At what institutions did Nobel laureates do their prize-winning work? An analysis of biographical information on Nobel laureates from 1994 to 2014

In this study we examined the institutions (and countries) the Nobel laureates of the three disciplines chemistry, physics and physiology/medicine were affiliated with (from 1994 to 2014) when they did the decisive research work. To be able to frame the results at that time point, we also looked at when the Nobel laureates obtained their Ph.D./M.D. and when they were awarded the Nobel Prize. We examined all 155 Nobel laureates of the last 21 years in physics, chemistry, and physiology/medicine. Results showed that the USA dominated as a country. Statistical analysis also revealed that only three institutions can boast a larger number of Nobelists at all three time points examined: UC Berkeley, Columbia University and the Massachusetts Institute of Technology (MIT). Researcher mobility analysis made clear that most of the Nobel laureates were mobile; either after having obtained their Ph.D./M.D. or after writing significant papers that were decisive for the Nobel Prize. Therefore, we distinguished different ways of mobility between countries and between institutions. In most cases, the researchers changed institutes/universities within one and the same country (in first position: the USA, followed, by far, by the United Kingdom, Japan and Germany).


Introduction
The Nobel Prize is the most prestigious and renowned research prize for outstanding contributions in physics, physiology/medicine, literature, and peace, and it attracts widespread attention not only within but also outside the world of academia and science (www. nobelprize.org). Since research prizes in general (and the Nobel Prize specifically) can be used as indicators of research achievements, and as information on research prizes is usually well accessible, numerous scientometric studies investigating Nobel laureates have been conducted. For example, Zhou et al. (2014) examined 362 landmark papers written by Nobel laureates in physics from 1901 to 2012 using bibliometric methods (Journal Impact Factor, citations of landmark papers, country where the journal is published). In two recent studies, Chan et al. (2015a) and (b) looked at alterations of co-authors on the laureates' publications, and Wagner et al. (2015) compared Nobel laureates with a matched group of scientists to examine productivity, impact, and research networks. In further studies, publications by Nobel laureates have been used, for example, to validate (newly suggested) bibliometric indicators (Antonakis and Lalive 2008;Aziz and Rozing 2013;Rodríguez-Navarro 2011a, b), to test the quality of Google Scholar as a source for citation data (Harzing 2013;Patel et al. 2013), to predict the awarding of Nobel Prizes (Ashton and Oppenheim 1978), to study the uncitedness of publications by reputable scientists (Egghe et al. 2011;Heneberg 2013), to determine the effect of the Nobel Prize on the citation impact of publications by a Nobel laureate (Frandsen and Nicolaisen 2013;Gingras and Wallace 2010;Mazloumian et al. 2011), and to investigate the relationship between the number of highly-cited papers and the awarding of the Nobel Prize (Chuang and Ho 2014;Laband and Majumdar 2012).
Previous studies not only examined Nobel Prizes and Nobel laureates using bibliometric methods, but also analyzed the event itself and the person as such from a sociology-ofscience perspective. Becattini et al. (2014) studied the time delay between when a scientist makes a prize-winning discovery and is recognized for it with the Nobel Prize. They found that from the very beginning of the Nobel Prize awards, this lag time has continuously increased and Nobel Prize winners have become proportionally older and older at the time of their awards. On average, the lag time is almost twice as long in chemistry (9 years) and physiology/medicine (11 years) than in physics (5 years) (Chan and Torgler 2013). Before being awarded the Nobel Prize, most Nobel Prize winners received a striking number of other awards (Chan et al. 2014a, b) or were invited more frequently than other scientists to join scientific societies (Chan and Torgler 2012;Chan et al. 2016). In a series of studies, Campanario (1993Campanario ( , 1996Campanario ( , 2009) examined resistance in the Nobel laureates' scientific communities to recognizing the later honored work (for example, a journal rejecting the paper on the later prize-winning work). Campanario's studies clearly pointed out that rejections and resistance in the scientific community actually occurred. Stephan and Levin (1993), Jones and Weinberg (2011) looked at the age of Nobel Prize winners and examined the relationship between age and scientific productivity and creativity.
With this paper, we take up from one of the most important empirical studies in this area entitled ''Scientific elite. Nobel laureates in the United States'' by Zuckerman (1977). The author tracked all Nobel laureates in the USA awarded the prizes from 1901 to 1972. Zuckerman (1977) focused on the question, which social factors and social conditions ''make'' Nobel laureates. The author investigated their social development, educational background, collaboration with other authors, and the specialty of the prize-winning research. For this purpose, Zuckerman (1977) interpreted the data also on the basis of interviews with laureates (e.g. in order to explain the mobility of laureates who ''moved '' or ''stayed''). Zuckerman (1977) examined the prize-winning research, looking at US research institutions where the prize-winning work was done (see Zuckerman 1977, p. 170, table 6-3). Such research institutions are of special interest generally, because it is considered that they provide very good research conditions. A similar research approach was used by Ye et al. (2013), who analyzed the awards of 66 Nobel laureates in physiology/ medicine during the period from 1983 to 2012. They reported one to at most four landmark papers describing the work decisive for Nobel Prizes. Furthermore, they listed the journal of the most cited work (see Ye et al. 2013, p. 536, Table 2). Knowing the institution with which the researcher is affiliated at the time of doing the decisive work is important, because many Nobel laureates were awarded the Nobel Prize many years after they did the prize-winning work, and because researchers tend to be mobile-that is, they often change research institutions.
Besides Zuckerman (1977), we take up from another important empirical study in the area of Nobel Prize analysis: Hillebrand (2002) undertook a biographical analysis of the Nobel laureates in physics from 1901 to 2000. In the analysis, he considered information about the Nobel laureates, e.g. age, teamwork or migration within countries. He focused on the laureates' curriculum vitae, e.g. their social responsibilities. The author concluded that ''success is made more likely by an early interest in science, a good education, hard work, mobility (on occasion), as well as a generous portion of luck'' (Hillebrand 2002, p. 93).
In this study we examined the institutions the Nobel laureates of the three disciplines chemistry, physics and physiology/medicine were affiliated with (from 1994 to 2014) when they did the decisive research work. To be able to compare the results, we also identified the Nobel laureates' institutional affiliations at the time point of their obtaining a Ph.D./ M.D. and at the time point of their being awarded the Nobel Prize. In addition, we examined the Nobel laureates' mobility across the three time points.

Sources
The names of all Nobel laureates and their institutional affiliations on the day of the Prize announcement were found on the Nobel Prize website, www.nobelprize.org. The website also provides a broad summary statement naming the research achievement or discovery for which the prize was awarded as well as information on whether the prize was awarded to one person or shared by two or three persons maximum. This summary statement was the most important and the only (official) indication of the honored research work/papers of the Nobel laureates.
In the three prize categories (physiology/medicine, chemistry, physics), 155 scientists were awarded the Nobel Prize in the 21 years from 1994 to 2014. Compared to the prize categories physiology/medicine and chemistry, most prizes (55) were awarded in physics. There were 50 Nobel laureates each in physiology/medicine and chemistry.
None of the Nobel laureates examined here was awarded a second Nobel Prize, which has occurred only four times in the past. But we found differences in the categories with regard to whether the Nobel Prize was awarded to one person or shared by two or three persons: whereas in the period from 1994 to 2014 the Nobel Prize for physics was shared by two or three persons every year (without exception), the Nobel Prize for chemistry was awarded to one person (unshared) five times (Georg A. Olah 1994, Ahmed H. Zewail 1999, Roger D. Kornberg 2006, Gerhard Ertl 2007, and Dan Shechtman 2011. In physiology/ medicine, the prize was awarded to one person (unshared) three times (Stanley Prusiner 1997, Günter Blobel 1999, and Robert G. Edwards 2010. In most cases, information on the date and place where the Nobel laureates obtained their Ph.D./M.D. was available in the Encyclopaedia Britannica (see www.britannica. com). But the Britannica was mainly important as a reference work for information on the course of the Nobel laureates' careers. The information summarized in Encyclopaedia Britannica made it possible to narrow down the time frame in which the prize-winning work was done or to find out what the Nobel laureates' major research achievement was. The Encyclopaedia Britannica was the most important source of information next to the Nobel Prize website. However, as there are sometimes gaps in accounts of the Nobel laureates' careers in the Encyclopaedia Britannica, it made sense to also consult current university and institute web pages, which in most cases provide information on their Nobel laureates. In this way, information was obtained on the workplaces of all 155 Nobel laureates examined.

Determining the prize-winning work
There is no one decisive publication as such, which could clearly indicate the affiliation, where the laureate did his decisive work. The Nobel Committees did not name any relevant publications as official justification. However, the summary statement issued by the Committee (www.nobelprize.org), naming the research achievement or discovery for which a Nobel Prize was awarded, narrows down the topic of the prize-winning work.
Utilizing the Nobel Committee's reason for which the prize was awarded and the information on a laureate's career taken from the Encyclopaedia Britannica, we identified the paper(s) in which the researcher described the prize-winning work. We searched for landmark papers of the laureates in literature databases such as Web of Science (WoS, Thomson Reuters) or Scopus (Elsevier), that-among other things-capture citation counts and the addresses of authors of publications (Scopus since 1996). To determine lacking publications about the prize-winning work, we searched for books via Google Scholar and used further databases (e.g. ProQuest or Wiley Online Library). For example, K. Tanaka's (Nobel Prize 2002 in chemistry) prize-winning paper (Tanaka et al. 1988) was found neither in WoS nor in Scopus, but in Google Scholar.
In any case, we took the author's institutional affiliation named in the paper as the institution where the researcher wrote the prize-winning publication. Almost always, not just one paper, but rather several papers on the same topic (up to five) came into question. Based on the one or several relevant publications, we then determined the institutions where the Nobel laureates were working when they gained the important research results.
A difficulty encountered in this search was that, in some cases, Nobel laureates published several papers on the relevant topic within the same year. In that case, we used further analysis methods. After searching WoS and Scopus for all publications by the correct author (established clearly by using initials, biography, etc.), we examined the content of the most highly cited papers more closely by reading the abstract and, when necessary, checking the full text as well.

Results
The characterization of the prize-winning publication(s) As to the document type of publication, we found that in most cases (about 95 %), the prize-winning work was published in the form of an 'article,' which is the most relevant document type in the natural sciences. Occasionally, we found signed letters as a document type for the prize-winning work of F. Englert, (Nobel Prize 2013 in physics) in Englert and Brout (1964), for the laureate's work of H. Kroemer (Nobel Prize in physics 2000) in Kroemer (1963) and for the prize-winning work of P. Mansfield (Nobel Prize 2003 in physiology/medicine) in Mansfield (1977). In addition, we found two book chapters by J. O'Kneefe (Nobel Prize 2014 in physiology/medicine) in O' Keefe and Nadel (1978) and by G. Blobel (Nobel Prize 1999 in physiology/medicine) in Blobel and Sabatini (1971). Two further papers were published as a meeting abstract-R. F. Furchgott (Nobel Prize 1998 in physiology/medicine) in Furchgott et al. (1987)-and as note-P. A. Grünberg (Nobel Prize 2007 in physics) in Binasch et al. (1989). In physics, at least one researcher, J. Kilby, was awarded the Nobel Prize, who applied in 1959 for a patent using his work on the invention of the world's first integrated circuit (IC) chip (Kilby 1959, Patentnumber:US3072832). He worked at Texas Instruments Inc. (Bell liscensee) in Dallas (TX) as an employee from 1958 to 1970. The results show that in addition to publications in renowned scientific journals like Nature, Science, Cell and Physical Review Letters, less well-known journals also helped the way to the Nobel Prize. For example, we found a non-English publication (in French) of Y. Chauvin (Nobel Prize 2005 in chemistry) in Herisson and Chauvin (1971).
Regarding the content of the prize-winning work of all 155 Nobel laureates, we defined three categories: (1) development of methods, e.g. useful in medical diagnosis or chemical ** A Gini coefficient of zero expresses equality among the values in a frequency distribution; a Gini coefficient of one maximum inequality synthesis, (2) discoveries of natural mechanism and phenomena, and (3) making mature products. In our analysis, medicine leads with 18 discoveries of natural mechanism (85.7 %), followed by physics with 15 (71.4 %) and chemistry with 10 (47.6 %). In contrast, chemistry ranks first with nine methodical prize-winning works (42.9 %), followed by physics with seven (33.4 %) and physiology/medicine with three (14.3 %). The definition of a mature product is an industrial product development, like the invention of LED-lights (Nobel Prize in physics 2014, Nakamura, Amano and Akasaki) and ''for basic work on information and communication technology'' (Nobel Prize in physics 2000, Kroemer, Zhores, Kilby). Mature products were seldom described in decisive publications; only chemistry and physics hold two prize-winning works (both 9.5 %), physiology/ medicine has none. A peculiarity in the awarding of the Nobel Prize during the examined time period in both physics and chemistry was of interest: Six scientists with a university degree but without a Ph.D. or M.D. were awarded a Nobel Prize: Koichi Tanaka  Institutions with which the Nobel laureates were affiliated Table 1 lists the institutions at which Nobel laureates of the last 21 years obtained their Ph.D./M.D., at which they made their prize-winning discovery, and with which they were affiliated at the time of the Nobel Prize award. In order to see whether the institutions listed in the table are also top-rated institutions in university rankings, we included the ranking positions provided by Claassen (2015, Fig. 2, p. 800). Claassen (2015) presents a metaranking of universities which is based on major university rankings. The comparison shows that-with the exception of the Nagoya University-ranking positions below 30 are more frequently among the institutions where the Ph.D./M.D. was obtained than among the institutions where the prize-winning work was done or the Nobel Prize awarded.
We also calculated Gini coefficients as a measure of statistical dispersion for the number of scientists in Table 1. These coefficients indicate that the scientists are more equally distributed among the institutions where the prize-winning work was done (Gini = 0.17) and the Nobel Prize awarded (Gini = 0.19) than among the institutions where the Ph.D./M.D. was obtained (Gini = 0.24). Obviously, certain institutions (like Harvard University) are not only able to recruit promising Ph.D./M.D. candidates, but offer also fruitful environments for starting a successful career in science.
In detail, Table 1 shows that most of the Nobel laureates obtained their Ph.D./M.D. in the USA at Harvard University (n = 14), the University of California, Berkeley (UC Berkeley) (n = 8), and Massachusetts Institute of Technology (MIT) (n = 6). Harvard University stood out from the others by far; with almost twice as many future Nobel laureates as UC Berkeley, it fulfilled the criterion as the most important university for future Nobel laureates.
Regarding affiliations while doing the relevant prize-winning work/paper, no single institution stood out with a very high number of persons: leading the list of institutions here were Cambridge University (U.K.) (n = 8), followed by UC Berkeley and Bell Laboratories, or Bell Labs (formerly AT&T Bell Laboratories and Bell Telephone Laboratories), with six persons each. Bell Labs, which took on the research functions for the American Telephone and Telegraph (AT&T) company-a North American telecommunications company-is one of the few companies that can boast Nobel laureates (see http://ethw.org/ Bell_Labs).
The research institutions with which a Nobel laureate was most frequently affiliated at the time of the Nobel Prize award were Stanford University (n = 10) and MIT (n = 6) (see Table 1). Looking at the research institutions across the different stages of the scientists' careers, there were only three institutions that fulfilled the criterion at all three time points (Ph.D./M.D., prize-winning work/paper, Nobel Prize): UC Berkeley, Columbia University and MIT. A number of institutions were on the list for two of the three time points (such as Harvard University, Cambridge University U.K., Yale University and Technion-Israel).
The results make clear that the Nobel laureates did their prize-winning work/paper mainly at institutions in the USA. Still, the results differ greatly regarding the institution at which the prize-winning work was done: in Zuckerman's (1977) study, the two most Australia 3 frequent institutions where the prize-winning work was done were Harvard University (n = 13) and Columbia University (n = 9), but in this study they were Cambridge University U.K. (n = 8), University of California, Berkeley (n = 6) and AT&T Bell Labs (n = 6) (see Zuckerman 1977, p. 171, table 6-3). In this study, only four Nobel laureates did their prize-winning work at Harvard University. The differences in the results of both studies are probably due to the fact that the time periods investigated in the studies were not of the same length and also that the historical and social contexts were different: Zuckerman (1977) looked at the period from 1901 to 1972, whereas this study examined the period from 1994 up to 2014. Similar to the results of this study are the results by Charlton (2007), who examined revolutionary biomedical science between 1992 and 2006 using Nobel Prizes, Lasker awards (clinical medicine) and Gairdner awards. However, Charlton's (2007) study only looked at the time point of the Nobel Prize award and the field of biomedicine. In first place with the greatest number of Nobel laureates, Charlton (2007) found with n = 6 for MIT a similar number as this study, but also the University of Washington at Seattle (also with n = 6), which is not included in the list of institutions in this study (see Table 1). In addition to examining the institutions with which the Nobel laureates were affiliated, we also looked at countries for the three time points of Ph.D./M.D., work/paper, and the Nobel Prize award. Table 2 shows the distribution of the 155 Nobel laureates in different countries at the three time points. As in Table 1, we listed only those countries in which we found at least three Nobel laureates. The table shows clearly that the USA was the country having apparently the best conditions for promoting a Nobel Prize winner. A similar result was reported by two other studies, which examined Nobel laureates in biomedicine (Charlton 2007) and economics (van Dalen 1999). Chan and Torgler (2015), who looked at Nobel laureates in physics, chemistry, and physiology/medicine between 1900 and 2000, found that ''researchers educated in Great Britain and the US tend to attract more awards than other Nobelists'' (p. 847). Table 2 illustrates an interesting finding for Japan. Compared to other countries, Japan is very well placed, although previous bibliometric studies found that Japan does not perform well on field-normalized citation impact (Bornmann and Leydesdorff 2013). Even though the number of Nobel laureates in a country and citation impact is used as indicators (proxies) for measuring the quality of research, they appear to measure different aspects of quality.
Nobel laureates' mobility As described above, the different institutions counted different numbers of Nobel laureates when they obtained their Ph.D./M.D., did the prize-winning work/paper, and received the Nobel Prize. This result points to considerable mobility on the part of the Nobel laureates. Table 3 visualizes their institutional mobility, distinguishing five types of institutional mobility behavior: 1. The Nobel laureates were affiliated with one and the same institution across the three career stages (Ph.D./M.D., prize-winning work/paper, Nobel Prize award). 2. The Nobel laureates obtained a Ph.D./M.D. at one institution and then moved on to another institution, with which they were affiliated while doing their prize-winning work/paper, and received the Nobel Prize.
3. The Nobel laureates obtained a Ph.D./M.D. at one institution and then moved to another institution, where they did their prize-winning work/paper. They received the Nobel Prize while affiliated with a third institution. 4. The Nobel laureates obtained a Ph.D./M.D. and did their prize-winning work/paper at one institution. They received the Nobel Prize while affiliated with another institution. 5. The Nobel laureates obtained their Ph.D./M.D. at one institution and then moved to another institution, where they did their prize-winning work/paper. They then returned to the first institution, with which they were affiliated when they received the Nobel Prize. Table 3 shows the Nobel laureates' changes of affiliations. The counts show clearly that only 10.4 % of the Nobel laureates remain at the same place during their entire career. The rest were mobile either after obtaining their Ph.D./M.D. namely 78.5 % or after doing their prize-winning work namely 89.6 %. These percentages suggest that successful careers are related to mobility.
We also looked at mobility on a country basis, since the Nobel laureates changed institutions not only within a country but also across countries. Again, we found five different types of mobility: 1. The Nobel laureates were in one and the same country across their three career stages (Ph.D./M.D., prize-winning work/paper, Nobel Prize award). 2. The Nobel laureates obtained a Ph.D./M.D. in one country and then moved to another country, where they did their prize-winning work/paper and received the Nobel Prize. 3. The Nobel laureates obtained their Ph.D./M.D. in one country and then moved to another country, where they did their prize-winning work/paper. They received the Nobel Prize while they were working in a third country. 4. The Nobel laureates obtained a Ph.D./M.D. and did their prize-winning work/paper in one country. They received the Nobel Prize while they were working in another country. 5. The Nobel laureates obtained their Ph.D./M.D. in one country and then moved to another country, where they did their prize-winning work/paper. They then returned to the first country, where they received the Nobel Prize. Table 4 shows the Nobel laureates' mobility behavior across countries. The results show that a large part of the Nobel laureates, 77 % (see Table 4), were not mobile across countries and worked in only one country during all time points.
Comparing Tables 3 and 4 indicates that Nobel laureates most frequently changed institutions within a given country. This fits the observation by Hillebrand (2002): ''Since 1950, almost all laureates have remained in the country in which they made their

Discussion
In modern science, evaluation of research is an increasingly important topic (Dahler-Larsen 2011;Power 1999). Whereas in the past the interest was in evaluating research presented in manuscripts or research proposals, today, entire institutions, research clusters and countries are being examined using indicators. Bibliometric indicators are certainly the most important class of indicators used (Moed 2005). With the aid of the underlying publication and citation data, evaluations can be done at any aggregation level-that is, from the individual researcher to entire continents. Bibliometric indicators can measure only the impact of science on science itself. But since science policy is interested in impact measurement above and beyond science and in other parts of society, scientometrics research is working on issues to measure this broader impact (Bornmann 2012(Bornmann , 2013. Research is conducted by persons. As research prizes are usually awarded to persons who have made outstanding scientific achievements, the prizes are also used as indicators of research performance (Rodríguez-Navarro 2011.
As opposed to bibliometrics, however, the criterion of a research prize has two major disadvantages that make it difficult to use research prizes as indicators: 1. Because citation rates vary widely across disciplines and the variation has little to do with scientific quality, citation scores are normalized for this difference across disciplines (Vinkler 2010). Only through normalizing the impact scores of research institutions (and other entities), conduct research and publishing in different disciplines can be compared to one another. Since it can be assumed that there are discipline-specific patterns also with research prizes (many prizes are awarded only in a specific field), a comparison of results on the number of research prizes per institutions having different disciplinary profiles is not possible. To our knowledge, no methods for producing normalized numbers of research prizes yet have been suggested (if they are possible at all). 2. The awarding of research prizes, especially Nobel Prizes, is a rather rare event. A number of conditions have to be met for a research prize to be awarded, and not all of these conditions have to do with research quality. It can therefore be assumed that with the Nobel Prize there have been many false negatives: a number of important scientists that actually deserved the prize for their research findings or discoveries did not win a Nobel Prize for the various reasons (that had scarcely to do with research quality). The results of this study demonstrate that the research prize as a rare event is based on a small numbers of cases-even when the data is evaluated at the level of countries. With only a small number of cases, there is always the risk that results will be unreliable.
For these reasons, the results of this study should be handled with caution if they are used in an evaluative context. They may better be used to draw the public's attention to topics investigated by the laureates (Chan et al. 2014a, b).
In this study we looked at the institution and country where a Nobel laureate did the work and was later awarded a Nobel Prize. To better understand the results of this time point, we in addition examined the number of Nobel laureates at the time points of obtaining their Ph.D./M.D. and when receiving the Nobel Prize. The results of the country analysis revealed that the USA dominates the country ranking. The institutional analysis shows that three institutions have a large number of Nobel laureates at all three time points: UC Berkeley, Columbia University and the Massachusetts Institute of Technology (MIT). The mobility analysis made clear that most of the Nobel laureates were mobile either after obtaining their Ph.D./M.D. or after doing the prize-winning work/paper. In most cases, the researchers moved from one institution to another within the same country (in the USA).
Explaining their individual motivations for moving or staying (using methods of qualitative research) could be a topic for future studies.
Despite the low numbers of events (n = 155 laureates) the results of this study in part resemble the findings gained by other studies. Analyzing citation impact, it was shown by Bornmann and Leydesdorff (2013) that countries such as USA, U.K., and Germany, having a high population number and a well working economy, lead in science. Bornmann and Bauer (2015) evaluated the list of 3216 researchers, who met the criteria of being highly cited researchers based on papers published between 2002 and 2012. They determined the number of highly cited researchers per institution and came to similar institutional rankings as shown in this study. A major difference to the current study is the fact that the Chinese Academy of Sciences ranks in the top ten of highly cited researchers in contrast to none Chinese Nobel laureate in one of the three disciplines before 2015 (Y. Tu received the Nobel Prize in physiology/medicine in 2015). We expect to see more Nobel laureates from China in the near future.
In this study, we examined a time point that was hardly previously analyzed: the career stage during which the researcher did the prize-winning work/paper. As laureates usually receive the Nobel Prize many years after this time point, the usual perspective in scientometrics, which attributes research achievements to institutions and countries at the award of the Nobel Prize, should be complemented by the perspective, which attributes achievements to the location where the researcher did the prize-winning work/paper.   Table 6 Detailed information on Nobel laureates in physiology/medicine Year      Akasaki, Amano, Nakamura: ''for the invention of efficient blue light-emitting diodes which has enabled bright and energy-saving white light sources''