared in one the core journals. Comparing the
245: normalization of the bins to that of the ensemble of e-prints on
246: astro-ph positions $10$--$40$, given by the dotted lines in
247: Fig.~\ref{fig:zipf_coeff}, we find that the PCE is significant for the
248: first six astro-ph positions for both data sets. The difference
249: between the normalization at position $1$ and the articles on
250: positions $10$--$40$ is significant at the $7\sigma$ level for all
251: e-prints and at the $4.7\sigma$ level for core journal articles.
252:
253: We have now established that the PCE is present and highly
254: significant. Most scientists will, however, wonder how the different
255: normalizations of the Zipf law translate into citation counts. By
256: restricting an analysis of average citations to the range over which
257: the power law holds, we avoid both the tail of exceptionally highly
258: cited papers and the bulk of mostly ignored publications. We determine
259: average citation counts in the range $-4.5 < \ln(r/M) < -2$ by
260: integrating over the normalized Zipf distribution. We find an average
261: citation count of $95.4\pm11.4$ for core journal articles and of
262: $89.8\pm9.0$ for all e-prints on astro-ph position~$1$. Core journal
263: articles appearing at positions $10$--$40$ are on average cited
264: $54\pm1.6$ times, while the mean citation count for all e-prints at
265: these position is $44.6\pm0.9$. The overall higher number of citations
266: for articles from the core journals is in agreement with the findings
267: of \citet{2004BAAS...36.1654S}.
268:
269: We propose three possible explanations for the observed dependence of
270: citation counts on astro-ph position:
271: \begin{itemize}
272: \item The Visibility Bias (VB) postulate -- Papers appearing at the
273: top of the astro-ph listing are seen by more people and thus cited
274: more often than those further down the list, where the attention of
275: the astro-ph readers might decrease;
276: \item The Self-promotion Bias (SP) postulate -- Authors tend to
277: promote their most important works, and thus most citable articles,
278: by placing them at prominent positions;
279: \item The Geography Bias (GB) postulate -- The submission deadline
280: preferrentially puts those authors at the top of the listing whose
281: working hours coincide with the submission deadline. This group
282: already has higher citation counts for other reasons.
283: \end{itemize}
284: The last postulate merits some further explanation.
285: \citet{2007EurRev..15..3S} noticed that US American authors have a
286: higher fraction of highly cited papers than their European colleagues.
287: The submission deadline for the arXiv e-print server is 16:00~EST/EDT.
288: This is within the normal working time of astronomers in all of the
289: USA, while it is outside working hours for European
290: astronomers.\footnote{In this we of course ignore the few individuals
291: who do the traditional night time work of astronomers.} Assuming
292: for the moment that GB is not caused by VB, i.e., that US Americans do
293: not get cited more because they are preferentially at the top of the
294: astro-ph listing, the PCE could be explained by the dominance of
295: American authors at the relevant moment in time. This explanation of
296: GB ignores astronomical communities outside Europe and the USA, but
297: they are comparably small.
298:
299: To test the GB postulate we analyze the author affiliations of
300: e-prints appearing in core journals. We associate articles to one of
301: the following regions according to the country of the first
302: affiliation of the first author, where the number in parentheses
303: indicates the total number of articles from the respective region in
304: our database: Europe (6843), USA (5659), Asia (1693), North America
305: without USA (605), South America (412), Australia/Oceania (397).
306: Russia (166) and Turkey (32) are counted as Asian countries, Iceland
307: (3) as European. We treat the USA separately from the rest of North
308: America because \citet{2007EurRev..15..3S} made their observation only
309: for authors from the USA.
310:
311: We repeat the analysis of the Zipf plots separately for authors from
312: Europe and from the USA. We find that the global power law indices for
313: both author sets are marginally inconsistent with each other. This
314: could potentially indicate a more fundamental difference between
315: papers written by European and American authors but our data set is
316: too small to draw firm conclusions and a more detailed study is beyond
317: the scope of this work. We settle for using the slopes of the Zipf
318: laws that were found for the respective sample.
319:
320: The results of this Zipf plot analysis are shown in
321: Fig.~\ref{fig:zipf-pos-affiliation} and proof that the PCE is also
322: present for the sample of European authors. This rules out the GB
323: postulate as the single explanation for the observed PCE in the whole
324: sample. Note that the higher normalization of European authors does
325: not mean that these are more cited than their US American colleagues.
326: The difference is simply due to the different power law indices.
327:
328: We emphasize that European authors who appear at the top of an
329: astro-ph listing had -- in all likelihood -- to submit their
330: manuscript well outside their normal working hours. It is thus
331: reasonable to assume that a conscious effort was made to gain this
332: position and to self-promote the work presented in these e-prints.
333: This is further supported by a comparison with the same statistics
334: made for US American authors. A significant increase of citations at
335: the top of the astro-ph listings is also seen for these authors but
336: the effect is not nearly as prominent as for their European
337: colleagues. A fair fraction of American authors will appear at or near
338: the top of the astro-ph listing just by chance, without any attempt at
339: self-promotion. They dilute any SP signal in this sample.
340:
341: So far we could rule out GB as a significant contribution to the PCE
342: and found evidence for SP as a source of the PCE. Analyzing the
343: contribution of VB is much more difficult. In principle this would be
344: possible by examining the submission times of e-prints and grouping
345: them into two samples; one that is submitted so shortly after the
346: deadline that it is statistically expected to be self-promoted, and a
347: second one that is submitted long enough after the deadline to exclude
348: self-promotion. Unfortunately, the arXiv stores only the submission
349: time of the last replacement that was made without generating a new
350: version number. It does not keep the initial submission time of an
351: e-print. Without the latter it is not possible to disentangle VB and
352: SP.
353:
354: Figure~\ref{fig:zipf_coeff} seems to show a marginally significant
355: drop of the power law normalization for articles at the bottom of very
356: long astro-ph mailings. This is most likely an artefact caused by the
357: growth of astro-ph over the period under investigation here. Far fewer
358: listings had more than 40 entries in 2002 and 2003 than in 2004 and
359: 2005. These articles had less time to garner citations, an effect
360: which we, as mentioned previously, do not correct for. We find that
361: the decline in citation rates at the bottom vanishes if we restrict
362: the citation analysis to e-prints that appeared in 2004 and 2005.
363:
364: \begin{figure}
365: \plotone{f3.eps}
366: \caption{Normalization of the citation distribution power-law for
367: European (red) and American (blue) authors of core journal papers.
368: Note that the binning is coarser than in Fig.~\ref{fig:zipf_coeff}
369: to account for the smaller number of articles in each sample.}
370: \label{fig:zipf-pos-affiliation}
371: \end{figure}
372:
373:
374: \section{Summary and conclusions}
375: \label{sec:disc-concl}
376: %
377: We investigated the positional dependence of citation counts of
378: e-prints on their position in daily arXiv:astro-ph mailings. We found
379: that articles at or near the top of these listings receive
380: significantly more citations than articles at positions $10$ and
381: higher. This positional citation effect is present at the $7\sigma$
382: level for all e-prints and for the set of those e-prints also
383: appearing as articles in one of the core journals in Astronomy and
384: Astrophysics with a significance of $4.7\sigma$. Restricting the
385: analysis to the range over which the citation distribution follows a
386: power law, the difference in normalization factors translates to a
387: factor 2 difference in average citation counts. E-prints at the top of
388: the astro-ph listing are on average cited $89.8\pm9.0$ times while
389: articles between positions $10$ and $40$ receive only $44.6\pm0.9$
390: citations. The PCE is significant for the first six positions in the
391: astro-ph mailings.
392:
393: We proposed three possible explanations for the observed PCE. By
394: analyzing the affiliations of authors we could exclude the hypothesis
395: that the PCE is caused by a geographical bias, namely by
396: preferentially putting US American authors at the top, who already
397: have higher citation rates than their European colleagues. Confirming
398: that the PCE is present and more pronounced for European authors, who
399: usually can gain the top position on astro-ph only by submitting
400: outside their normal working hours, we could conclude that
401: self-promotion plays a role in creating the PCE.
402:
403: We cannot present firm statistical evidence that the PCE at the top is
404: due to higher visibility. However, the PCE is present down to astro-ph
405: position 6, where the submission times of e-prints suggest that only a
406: very small fraction of manuscripts at this position was submitted with
407: the intention of self-promotion. Of course, this statement has to be
408: made with some care since the submission times reported by arXiv are
409: often not the original submission times as explained above. We,
410: however, observe that less than $5\%$ of e-prints at position 6 have
411: reported submission times within half an hour after the deadline,
412: while this fraction for e-prints at the top position is $62\%$,
413: suggesting that VB plays some role.
414:
415: A situation in which the number of citations a publication gets does
416: not solely depend on its merits is unsatisfactory. One may be tempted
417: to suggest to randomize the order of e-prints in an astro-ph mailing,
418: instead of listing them sorted by submission time. This proposal has
419: severe weaknesses. On the one hand, the PCE in a randomized listing
420: would only vanish if the observed effect is completely due to SP bias,
421: since any visibility effects would immediately recur in a randomized
422: listing. On the other hand, if in fact SP bias is the sole cause of
423: the PCE there is no reason to change anything. It is important to
424: understand that SP means that papers at the top are on average
425: intrinsically more citable -- or better by some metric -- than papers
426: further down the list. A PCE caused only by SP bias is not in
427: contradiction to merit based citation counts. Hence, randomizing the
428: astro-ph listings would not solve the visibility problem and only
429: remedy a non-existing problem.
430:
431: Any visibility problem is connected to the length of astro-ph
432: listings. Shorter listing would naturally give the same or almost the
433: same visibility to all articles, irrespective of their position.
434: Looking at the arXiv submission history, it is obvious that the number
435: of astro-ph e-prints will not decrease but only increase with time. A
436: possible method leading to shorter listings would be to split astro-ph
437: into several subject-classes, like the other three big archives
438: ``hep'' (High-Energy Physics), ``cond-mat'' (Condensed Matter), and
439: ``math'' (Mathematics) are subdivided into smaller categories.
440: Ideally, most researchers would have to look at only one or two more
441: specific and hence much shorter listings per day. For example, a
442: cosmologist could avoid seeing e-prints on Solar System objects if
443: he/she chooses not to read a possible Solar System subject-class.
444:
445: A recent approach to tackle the increasing volume of astro-ph is the
446: arxivsorter\footnote{\texttt{http://arxivsorter.org/}}
447: \citep{2007Arxivsorter..M}. Arxivsorter aims to sort daily, recent, or
448: monthly astro-ph listings by relevance to an individual reader. The
449: underlying idea is that scientists through co-authorship form an
450: interconnected network of authors. By specifying a few authors
451: relevant to a reader's fields of interest, the ``proximity'' of a new
452: e-print in the author network can be calculated. This proximity seems
453: to be a good proxy for relevance to a reader's interests. Arxivsorter
454: has some problems with non-unique -- mostly Chinese -- names, and a
455: small percentage of authors not connected to the global cluster of
456: authors. It is worth pointing out that arxivsorter just re-orders the
457: papers without any loss of information, a clear advantage over a
458: possible split into subject-classes.
459:
460:
461: \acknowledgements This research has made use of NASA's Astrophysics
462: Data System Bibliographic Services. I am very grateful to Maryam
463: Modjaz, who identified a crucial error in a previous version of the
464: analysis presented here. Brice M\'enard and the referee Michael Kurtz
465: gave a number of excellent suggestions, which improved this paper. I
466: thank Uta Grothkopf for a careful reading of the manuscript.
467:
468:
469: \begin{thebibliography}{10}
470: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
471:
472: \bibitem[{{Habing}(2007)}]{2007EurRev..15..3S}
473: {Habing}, H. 2007, {European Review}, 15, 3
474:
475: \bibitem[{{Henneken} {et~al.}(2006){Henneken}, {Kurtz}, {Eichhorn},
476: {Accomazzi}, {Grant}, {Thompson}, \& {Murray}}]{2006JEPub...9....2H}
477: {Henneken}, E.~A., {Kurtz}, M.~J., {Eichhorn}, G., {et~al.} 2006, Journal of
478: Electronic Publishing, 9, 2
479:
480: \bibitem[{{Kurtz} {et~al.}(2005){Kurtz}, {Eichhorn}, {Accomazzi}, {Grant},
481: {Demleitner}, {Henneken}, \& {Murray}}]{2005IPM....41.1395K}
482: {Kurtz}, M.~J., {Eichhorn}, G., {Accomazzi}, A., {et~al.} 2005, Information
483: Processing and Management, 41, 1395
484:
485: \bibitem[{{Kurtz} {et~al.}(2000){Kurtz}, {Eichhorn}, {Accomazzi}, {Grant},
486: {Murray}, \& {Watson}}]{2000A&AS..143...41K}
487: {Kurtz}, M.~J., {Eichhorn}, G., {Accomazzi}, A., {et~al.} 2000, \aaps, 143, 41
488:
489: \bibitem[{Lawrence(2001)}]{2001Nature.411.521L}
490: Lawrence, S. 2001, Nature, 411, 521
491:
492: \bibitem[{{Magu\'e} \& {M\'enard}(2007)}]{2007Arxivsorter..M}
493: {Magu\'e}, J.-P. \& {M\'enard}, B. 2007, {Arxivsorter documentation, {\tt
494: http://arxivsorter.org/doc}}
495:
496: \bibitem[{{Metcalfe}(2005)}]{2005BAAS...37..555M}
497: {Metcalfe}, T.~S. 2005, in Bulletin of the American Astronomical Society, 555
498:
499: \bibitem[{{Redner}(1998)}]{1998EPJB....4..131R}
500: {Redner}, S. 1998, European Physical Journal B, 4, 131
501:
502: \bibitem[{{Schwarz} \& {Kennicutt}(2004)}]{2004BAAS...36.1654S}
503: {Schwarz}, G.~J. \& {Kennicutt}, Jr., R.~C. 2004, in Bulletin of the American
504: Astronomical Society, 1654
505:
506: \bibitem[{{Spergel} {et~al.}(2003){Spergel}, {Verde}, {Peiris}, {Komatsu},
507: {Nolta}, {Bennett}, {Halpern}, {Hinshaw}, {Jarosik}, {Kogut}, {Limon},
508: {Meyer}, {Page}, {Tucker}, {Weiland}, {Wollack}, \&
509: {Wright}}]{2003ApJS..148..175S}
510: {Spergel}, D.~N., {Verde}, L., {Peiris}, H.~V., {et~al.} 2003, \apjs, 148, 175
511:
512: \end{thebibliography}
513:
514:
515: \end{document}
516: