Lecture 6

LECTURE 6. SCIENTIFIC NETWORKS

Sociology of science

- It aims to understand how knowledge is produced and validated - It asks questions regarding the structure of power, hierarchies,

controversies, and alliances which underlie the work of science itself.

Various traditions:

• Merton• Kuhn• The strong programme• Laboratory studies• ANT• Bourdieu

MertonRobert King Merton (July 4, 1910 – February 23, 2003) was an American sociologist. He spent most of his career teaching at Columbia University. In 1994 Merton won the National Medal of Science for his contributions to the field and for having founded the sociology of science

First sociology of science: study of how science is produced, and of scientific careers.

Differentiation for sociology of knowledge (Mannheim), which, in line with Marx and Engels, claimed that people's ideologies, including their social and political beliefs and opinions, are rooted in their class interests, and more broadly in the social and economic circumstances in which they live. Study of ideologies and the social context in which things are believed to be true.

Merton studies researchers and scientific institutions within a structural functionalism perspective.

Studies of RECOGNITION: how scientists obtain scientific success.Correlations between quantitative indicators:• Number of publications (many scientists publish many articles of low significance, few

scientists produce few articles of high significance)• Honorific awards and membership of honorific societies as forms of recognitions• Positions in top ranked departments• Number of citations• Nobel prices

The goal is to verify that the system of rewards is justifiable: the scientific institution is organised around a system of rewards whose FUNCTION is to confer recognition to the most valid scientists.

Science is organised in a scientific community with its own allocation system: the more creative researchers are pushed to produce more, the less creative are relegated in administrative roles.Recognition at an early start (University students who get a research assistant position in a prestigious department). Late boomers are rare.

WEAK PROGRAM. This approach accept a logistic and objective definition of science: the scientific method and the logic principle is not under discussion.

Macro approach: individual variables are aggregated and correlated.

It justifies science by justifying inequalities as a logic outcome of meritocracy: the reward system rewards the best scientists an it is consistent along their careers

BIBLIOMETRY provides scientific administrators with a rational tool to evaluate the outputs of science, with no attention to the quality of citations (eg: positive VS negative citations)

NO ATTENTION TO CONFILCTS

KuhnThomas Samuel Kuhn (July 18, 1922 – June 17, 1996) was an American physicist, historian, and philosopher of science whose controversial 1962 book The Structure of Scientific Revolutions was deeply influential in both academic and popular circles, introducing the term "paradigm shift”.

Evolution of science is not a continuous process but a series of periods of normal science followed by revolutions.VSScience as a process of accumulation• Science is shaped by paradigms: sets of shared beliefs among members of a speciality area,

where interests converge around a specific set of problems identified as significant for the advancements of knowledge. By supporting the same paradigm, researchers also share specific understandings of which research techniques are appropriate for investigating these problems and a sense of identity that is constructed via interpersonal networks and the processes of information sharing.

• Rules of scientific method do not correspond to what happens in practice: in normal science scientists do not work to discover new theories, but a-critically accept the paradigmatic ones, and work toward building empirical proof of them.

• The paradigm determines the questions that can be asked, and science develops as a world independent from its social environment: AUTONOMY OF SCIENCE

Paradigms reach a point of depletion when there are no more interesting questions to explore. Revolutions happen when scientists start focussing on errors of the established paradigm. Thinking outside the box.The revolutionary scientists are people who can manage a paradigm so well that they are also able to overcome it.

The strong programme

Edinburgh schoolBloorRationality, objectivity and truth are local sociocultural norms, shared conventions.Scientific norms are like linguistic norms (Wittengstein): they are associated to specific sociocultural groups, whose practices are regulated by conventions only valid for those groups.

BarnesSocial interests influence the content and development of science through tactics of persuasions, strategic expedience, cultural dispositions. Scientific actions are not determined by the nature of things or by logic possibilities, but by social factors.

Bath schoolCollinsFocus on interactions between scientists: facts are collective constructions constituted within the interaction between who produces it and who receives it and attempt to replicate it.Falsifiability is an idealistic vision of the practice of science, where controversies are not resolved via replications, but with negotiations of a core set of interested researchers.

Laboratory studies

Knorr-Cetina, Gilbert and MulkayScientific objects are not only technically produced in laboratories, but also constructed with literary skills of persuasion and with political strategies for making alliances and mobilising resources.

Study of laboratory notes. Scientists do not normally mention unfavourable results, and present the favourable ones with rhetorical strategies that give strength to what they have found.They give the impression that arts, creativity and intuition does not have anything to do with production of science, pure result of the scientific method.

EMPIRICAL REPERTOIRE: typical of published papers, written in conformity with the empirical representation of scientific action

CONTINGENT REPERTOIRE: practical skills, traditional tricks, informal dialogue within scientists that does not follow the logical steps of scientific method.

Gilbert and Mulkay, Opening the Pandora’s box, 1984, p. 176

What he wrote (empirical repertoire) What he meant (contingent repertoire)

It has long been known that I haven’t bothered to look the reference

While it has not been possible to provide definite answer to these questions

The experiment didn’t work out, but I figured I could at least get a publication out of it

The W-PO system was chosen for detailed study

The fellow in the next lab had someone already prepared

Accidentally strained during the mounting Dropped on the floor

Handled with extreme care throughout the experiment

Not dropped on the floor

Typical results are show The best results are shown, i.e. those that fit the dogma

Agreement with the predicted curve isExcellentGoodSatisfactoryFair

FairPoorDoubtfulImaginary

ANT (Actor Network Theory)Latour and Callon

Actor–network relates different elements together into a network so that they form an apparently coherent whole. These networks are potentially transient, existing in a constant making and re-making. This means that relations need to be repeatedly “performed” or the network will dissolve. They also assume that networks of relations are not intrinsically coherent, and may indeed contain conflicts. Social relations are only ever in process, and must be performed continuously.

Actants denote human and non-human actors, and in a network take the shape that they do by virtue of their relations with one another. It assumes that nothing lies outside the network of relations. As soon as an actor engages with an actor-network it too is caught up in the web of relations.

A car is an example of a complex system. It contains many electronic and mechanical components, all of which are essentially hidden from view to the driver, who simply deals with the car as a single object. This effect is known as punctualisation.When an actor network breaks down, the punctualisation effect tends to cease as well. In the automobile example, a non-working engine would cause the driver to become aware of the car as a collection of parts rather than just a vehicle capable of transporting him or her from place to place. Depunctualization is like the opening of a black box. When closed, the box is perceived simply as a box, although when it is opened all elements inside it becomes visible.

Bourdieu

Field theoryAttention to the structural possibilities and constraints that shape scientific production. The field of science is a social field of strengths with its own structure and battles to conserve or change the hierarchical structure that is produced by the field itself. Scientists build up the field through their relationships, which means that the field structure is generated by actors' relationships. The possibility, for every actor to deform the field depends on his/her weight, which consist in the relative amount of symbolic capital which everyone owns. The symbolic capital is “the form that one or another of capital species [economic, cultural and social capital] takes when it is grasped through categories of perception that recognize its specific logic or, if you prefer, misrecognize the arbitrariness of its possession and accumulation”. This symbolic capital assumes a special feature in the scientific field and it is called scientific capital, made by connaissance and reconnaissance.Victory is measured in terms of the amount of scientific capital attributed to each scientist (and the institutions to which he or she belongs) by his or her peers. Therefore the stake is internal to the field and grants its autonomy from other social spheres.

Authors and schools

Elements of analysis Approach Limits/advantages

Merton PublicationsPrizesIndividual affiliations

Macro No controversiesMethods of construction of scientific object is not the object of sociology of science

Kuhn Themes Macro Focus on controversies, communities boundaries defined by sets of problems

Edinburgh School

Local socio-cultural norms Micro Focus on local interactions rather than on structures and dispositions

Bath school

Laboratory studies

Scientific objects as cultural constructions

Micro Regularization in an empirical repertoire is a local and individual strategy to gain prestige.

ANT Networks of human and non humans Micro No empirical detection of networksLock in effects of networks are rigid

Bourdieu Individual variables (combination of capitals) are abstracted to form position in the social structure

Micro and Macro

No empirical detection of networks of concrete relationships

Network approaches to the study of science

• Not limited to the micro level of local interactional mechanisms, and not jumping from the micro to the macro abstracting individual features: SNA focuses on empirical connections

• Relationships between scientists cannot be reduced to co-authorship and co-citations, but involve a wider set of interactions, from competition for research funding to affiliation to different organizations

Two traditions:

1. PATH ANALYSIS2. MULTILEVEL NETWORKS

PATH ANALYSIS

Focus on empirical connections to test the robustness of various theories in sociology of science and to study the evolution of several disciplines

Diana Crane, 1969, Social Structure in a Group of Scientists: A Test of the "Invisible College“, American Sociological Review, Vol. 34, No. 3, pp. 335-352

The existence of "invisible colleges" has been difficult to prove:• The boundaries of research areas are difficult to define since most scientific

work can be classified in numerous ways. • Scientists have many contacts with other scientists in their own research

areas and in other fields, some fleeting, some lasting. If social organization exists in a research area, it is of a highly elusive and relatively unstructured variety.

The existence of social organization within a research area may be inferred (a) if scientists who have published in the area have more social ties with one

another than with scientists who have not published(b) if scientists who have published in the area can be differentiated by degree

of social participation within the area.

Informal communication regarding research findings, research-in-progress, and research techniques represents one way in which members of a problem area can be linked to one another.- information about informal communication was obtained only from those

currently engaged in research in the field, a subsample of 52 members from a total of 147 respondents.

In addition to informal communication, several other types of ties between scientists exist. Collaboration occurs in several ways.- In the case of formal collaboration communication between two or more

scientists about their research was so important that it received formal recognition in the publication itself.

- Another form of collaboration takes place when a student writes a thesis under the direction of one or more teacher

- intellectual linkages represented by the influence of one scientist's work upon that of another. Citation references in journal articles are direct indications of such influence (can also be measured by asking scientists to name others who have influenced them in their selection of problems and techniques )

NB: many articles include numerous citations, but the relative importance of each citation may vary consider-ably from a reference to a scientist whose work has had a very strong influence on the author to a scientist whose work is relevant only in connection with a minor point

Combining several indicators into a composite index supplies further information about the group. Some members may be related to other members through influences on the selection of problems or techniques, others through some type of collaboration or through informal communication, but, if social organization does exist in a research area, most members should be related to others in at least one of these ways.

The choices of members of a group are arranged in matrix form where one axis represents choices made by a member of the group; the other axis, choices received by a member

Measures:• Degree (indegree and outdegree)• Density• Analysis of subgroups:Members of the problem area were divided into five groups on the basis of productivity

(1) 8 High Producers, each of whom had published more than ten papers in the area(2) 11 Moderate Producers, who had published four to ten papers in the area; (3) 33 Aspirants, who had published fewer than four papers in the area. (4) 9Defectors, each of whom had published four to ten papers in the area(5) 86 Transients, each of whom had published fewer than four papers in the area

Boundaries

At least four methods of locating members of a problem area are possible: the use of bibliographies, abstracting services, citation networks and sociometric data.

Selected scientific field:The diffusion of agricultural innovation

Questionnaires were sent to 172 of the 221 scientists listed,10 both junior and senior authors; 147 replies were received. Each respondent was sent a letter which included references to his publications as listed in the bibliography and was requested to respond to the questionnaire with respect to those publications only

Out of a total 1351 choices made by all respondents on all the different types of ties, outsiders were named 684 times (51%) and problem area members 667 times (49%).Does this suggest that a social group within the area did not exist? the majority of "outsiders" were selected only once. Within the problem area, about half the members were never named, but most of these scientists had been relatively unproductive. On the other hand, 26, or 27% were named more than five times. 15, or 7%, were named more than ten times. Thus, the social organization of the problem area appeared to be centered around a small and relatively productive proportion of the total membership.

Respondents were asked to indicate if they had any personal acquaintance with the scientists whom they mentioned as having influenced their selection of problems in the area. • of 246 choices of problem area members, 76% were designated as personal

acquaintances• of 256 choices of outsiders, 57% were indicated to be personal acquaintancesThis outcome suggests that to some extent the influence of outsiders was exerted through publications, while that of insiders was exerted through personal contact.

LONGITUDINAL ANALYSIS• During the first ten years of activity in the area, only 5 % of the members were active.• Between 1948 and 1958, the number of authors entering the area doubled every three

years. • After 1958, the number of authors entering the field doubled every five years• The increase per year stabilized at approximately 17 authors per year.

The trend for the number of publications was similar. • Between 1951 and 1960, 49% of the authors entered the field, most of this increase

occurring during the latter part of the decade. • Another 46% entered between 1961 and 1966.

Trends for high producers:• Two High Producers entered the field in the first decade• In the middle of the second decade, each of these High Producers had a student who• also became a High Producer. Another thesis director and his student, both High

Producers, entered the field at the same time along with a collaborator who also became a

• High Producer. • By 1957, all of the High Producers had entered the field. • Although almost two-thirds of the Aspirants and more than one-third of the Transients

entered after 1960, only 11 % of the more productive scientists (those with more than three publications in the area) entered the field in that period.

SUBGROUPS ANALYSIS

On the basis of collaboration and student-thesis director relationships, members of the area were assigned to distinct subgroups of varying sizes A scientist was assigned to a particular group of collaborators if he had a published Collaboration with at least one of its members or had been the student or thesis director of at least one of its members.

• Before 1956, the group consisted of small groups of collaborators and student-teacher pairs and a number of isolates.

• After 1956, when the group as a whole increased in size, some of these small groups expanded. Two large groups emerged, with 27 and 32 members respectively, as well as several medium-size groups with five to 13 members. A number of new small groups with 2 to 4 members and numerous isolates appeared

It seems plausible that these large groups could have exercised control over the direction of research in the area. Since their approach to the field was so visible due to the large number of publications produced by these groups, scientists outside these groups who had other types of approaches might have found it difficult to exert a comparable influence

In this tradition, many network studies• Burt, R.S., 1978/79. Stratification and prestige among elite experts in methodological and

mathematical sociology circa 1975. Social Networks 1, 105–158. • Hummon, N.P., Carley, K., 1993. Social networks as normal science. Social Networks 15, 71–

106. • Hummon, N.P., Doreian, P., 1989. Connectivity in a citation network: the development of

DNA theory. Social Networks 11, 39–63. • Liberman, S., Wolf, K.B., 1998. Bonding number in scientific disciplines. Social Net-works 20,

239–246. • Lievrouw, L.A., Rogers, E.M., Lowe, C.U., Nadel, E., 1987. Triangulation as a research strategy

for identifying invisible colleges among biomedical scientists. Social Networks 9, 217–248.

Lazega et al. (2008): multi-level networks of two systems of superposedand partially interlocked interdependencies, one inter-organizational, the other inter-individual.

Combination of centrality measures at the individual level and at the organizational level: FISHES AND PONDS

Multilevel networks

Important distinction in the conceptualization of micro and macro levels. Micro level: the network of individual interactions, together with attributes.Macro level: the network of inter and intra ties within and between organizations.

Meso level links the two together via bipartite networks, accounting for interdependencies through matrix algebra.

Bellotti E., 2012, Getting funded. Multi-level network of Physicists in Italy, Social Networks, 34: 215-229

The data

Networks: - 10 bipartite networks of “people by projects”, one for each year from 1997 to 2006 summed up to observe overlap- University affiliation: “people by university” matrix obtained by attribute- “people by projects” multiplied by “university by people” = “university by projects”- “people by people” valued network of number of projects in common,and the equivalent valued network for Universities. Both networks are undirected

3116 physicists working in 73 Universities; 1122 of them, who work in 66 Universities, have been funded during the 10 years under analysis

Attributes:- University affiliation- Individual affiliation to physics sub-disciplines (experimental physics, theoretical and mathematical physics, material physics, nuclear and subnuclear physics, astronomy and astrophysics, earth system physics, applied physics, and history and didactic of physics) + scientists from other areas = interdisciplinarity of projects- Rank of scientists (full professor, associate professor, and researcher)- National coordinators

DEPENDENT VARIABLE: TOTAL AMOUNT OF FUNDING RECEIVED BY EACH SCIENTIST OVER THE 10 YEARS

Analysis of attributesH1. On average, national coordinators are more likely to obtain a larger amount of money for research than researchers who never lead a research group.

H2. On average, full professors are more likely to obtain a larger amount of money for research than other ranks.

H3. On average, researchers working on experimental physics, astronomy, and material physics, and in other disciplines, are more likely to obtain a larger amount of money than researchers working in other sub-disciplines.

42%

23%

6%

29%Professor

Associate

Researcher

Other rank

24%

10%

17%4%19%

1%

6%

1% 18%Experimental

Theoretical

Material

Nuclear

Astronomy

Earth

Applied

History and didactic

Other discipline

8 people (0.7%) have been national coordinators 4 times, 29 people (2.5%) 3 times, 64 people (5.7%) 2 times, 215 people (19.1%) 1 time, and 806 people (72%) have never been national coordinators.

Analysis of the micro level of collaborations

H4. On average, researchers with higher E-I index values for sub-disciplines are more likely to obtain a larger amount of money for research than the ones with lower values.

Two strategies of getting connected: brokerage VS closure

H5. On average, researchers with higher ego brokerage scoresand/or higher egonetwork density are more likely to obtain a largeramount of money for research than researcher with low values in one or both measures.

H6. On average, researchers working in core institutions are more likely to obtain a larger amount of money for research than researcher working in peripheral ones.

Analysis of the macro level of collaborations

Bari

BasilicataBenevento

Bergamo

Bologna

Brescia

Cagliari

Calabria

Camerino

CampusBioMedicoRoma

Catania

CatanzaroMagnaGraecia

Cattolica

ChietiPescara

Ferrara

Firenze

Genova

IASF

Insubria

LAquila

Messina

Milano

MilanoBicocca

ModenaeReggioEmiliaNapoliFedericoII

NapoliParthenope

NormalePisa

OssAstrofArcetri

OssAstrofCatania

OssAstronBologna

OssAstronBrera

OssAstronCagliari

OssAstronCapodimonteNapoli

OssAstronColluraniaTeramo

OssAstronPadova

OssAstronPalermo OssAstronRoma

OssAstronTorino

OssAstronTrieste

Padova

Palermo

Parma

Pavia

Perugia

PiemonteOrientale

Pisa

PoliBari

PoliMilano

PoliTorinoPolitecnicaMarche

Roma

RomaTorVergata

RomaTre

SISSA

Salento

Salerno

Sassari

SecondaNapoli

Siena

Torino

Trento

Trieste

Tuscia

UdineUrbino

Venezia

Analysis of the meso level of collaborations

Big fishes: scientists with a valued degree above the median = 5. 535 BF, 587 LFBig ponds: Universities with a number of appointed physicists above the mean = 42. 32 BP, 34 LP

Fish and Pond BP LP

BF 424 111

LF 460 127

H7. On average, big fishes in big ponds (BFBP) are more likely to obtain a larger amount of money for research than little fishes in big ponds (LFBP), big fishes in little ponds (BFLP) and little fishes in little ponds (LFLP).

Modelling funding achievements*

* checked by J. Koskinen using Network Disturbances model to correct for potential lack of independence between cases, following Leenders (2002)

Results

Levels H Significance R2

MicroAttributes

H1 confirmed Being a national coordinator is significant, and on average it increases the amount of funding

13.8%

H2 confirmed Rank is significant: full professors are on average more funded than others 17.6%

H3 not confirmed Working in experimental physics is significant, but it does not increase the average amount of money obtained by full professors.Working in applied physics, history and didactic, and other disciplines significantly decreases the amount of funding

18.9%

MicroIndividual ntw

H4 not confirmed E-I index calculated for disciplines is not significant 33.8%

H5 confirmed Brokerage is highly significant. Closure also has a positive impact on the average amount of funding, even if not as strong as brokerage, but the effect is positive only when brokerage and closure are included together in the model, closure alone not being significant.Brokerage strategy does not work for astronomers: the values for their discipline become negatively significant when brokerage values are taken into account. It also diminishes the gap between full professors and other physicists’ ranks

MacroOrganizational ntw

H6 not confirmed Working in a core institutions is not significant 33.8%

MesoFish and ponds

H7 partially confirmed

Little fishes are generally penalized against the big fishes, but little fishesin big ponds are more penalized than little fishes in little ponds.The meso level variables diminish the positive effect of brokerageand closure

35.0%

Recap

Several approaches in sociology of science

• Merton• Kuhn• The strong programme• Laboratory studies• ANT• Bourdieu

Two main approaches in SNA• Path analysis• Multilevel networks