Downloaded by guest on September 25, 2021 nipratfcto h etrhppoes xeinewith experience process: journal. specific mentorship a the within publishing of on but facet directly, mentorship important on pres- an not in focus, publishing we on Here rests venues. success tigious career answers, select of meaningful to aspect learning providing important an of and skill questions core scientific by the trained relevant Beyond example, were 6). For laureates (1, careers. Nobel laureates of other their fraction in large later improbably prob- an success higher a achieving have of mentors ability successful with interact who entists chaperone strong a scien- by by characterized played venues and effect. in role expertise, publish experience, the necessary to on the skills acquire light same to the training shed within senior tific published findings as previously multi- not appear Our high-impact has to journal. she For unlikely or he is sciences. if scientist author natural a the journals, effect sciences disciplinary for mag- biological the and science, different weaker medical of a and within branches pronounced has different more effect in being journal. chaperone journals scientific for the nitude same that the illustrate status consid- within senior by We publications into effect” Here multiple transition “chaperone 5). scientists this given (4, inexperienced of aspect how generations key ering scientific between a how along quantify of role we passed and the is publishing of understanding scientific excellence mathematical quantitative in less apprenticeship the is of scien- sci- like there renowned prominent studies between (4), bonds entists of single-topic such document scientific lineages While project that genealogy through 3). notion along (2, the tists passed supports is mentors status accomplished of S science of science prestigious in publish to required skills venues. the experience on necessary and scientific the expertise, acquiring specific and toward the within paths by on publishing the authored on in light journals, papers experience shed by to findings played Our relative the role (PIs). impact venues, investigators average high-impact principal higher result- new of a specifically case in implications, the significant ing has in effect that sciences. chaperone natural show for in we weaker magnitude and Finally, sciences pronounced different biological more and being a medical science, in has of branches effect different in illus- chaperone We journals the venue. publication that particular a trate transition within scientists status how senior capturing into quanti- effect,” we by “chaperone order Here, this author journal. fying of same understanding the she quantitative or a within he develop if published author previously senior as not appear has to scientific unlikely journals, is multidisciplinary high-impact scientist top a in crafting where evident in particularly for is role (received This 2018 critical 6, work. September a Debenedetti G. plays Pablo Member Experience Board Editorial by accepted and 2018) MA, 15, Boston, February University, review Boston Trunfio, Paul by Edited and 02115; Italy; MA Torino, Boston, 10126 School, Medical Harvard Hospital, Women’s Hungary; www.pnas.org/cgi/doi/10.1073/pnas.1800471115 Hungary; Budapest, 1051 University, European Central Science, Network 02115; MA Boston, Universit Mathematics, Applied a Lehmann Sune and publishing Sekara scientific Vedran in effect chaperone The eateto ple ahmtc n optrSine ehia nvriyo emr,D-80KnesLnb,Denmark; Lyngby, Kongens DK-2800 Denmark, of University Technical Science, Computer and Mathematics Applied of Department ngnrl hr r elho niain htyugsci- young that indications of wealth a are there general, In a vdneta cetfi ihaheesaeotnprot anecdo- often are and achievers (1), high stratified scientific highly that is evidence tal institution an as cience g etrfrCne ytm ilg,Dn-abrCne nttt,Bso,M 02115; MA Boston, Institute, Cancer Dana-Farber Biology, Systems Cancer for Center | a d cetfi careers scientific ireDeville Pierre , hoyo odne atr aeds aoaoy nvriyo abig,CmrdeC30E ntdKingdom; United 0HE, CB3 Cambridge Cambridge, of University Laboratory, Cavendish Matter, Condensed of Theory k a,k,1,2 h il orIsiue nvriyo oehgn K20 oehgn Denmark Copenhagen, DK-2100 Copenhagen, of University Institute, Bohr Niels The ahlqed ovi,14 ovi-aNue Belgium; Louvain-la-Neuve, 1348 Louvain, de Catholique e ´ | mentorship b,c eata .Ahnert E. Sebastian , i opeiySineHb 00Ven,Austria; Vienna, 1080 Hub, Science Complexity d Albert-L , ´ eg f ´ es eateto ahmtc,CnrlErpa nvriy 01Budapest, 1051 University, European Central Mathematics, of Department uhr hs eadesaqeto hc sotnakdb sci- by asked often Nature junior in is publish a which you as question “Can journal a entists: same address of the we (PI) in Thus, investigator published sciences, author. principal previously the the has unravel across which paper To to vary a extent sequences authors. the these some explore of we for dynamics acceptance the of how rate 10). that patterns (9, the the analyz- publications unveiling increase by dish,” multiauthor “petri (5) a scientific provide work sequences study of Such scientific to dynamics crafting order the in author ing experience use of we role work, the of this terms In in (13–16). heterogeneous experience usually is multiple composition by researchers performed whose often authors (11, junior are mentorship works need that high-impact but Further, topics shown 12). innovative often more been on is work recently to labor as tend has of such division it roles, This symbiotic; specialized analyses. the more statistical establishing play contributing research, fig- generally 10). the mentor (9, authors editors shaping a journal Middle in with is corresponding role author and structure, a last paper’s who plays the scientist while who early-career research, ure an the is out author sci- physical carries first in the increasingly and typically, project biological ences a in in example, scientist For each 8). of (7, role the regarding signals important 02i 8 cetfi oras oeigtefilso mathe- of fields (see the medicine covering and biology, journals, chemistry, scientific physics, matics, 386 in 2012 spans analysis in described our as However, journals, factor. multiple high-impact a with journal a ulse nieDcme 0 2018. 10, December online Published 1073/pnas.1800471115/-/DCSupplemental. y at online information supporting contains article This 2 PIs chaperoned 1 and established, new, of ( values of the proportion and the time over about Data deposition: Data the under Published V.S., and findings; the discussed paper. the and the wrote explain interpreted S.L. to the and authors calculations R.S., curated all A.-L.B., analytical and the model; collected made R.S., R.S. null V.S., S.E.A. and alphabetical analysis; research; P.D., the the V.S., conducted led analyses; V.S. S.L. and data; and measures R.S. designed study; S.L. the and conceived S.L. contributions: Author hsatcei NSDrc umsin ..i us dtrivtdb h Editorial the by invited editor guest a is P.T. Submission. Direct PNAS Board. a is article This interest.y of conflict no declare authors The at of website recordings NAS of video the Academies and on National program and available the complete are of The Center presentations CA. Beckman Irvine, most Mabel in and Engineering Arnold and the Sciences Decem- held at Developments,” 2017, Sci- Technology 4–5, of and Academy ber Science National the Visualizing of and Colloquium “Modeling Sackler M. ences, Arthur the from results paper This https://github.com/SocialComplexityLab/chaperone-open). ..adRS otiue qal oti work.y this to equally robertasinatra@ contributed R.S. [email protected], and S.L. Email: addressed. be [email protected] or may gmail.com, correspondence whom To aszl ´ h re fatosi utato cetfi rilsprovides articles scientific multiauthor in authors of order The ecnie . ilo aespbihdbten16 and 1960 between published papers million 6.1 consider We visualizing y Barab o ´ eoe” oeta eew take we here that Note before?”. c PNAS etrfrCmlxNtokRsac,Nrhatr University, Northeastern Research, Network Complex for Center .y h asi hnigDvso fNtokMdcn,Bihmand Brigham Medicine, Network of Division Channing ´ | NSlicense.y PNAS eebr1,2018 11, December c,e,f,g,h,1 c , C and , oet Sinatra Roberta , j aaSineLbrtr,IIFoundation, ISI Laboratory, Science Data y C Nature alphabet | aeil n Methods. and Materials r rvddfrec ora nGitHub on journal each for provided are fyuhv ee ulse in published never have you if o.115 vol. www.pnas.org/lookup/suppl/doi:10. Nature www.nasonline.org/modeling y c,e,f,i,j,1,2 | b o 50 no. eatetof Department sa xml of example an as e etrfor Center , | 12603–12607 Materials
SOCIAL SCIENCES APPLIED PHYSICAL COLLOQUIUM SCIENCES PAPER and Methods for data processing and name disambiguation). large fraction of repeated names in the PI spot. In Fig. 1B, Bot- Included are the top 3 multidisciplinary journals: Nature, Science, tom we show Nature, a journal with an interdisciplinary audience, and PNAS. In our analysis, we assume that the PI is listed last in which has undergone a strong change over the past 10 y, with the a paper’s author list, a common practice in many scientific fields fraction of new authors dropping significantly. This indicates that (9, 10, 17). Note, however, that our analysis is not affected if the it is becoming increasingly rare to publish as the senior author author list of some papers does not mirror seniority roles (Mate- in Nature without previous publishing experience in the jour- rials and Methods). For all papers in each journal, we divide PIs nal. A possible explanation for this development is an increasing into three categories: New PIs are those who have not published number of authors specializing in writing papers for high-impact previously in that specific journal, chaperoned PIs are those who general audience journals, eschewing the more traditional pat- have appeared before only as junior (nonlast) authors, and estab- tern of publishing primarily in specialized journals and sending lished PIs are those already previously listed as a last author in only selected results to high-impact multidisciplinary journals. the journal (Fig. 1A). By definition, the last author in any given To understand the role of journal-specific experience, we publication can be classified only in one of these categories. For investigate the chaperoned authors more closely. Chaperoned example, F. J. Weissing’s first paper in Nature was as last author, authors are senior authors who have published in the journal so he is labeled as a new PI in Nature for that year (1999). In 2007, previously as nonlast authors (Fig. 1A). Due to prior experience Weissing published in Nature as last author again, but because with the process, chaperoned PIs already have gone through the of the previous publication, we categorize him as an established intensive process of preparing a manuscript for a high-impact PI in Nature in 2007. This 2007 Nature paper was coauthored journal and absorbed tacit knowledge on how to frame the mes- by three other scientists, one of them being O. Leimar. A year sage appropriately for the journal audience, how to strike the later, Leimar published a paper as last author in Nature and is right tone in the cover letter, how to structure the supporting therefore marked as a chaperoned PI in Nature for that year. In information, and the subtleties of how to constructively interact Fig. 1B we show the fraction of new, established, and chaperoned with editors, mastering layers of information that is usually invis- authors over time for three scientific journals. ible to those reading a paper. Hence, the senior author acts as a The proportion of these three kinds of author is substan- chaperone simply through guiding the submission process. Hav- tially different depending on the journal. New England Journal ing experienced the entire publication process once increases the of Medicine (NEJM) is an example of a journal where the high- chances of publishing in similar journals again, since the author is est fraction of senior authors is new (Fig. 1B, red line), signaling familiar with their particular idiosyncrasies. In Fig. 1B, the chap- that repeat authorship is less common; i.e., the medical commu- eroned fraction hovers at around 0.1–0.2 for all three journals, nity tends to submit only the most groundbreaking work to this but shows an increasing trend over time for NEJM and Nature. high-impact general interest journal. As a point of contrast, we Thus in both NEJM and Nature, high-impact journals with a wide show Physical Review D as an example of a journal where estab- audience, the fraction of new authors decreases over time, while lished PIs are predominant, a tendency which increases over time the fraction of chaperoned authors slightly increases. In other (Fig. 1B, blue line). This picture arises when some authors spe- words, it is becoming harder to publish in Nature without having cialize in writing for a particular disciplinary journal, leading to a published in Nature before.
AB
Fig. 1. Probability of being listed as PI in Nature given previous publication history. (A) Terminology of authors. The last authors of all papers published each year in Nature are divided into three categories: new authors that have never published in Nature before, chaperoned authors that have published in Nature before only at junior level, and established authors that have already previously published as last authors. (B) Change in author fractions over time for three journals, displaying different trends over time. While in the New England Journal of Medicine (NEJM) the proportion of different PIs tends to be equally balanced over time, in Physical Review D this proportion tends to become more unbalanced, with the fraction of established PIs increasing. For author fractions in PNAS see SI Appendix, Fig. S2.
12604 | www.pnas.org/cgi/doi/10.1073/pnas.1800471115 Sekara et al. Downloaded by guest on September 25, 2021 Downloaded by guest on September 25, 2021 ueo h hprn fet oeta h magnitude the that Note effect. chaperone the of tude h admodrn.I eea,tecaeoephenomenon chaperone to the respect with general, authors In of ordering. order random the the in as changes productivity, significant However, volume randomization. ture the individual in and preserved are size these team by of affected order be call the We permuted paper. randomly each have in we Therefore, names where author (21). model where relevant null system not a a is consider names compare we author we First, of 20). ordering (19, the models occurring those null with compare two chaperoned we and in new sciences, of values the observed across the apprenticeship of importance through phenomenon erone an ran- the position compared. last fact reasons, to these the nonlast For account from domly. into transition the take make not can does author also pro- It individual (18). and size ductivity team typical like conventions, publishing of eaae al. et Sekara (see model null the of properties analytical the of because while 0.5 distributions around peak all tions one chaperone. than a smaller as effect chaperone act a author (c has senior ques- journal a in a of have journal if fraction to the Conversely, the important in is than publish it smaller to tion, is (c that, authors and one chap- authors new than the chaperoned of greater fraction If is the journal authors. that a chaperoned for and effect new erone between balance a IAppendix hc (SI fields scientific compare to us enable 2. Fig. with of experience effect previous chaperone without A journal journal. chaper- a within that and in associated that difficulty new publishing the captures with authors between effect chaperone proportion last The the PIs. the other oned In in of compare time. over position we transition number a the last words, such made made the never the have have journal to with the time position order over nonlast that author a authors from of transition number the paring hc /c /c ocretfrteecvasi h unicto ftechap- the of quantification the in caveats these for correct To eqatf h hprn effect chaperone the quantify We P(C) P(C) < c random random 0.0 0.1 0.2 0.0 0.1 0.2 oee,i fetdb edseiccaatrsisand characteristics field-specific by affected is however, , ,pbiaini airfrnwatos h pcfi value specific The authors. new for easier is publication 1), oprsno hprn fetbtensinicfils erydsrbtosfrteps 2yaeclasdit igedsrbtosand distributions single into collapsed are y 12 past the for distributions Yearly fields. scientific between effect chaperone of Comparison i i interdisciplinary physics
0 Alphabet c ' to 0.91, C alphabet c random ' Mathematics hc 8 icxnrn u et oevr lutae httedsrbtosaedistinguishable are distributions the that illustrates moreover, test, sum rank Wilcoxon A 1.68. 1 /c itiuin r niae ngray. in indicated are distributions random where , c C fdfeetfilscno edirectly be cannot fields different of i c chemistry n ob bet opr the compare to able be to and c 2 random Mathematics ' c Medicine 1.01, C c 1 = stertootie in obtained ratio the is = o ora ycom- by journal a for hc mle htteeis there that implies c .Frtedfeetdsilnsw n naeaethat average on find we disciplines different the For S1). Fig. and 2 section , /c /c 3 random random > 0 i
Physics means it 1), C medicine h magni- the C oscap- does cannot ' 1 and 1.21, 1 Purity offield rniinn ols uhrcag safnto fnme of 3A number Fig. of In function author? of a a nonlast as probability a answer change as the however, author occurrences does last can, How to unable We namely, transitioning are exactly. question, do we related we question papers, Since closely rejected that impact author? for author answer last statistics nonlast as to to a publishing access day as have question: one not journal important for a an odds in one’s prompts publishing authors does last How new and eroned within authorship senior and them important journals. is junior as in high-impact publication between publishing them with transitioning of experience in for probability that published illustrating the PI, having and as between researcher junior relationship a clear there fields, growing a these effect For is journals. chaperone general-topic effect and strongest the the biology with of within publications medicine, and magnitude all chemistry, physics, the for across conventions see order authorship We that alphabetical fact (22). dictate the to mathematics due expe- part purity in that levels. in “the senior likely signs and about is junior few This intuition between very transition collective the show influences the Mathematics rience with (23). line sciences” in of is journals. 2 interdisciplinary for Fig. and above mentioned fields five the and (Materials effects of confounding deviation any Methods). the of stripped nutshell, effect, a erone In sorted ). 1 are section from construct lists Appendix, we author (SI model, one all second where this system on C compare Based a (22). on fields impor- also alphabetically based some an we is is in Therefore, which sorting Second, convention. posi- author way. last tant alphabetical significant the mathematics), in statistically authors (e.g., a new in of appearance tion the than frequent when occurs CC alphabet hc sesn h xsec fa naacdpooto fchap- of proportion unbalanced an of existence the Assessing of distribution the show we 2, Fig. In
/c Chemistry random c IAppendix SI 2 random = c i biology /c Physics C Biology and alphabet PNAS > ' eto 1 section , ati more is nonlast→last transition the when i.e. 1, c 1 hl h fetfritricpiayjunl is journals interdisciplinary for effect the while 1.41, 3 alphabet ausof Values . |
eebr1,2018 11, December Medicine 0 o proof). a for rvdstemgiueo h chap- the of magnitude the provides C alphabet 1 P C srpeetdb h colored the by represented is | Biology than smaller typically are C 5 The 0.05. o.115 vol. swl as well as hc eseta,i the in that, see we /c Interdisciplinary 2 random | C c alphabet o 50 no. Chemistry Interdisciplinary C to i math alphabet c distribu- | alphabet ' 12605 3 0.73, for c ,
SOCIAL SCIENCES APPLIED PHYSICAL COLLOQUIUM SCIENCES PAPER A our findings suggest that experience of publishing within spe- cific journals can play an important role in acquiring long-term 0.25 scientific impact. Taken together, our results add a piece to the puzzle of how mentor–prot´eg´e relations function more generally (4, 5, 25–28), 0.20 where a full picture of the relation will also draw on understand- ing how teams are assembled and produce knowledge (13, 29). 0.15 In this sense, additional research is needed to understand the complex processes that drive the differences between new and Nature 0.10 chaperoned authors. By focusing on the role of experience within
Probability of transitioning to last Physical Review D journals and fields, we deliberately average over authors with
1 2 3 4 very different levels of success and do not account for the fact Number of publications as non-last that good prot´eg´es tend to find good mentors; nor do we include B the fact many young authors leave science altogether. There- 160 Established fore, it is important to stress that our results are not designed 140 Chaperoned to answer the deeper questions about mentor–prot´eg´e rules, but New 5 point to the general structures in how knowledge needed to write c 120 for certain journals is different across the sciences, with high- act, act,
p 100 impact, interdisciplinary journals showing a particularly strong
e im 80 effect. Thus, while the available data here do not allow us to strictly 60 Averag pinpoint which facet of experience is most important to suc- 40 ceed in science or which share of a senior author’s apprentices are successfully chaperoned (30), we have demonstrated that 1990 1995 2000 2005 2010 the chaperone effect does indeed exist, showing that the ability Year to publish in certain venues is something that junior scientists Fig. 3. The advantages of chaperoned and established PIs. (A) The probabil- learn from senior colleagues. Further, we have demonstrated ity of transitioning to last author as a function of number of occurrences as that apprenticeship is not just about membership in the exclusive nonlast author for a specialized journal, Physical Review D, and an interdisci- club of having published in Nature or another prestigious jour- plinary journal, Nature. (B) Average impact of papers in Nature, quantified nal, but that papers by chaperoned authors have greater scientific with citations after 5 years from publication (c5), for papers authored by impact than papers by new PIs. In addition, we show that the new, chaperoned, and established last authors. magnitude of the chaperone effect varies across scientific fields (23). The chaperone effect is most strongly expressed in pres- tigious multidisciplinary journals, demonstrating that the highly case of Nature, this probability grows significantly from 10% after specialized skill set required to publish in these venues is passed one publication to nearly 20% after four publications as nonlast along more strongly than any field-specific expertise. author (for the study of the chaperone effect in PNAS, see SI Appendix, Figs. S3 and S4). In contrast, the same transition prob- Materials and Methods ability is 25% in the case of the highly disciplinary Physical Review Data. We use publication data provided by the Web of Science database D and does not change with additional publications as nonlast (www.webofknowledge.com), purchased for research purposes by some of author. the authors of this publication in 2013. The database includes several types There is a critical aspect of the chaperone effect that we have of scientific outputs such as articles, letters, reviews, editorials, and abstracts not yet explored: Does experience with publication within a cer- from 1898 to 2012 across more than 22,000 scientific journals from broad tain journal play a role in the scientific impact of subsequent domains, resulting in a set of more than 50 million papers. For each paper, papers that the PI published in the same venue? Could it be that the dataset includes more information on the date of publication (month, day, year), the journal name and journal issue, author names with the order new, chaperoned, and established last-author papers receive dif- they appear in the article, their affiliations, and the references to past arti- ferent levels of recognition from the scientific community? Our cles indexed in the database. For Nature we downloaded the full publication initial hypothesis was that papers authored by new PIs might history using the Nature opensearch Application Programming Interface. have higher impact, since their lower odds of being published For our analysis, we focused on publications from 1960 to 2012 pub- might signal a higher significance of the reported discoveries for lished in interdisciplinary journals (Nature, Science, and PNAS), as well as the scientific community. To test this hypothesis, we quantified in journals associated to five distinct scientific fields: medicine, biology, the impact of each paper by measuring c5, its citations after 5 y mathematics, chemistry, and physics. To identify the journals belonging to from publication, a measure that is not affected by the specific- each category, we first parsed dedicated Wikipedia pages containing lists field citation dynamics (24). This allowed us to directly calculate of journal names associated to specific scientific fields and then matched these with the journals in the database (31). In total we identified 97 the average impact over time for three categories of papers: biology, 337 medicine, 243 physics, 248 mathematics, 138 chemistry, and 3 those with chaperoned, established, and new PIs. In the case of interdisciplinary journals. Nature the result is striking (Fig. 3B). We find that papers with Next, we extracted the publications associated to each of these catego- established and chaperoned PIs have indistinguishable impact. rized journals. To ensure dealing with original research, we collected only Contrary to what we expected, however, papers authored by new publications labeled as articles, letters, and reviews and that did not have a PIs in Nature receive on average only half the citations of papers title containing the terms comment, reply, errata, or retracted article. More- authored by chaperoned and established PIs, indicating a system- over, to have enough statistics, only the categorized journals fulfilling the atically lower scientific impact. The same pattern is observed in following criteria were taken into account for our analysis: The collected the entire group of interdisciplinary journals, suggesting this pat- publications associated to the journal span a period of at least 10 y, at least 1,000 collected publications were published in the journal overall, and at tern is consistent in these venues with high selection pressure least 100 collected publications were published each year in the journal. and only a small fraction of all scientists manage to publish as After this preprocessing, our data amount to (i) 795,558 publications PI. In more specialized field-specific journals, a difference can from 40 journals in biology, (ii) 1,350,936 publications from 128 jour- be also present, but the differences between the three categories nals in medicine, (iii) 1,753,641 publications from 117 journals in physics, of authors tend to be smaller (SI Appendix, section 3). Thus, (iv) 208,223 publications from 26 journals in mathematics, (v) 1,341,150
12606 | www.pnas.org/cgi/doi/10.1073/pnas.1800471115 Sekara et al. Downloaded by guest on September 25, 2021 Downloaded by guest on September 25, 2021 4 oe F uhyS ziB(08 ut-nvriyrsac em:Sitn impact, Shifting teams: research Multi-university (2008) B Uzzi production in S, teams Wuchty of BF, dominance Jones increasing The 14. (2007) B Uzzi BF, ideas. Jones fresh S, for Wuchty go scientists of Young 13. Bureau (2015) (National E ideas Callaway new out 12. trying and Age (2015) J Bhattacharya M, takes. Packalen it What 11. (2014) J Austin 10. 5 Milojevi 15. 7 ata 21)A miia nlsso h s fapaeia uhrhpin authorship alphabetical of use the of analysis empirical An teams. (2012) of success L and Waltman dynamics group 17. the Understanding (2016) JP Bagrow M, Klug 16. h aefratdnm n vlei h aesinicfil,ie,the i.e., field, (32). scientific feature same disambiguating accurate the an in already is evolve which journal, and same share name individuals distinct formatted two if same only within occur the names thus only can compare error we An as journal. low same the be that to individuals name, distinct formatted i.e., same the homonyms, share by induced indi- errors same expect the journal to We correspond same vidual. to name the formatted respec- within identical an “johnson,ms,” Johnson” publications by authored or of Suzy and “smith,j” sequence “Mary the format or considered the We Smith” to tively. “John converted be named thus author would An their into only. names letter their converted first and lowercase to publications collected the Disambiguation. Name Author (https://www.nature.com/opensearch/). accessible. opensearch journal are the data the about where Northeast- University Data to European visit Central reproduce research or to a University making possibility by ern the records offer raw we from but starting web, results our the on publicly shared be not from data Raw SocialComplexityLab/chaperone-open). salse,adcaeoe I vrtm n h ausof values the and time over PIs C chaperoned (vi and and established, chemistry, in journals from 72 from publications eaae al. et Sekara .(08 cdmcgnaoyo hoeia hscss vial ten.wikipedia.org/ at Available physicists. theoretical of genealogy genealogy.math.ndsu.nodak. Academic at (2018) Available 3. project. genealogy Mathematics (2018) (1973) S 2. Cole JR, Cole 1. .vnDj ,MnrO ae B(04 ulcto erc n ucs nteacademic the on success and metrics Publication (2014) LB Carey O, Manor D, Dijk van 9. Barab HW, Shen measures. for 8. Measures (2006) BE Lautrup collabora- AD, productivity, Jackson of S, Lehmann Patterns science: 7. in laureates Nobel (1967) H Zuckerman 6. comparison A prot mentorships: in informal and mentorship Formal (1992) of PD Gardner role P, The Walz GT, (2010) Chao LAN 5. Amaral JM, Ottino RD, Malmgren 4. alphabet wiki/Academic 2018. 14, February Accessed edu/. egah,adsrtfiaini science. in stratification and geography, knowledge. of 20920. Paper Working MA), Cambridge, Research, Economic market. job USA 1004. authorship. and tion, counterparts. nonmentored with contrast 45:619–636. and functions mentoring on performance. alAa c USA Sci Acad Natl cetfi publishing. scientific Sci Open Soc R Nature, 111:12325–12330. r rvddfrec ora nGtu ( GitHub on journal each for provided are 21)Picpe fsinicrsac emfrainadevolution. and formation team research scientific of Principles (2014) S c ´ urBiol Curr Nature and Science, Science 3:160007. genealogy s L(04 olciecei loaini science. in allocation credit Collective (2014) AL asi ´ 111:3984–3989. Informetrics J 24:R516–R517. 465:622–626. mSco Rev Sociol Am oilSrtfiaini Science in Stratification Social 316:1036–1039. Nature of theoretical Science aaaottepooto fnew, of proportion the about Data PNAS. efratdalato ae rsn in present names author all formatted We a edwlae o refrom free for downloaded be can 6:700–711. 32:391–403. 344:1422. Science hscss cesdFbur 4 2018. 14, February Accessed physicists. 322:1259–1262. Ui fCiaoPes Chicago). Press, Chicago of (Univ Nature 5,9 publications 251,294 ) 518:283–284. e fScience of Web https://github.com/ rcNt cdSci Acad Natl Proc Nature esPsychol Pers c , 444:1003– C Nature and , can- Proc eg ´ e ´ 1 iar ,Wn ,DvleP ogC Barab C, Song P, Deville D, Wang R, Sinatra compar- 21. for indicators citation of fairness the Testing (2012) C distributions: Castellano citation F, Radicchi of Universality 20. (2008) C Castellano S, the Fortunato of F, narrative misleading Radicchi The (2017) 19. DB Larremore A, Clauset AC, Morgan SF, Way 18. 2 ovkV,Sahie R(09 uhrnm iabgaini MEDLINE. in disambiguation name Author (2009) NR Smalheiser Barab VI, Torvik D, Wang 32. M, of Szell science P, the Deville in R, predictions Sinatra Data-driven (2017) 31. R Sinatra DB, Larremore determine A, mechanisms Clauset assembly Team (2005) 30. LAN Amaral J, Spiro B, Uzzi R, Guimera 29. Accessed https://xkcd.com/435/. at Barab C, Available Song in D, sciences. Wang authorship of alphabetical 24. Purity of use (2008) the R of Munroe analysis 23. empirical An (2012) L Waltman 22. 8 ig ,Rgn R hrnuP(09 h esamno?Alniuia assessment longitudinal A mentor? a gets Who qualitative (2009) A P Tharenou perspective: BR, mentor’s Ragins R, The Singh (1997) 28. SM Burroughs ML, Poteet TD, Allen developmental A work: 27. at mentoring investigation. Reconceptualizing (2001) empirical KE Kram An MC, Higgins mobility: career 26. and Mentorship (1992) TA Scandura 25. eutsoni i.2ta hr sasgicn ifrnebetween difference significant a the is with there consistent that datasets, Nonethe- 2 both chance. Fig. for by observed by in robust occurred ordered shown are this are result conclusions where authors our publications the less, also where but publications choice, include numbers 20.8% these and version chemistry, a within analysis: one, 30.9% our 17.7% medicine, another alphabetically within of are removes and authors versions This where account, publications two ordered. into understand disregarded perform have to publications we we such, where all results, As taking (17). the one, alphabetically affects authors this order how to Ordering. common Alphabetic to is Results of Robustness onain(..;adTeIdpnetRsac udDnak(S.L.). Denmark Fund Research Villum Independent The The (R.S.); Initiative University and European Grants (S.L.); Themes Central Foundation Intellectual by Research funded Euro- the Data” A.-L.B.); Scientific “Just The The project and CIMPLEX; of R.S.); (R.S. 641191 Grant Foundation and Office program, Templeton Framework A.-L.B. Force H2020 (to Commission, Air pean FA9550-15-1-0364 by the and with supported FA9550-15-1-0077 support was for Gates work Alex and feedback ACKNOWLEDGMENTS. S5). Fig. niiulsinicimpact. scientific individual counts. citation fractional of case The 6:121–130. domains: scientific across ison impact. scientific 17272. of measure objective an Toward trajectory. productivity faculty canonical nw icvData Discov Knowl Phys science. performance. team and structure network collaboration 2018. 20, September publishing. scientific ftersn trhypothesis. star rising the of agenda. research future and inquiry perspective. network Behav Organ 342:127–132. 11:791–796. C Science n hto h lhbtclnl model null alphabetical the of that and 13:169–174. 355:477–480. PNAS 3:1–29. Informetr J cdMngRev Manag Acad s L(03 uniyn ogtr cetfi impact. scientific long-term Quantifying (2013) AL asi ´ ..tak ihe zl o sfldsusosand discussions useful for Szell Michael thanks R.S. | ihnpyis 75.1% physics, within eebr1,2018 11, December Science oa Behav Vocat J ihnitricpiayjunl.Nt that Note journals. interdisciplinary within 6:700–711. falpbiain ihnbooy 14.4% biology, within publications all of 354:aaf5239. oa Behav Vocat J 26:264–288. rcNt cdSiUSA Sci Acad Natl Proc s L(06 uniyn h vlto of evolution the Quantifying (2016) AL asi ´ 74:11–17. s L(05 etr fphysics. of century A (2015) AL asi ´ rcNt cdSiUSA Sci Acad Natl Proc | 51:70–89. ncransinicfilsit fields scientific certain In ihnmteais 23.3% mathematics, within o.115 vol. e fScience of Web Science C alphabet | 114:E9216–E9223. 308:697–702. o 50 no. IAppendix (SI Informetrics J aa This data. 105:17268– C Trans ACM | Science 12607 Nat J ,
SOCIAL SCIENCES APPLIED PHYSICAL COLLOQUIUM SCIENCES PAPER