Research Policy 39 (2010) 869881

Contents lists available at ScienceDirect

Research Policy

journa l homepage: www.e lsev ier .com

Career patterns and competences of PhDs in scieknowle om

Hsing-fea Manchester B Boothb Universit Pa ce

a r t i c l

Article history:Received 15 JuReceived in reAccepted 3 MaAvailable onlin

Keywords:UK research-bCareer typesPhDsCompetences

plexexamdoctoment. Thiseer tyoreablef knowe regregar

technical occupations. In absolute terms, general analytical skills and problem solving capability acquiredfrom doctoral education are perceived as valuable in all three career types.

2010 Elsevier B.V. All rights reserved.

1. Introdu

In the eogy PolicyTechnologycerned withfuture shoron the econthere wouldand the perneering.

This conwas echoedthe 1987 Dethat the deply and calland enginee1987). In tha projectionindicated th

CorresponInnovation Res6PB, United Ki

E-mail add

0048-7333/$ doi:10.1016/j.ction

arly 1990s, in the publications Science and Technol-(OECD, 1991) and Technology-Economic Programme:and the Economy (OECD, 1992), the OECD was con-the prediction among several member countries of a

tage of scientists and engineers and its possible impactomy. This prediction was based on both the belief thatbe an increased demand for scientists and engineers

ceived decline in students interests in science and engi-

cern about a future shortage of scientic labour forcein policy reports in a number of countries. In the UK,partment of Education and Science White Paper statedmand for highly qualied manpower outstripped sup-ed for an increase in the number of graduate scientistsrs (Department of Education and Science White Paper,

eUSA, in1991, theBureauof LabourStatisticsdevelopedof the labour force covering 19902005. The projectionat for scientists, engineers and technicians as a group,

ding author at: Manchester Business School, Manchester Institute ofearch, University ofManchester, Booth StreetWest,ManchesterM15ngdom.ress: laredo@enpc.fr (P. Laredo).

demand could increase by up to 59% (Braddock, 1992). Alterna-tively, a 1990 study by the National Science Foundation projectedthat there would be a shortfall of 675,000 graduates in naturalscience and engineering by the year 2006 (Finn and Baker, 1993).

The concern raised during this period about the future shortageof scientists and engineers and the possibility that their technicalknowledge and talent may not be properly exploited was justi-ed by the belief that having qualied scientists and engineersworking within the boundaries of the conventional scientic andengineering occupations was a key factor contributing to nationaltechnological competitiveness and economic growth (Dosi et al.,1994; Freeman, 1992). Consequently, policy responses includeda series of programmes for training scientists and expanding thenumber of PhDs in science and engineering in member countries(OECD, 1991).

More than a decade later, policymakers are still concernedabout the shortage of scientists and engineers due to the contin-ued lack of interest in science and engineering among students,but this time the concern is not just about how to keep scienceand engineering graduates in their conventional occupations. Thecontemporary argument is that, in the new economy, the basisof competition has changed and is increasingly driven by knowl-edge and intangible assets, with knowledge production becomingmore widespread and widely distributed across a host of newplaces and actors, in many cases outside conventional technicaloccupations (David and Foray, 2002). Therefore, in contrast to the

see front matter 2010 Elsevier B.V. All rights reserved.respol.2010.05.001dge economy: The case of graduates fr

n Leea, Marcela Miozzoa, Philippe Laredob,a,

usiness School, Manchester Institute of Innovation Research, University of Manchester,ris-Est (ENPC, LATTS) Cite Descartes, 6 av. Pascal, 77455 Marne-la-Valle Cedex 2, Fran

e i n f o

ne 2009vised form 10 February 2010y 2010e 1 June 2010

ased university

a b s t r a c t

Based on data collected through a comresearch-based university, this papertypes of competences acquired fromtypes. The results show that employdestination for the survey respondentsis also the destination of the other careducation are regarded as relatively msubject areas is regarded as more valusubject areas (butmoregeneral typeomore general and transferable skills arthe general and transferable skills are/ locate / respol

nce and engineering in thea UK research-based university

Street West, Manchester M15 6PB, United Kingdom

survey of science and engineering PhD graduates from a UKines the different types of careers and towhat extent differentral education are regarded as valuable in the different careeroutside the conventional technical occupations is the maincareer type isnotonly successful at retaining itsmembers, butpes. Moreover, different types of competences from doctoralvaluable in different career types: knowledge directly tied toin academia/public research; both knowledge directly tied toledge rather than specialist knowledge inPhD topics) and the

arded as valuable in technical positions inmanufacturing; andded asmore valuable in employment outside the conventional

870 H.-f. Lee et al. / Research Policy 39 (2010) 869881

attitudes in the late 1980s and the early 1990s, policymakers havebegun to recognise that one of the reasons for supporting theproduction of larger numbers of science and engineering gradu-ates is that, in the new economy, with more and more sectorsadopting new technologies, the demand for scientists and engi-neers is increasing outside the conventional boundaries of scienceand engineering occupations in order to adopt, produce and dif-fuse knowledge efciently (The Dearing Report, 1997; Foray andLundvall, 1in the econincreasingpersonnel sis becomingFinnie, 199are displacetors (MiozzRoberts revUK, entitledandengineement in mR&D.1

Hence, rentists andIn the UK, sPaper entitUniversitiesexisting poengineeringat doctoralempirical reas a passpois visible ithat becauswork in higtions shouldHowever,wtoral qualithe lack ofPhDs in acMangematiGiret and RRobin and Cditional cara questionthe rationaimplicationcally the caPhDs are mventional Sextent theS&E doctorapations.

The papecareers foracquired frorelevant todistinctionpotentiallyventional Sin the analymethods an6 summaris

1 Documenthtm (accessed

2. Careers of S&E PhDs

Scholars in innovation studies have pointed out the contribu-tions of S&E PhD personnel to the economy. Pelz and Andrews(1966) stresin their moMangematimanpower

sitiesother froducaant

s high S&Mowt thaedgenowh mogut,This

ry hastryexteme aineserv

Theirdentstry

ernmper

athecade

ciente sece pra frofurths forymensed tymenall tRC (

PhDars awereernmivatethree990)raducal eimpempinan

auseat dto

r ofematFoxal rewith996; OECD, 2000). Moreover, with structural changeomy, including the decline of manufacturing and theimportance of services, the amount of highly skilleduch as scientists and engineers in the service sectorincreasingly signicant (Cervantes, 2001; Lavoie and

8; Lavoie et al., 2003), as many jobs and functionsd or outsourced from traditional manufacturing sec-o and Grimshaw, 2006). Indeed, the 2002 Sir Garethiew of supply of science and engineering skills in theSET for Success, clearly stated that many scientistsrsmakecontributions to theeconomythroughemploy-any sectors, not only through working in industrial

egardless of the change in rationale, the demand of sci-engineers has been increasing over the last decade.uch demand has been re-afrmed in the 2008 Whiteled Innovation Nation (Department for Innovation,& Skills, 2008). However, most of the discussions in

licy statements or reports are based on science and(S&E) graduates. Whether scientists and engineers

level are experiencing the same trend is a matter ofsearch. Traditionally, doctoral education was regardedrt to academia or public research organisations. Thisn the Harris Report (1996) in the UK which statede many postgraduate research students might go toher education institutions, higher education institu-provide themwith proper training related to teaching.ith the huge increase in the number of peoplewith doc-cations, many studies have expressed concerns aboutjob opportunities for science and engineering (S&E)ademia or public research organisations (Dany andn, 2004; Enders, 2002, 2005; Fox and Stephan, 2001;ecotillet, 2004; Mangematin, 2000; Martinelli, 1999;ahuzac, 2003; Stephan et al., 2004). Whether this tra-eer type is the dominant one for S&E PhD graduates isopen to empirical research. Thus, given the change inle for the demand of scientists and engineers and thes for S&E PhDs, this paper intends to explore empiri-reer types of S&E PhDs and to investigate whether S&Eost likely to be employed within or outside the con-&E PhD occupations. Also, the paper studies to whatdifferent types of knowledge and skills acquired froml education are perceived as valuable in different occu-

r is organised as follows. Section 2 reviews the types ofS&E PhDs. Section 3 explores the knowledge and skillsmdoctoral education and towhat extent theymight bedifferent career types, paying particular attention to thebetween the conventional S&E PhD occupations and theincreasingly signicant employment outside the con-&E PhD occupations. Section 4 discusses the data usedsis, which is based on a complex survey, the analysingd measures. Section 5 presents the results and Sectiones the discussion and conclusions.

online available at: http://www.hm-treasury.gov.uk/ent res roberts.24 January 2010).

univeron thetransfetoral esignicence ithroug1981;suggesknowlsuch kthrougand Ko2003).industto indu

Theover tiand Irvomyob1978.responin induin govout thewere r1991, atoral swas thand thon dat1999,schoolemplosurpasemplopassedthePPAsored68 yedentsin govthe prvey of1980/1after gelectristudiesondaryits dom

Bectern thtrainednumbe(Mangment (doctorpeoplesed that PhDandnon-PhDpersonnel differ signicantlytivations and the quantity and quality of their output.n (2001) also pointed out the special nature of doctoralbecause its members, on the one hand, are trained inand contribute to production of new knowledge and,r hand, serve as an important channel for knowledgem academia to industry if they enter industry after doc-tion. Indeed, it has been argued that one of the mostbenets to the economy from public funded basic sci-hly trained manpower for industry and governmentE doctoral education (Lardo, 2007; Martin and Irvine,ery and Sampat, 2005; Pavitt, 1991). These argumentst PhDs may bring either the most up to date scienticthey have produced or their capabilities in producingledge into industry and result in knowledge spilloversbility across different employment contexts (Almeida1999; Madsen et al., 2003; Rosenkopf and Almeida,means that the extent of S&E PhDs employment ins an impact on how academic research is transferred.nt to which S&E PhDs are employed in industry shiftsnd seems to become increasingly signicant. Martin(1981) surveyed PhDs trained in two UK radioastron-atories (Jodrell Bank andCambridge) between1945 anddata revealed that at the time of survey, rst jobs fors were 55% in academia, 22% in government and 17%and the most recent jobs were 46% in academia, 29%ent and 20% in industry. This indicates that through-iod, career patterns for radioastronomy PhDs in the UKr stable. Stephan (1996) showed that in the US, up tomia remained the largest employment sector for doc-ists although the proportion was decreasing. Industryond largest employment sector for doctoral scientistsoportion was increasing. Stephan et al. (2004), basedm the US Survey of Doctorate Recipients from 1973 toer pointed out that for those who have left graduatemore than 5 years in all science and engineering elds,t in industry grew so rapidly that by 1989, industryhe tenure-track academic sector as the most commont sector for S&E PhDs and by the mid-1990s, it sur-ypes of academic employment. A UK survey targetingParticlePhysics andAstronomyResearchCouncil) spon-students (DTZ Pieda Consulting, 2003) estimated thatfter awards ended, in 2003, 15% of the sponsored stu-either in permanent university research positions orent/public sector research positions and 54% were insector. Enders (2002) German case, based on a sur-cohorts of German doctorates (1979/1980, 1984/1985,in 1999, reported that in the long run (1520 years

ation), only 40% of mathematics graduates and 20% ofngineering graduates were in higher education. Thesely that inmany counties, academia is becoming the sec-loyment sector for S&E PhDs, while industry is gainingce as the major PhD employment sector.these observations may indicate an employment pat-

iverges from the traditional expectation that PhDs arebecome academics, this has led scholars to discuss aissues. These include: the incentives for doing a PhDin, 2000), expectations and realities regarding employ-and Stephan, 2001; Mangematin, 2000), value of thesearch training (Enders, 2002, 2005), employability ofadoctoraldegree (DanyandMangematin, 2004), deter-

H.-f. Lee et al. / Research Policy 39 (2010) 869881 871

minants of S&E PhD career outcome (Giret and Recotillet, 2004;Mangematin, 2000; Robin and Cahuzac, 2003), and how S&E PhDsmay contribute to research activities in industry (Mangematin,2000; Stephan et al., 2004), particularly their role in the commer-cialisationStephan et

Despiteplored. Firsof scientistture, thereS&E PhDs. Wregarded asmanyuniveing more S&strategies oS&E PhD jobis often takepositions inassociated was chemicais unclear wemployed ior in banksPieda ConsuPhDs whodesign/solu24% were inapart fromresearch/noattention tooccupationscal positiontime. Hence

(1) Wherejobs/sec

3. Knowle

3.1. Purpos

In the Utraced backPrincipals (stressed twenable gradtive disciplitraining ena

These twcation, schoby some asreect the d(Park, 2005perceptionand advancof scholarsha thesis; a thas been psional trainof the procedently. As Pnature, withpline to maresearch qucontributio

to a certain extent, it is reasonable to assume that if a PhD awardis granted, it is a guarantee that the receiver is equipped with in-depth knowledge in a specic discipline and has the capabilitiesto design and implement an independent piece of research. This

s thag, boo coleavject-

ng apmeder deer btenccognanceymenkno

yeeschared. Iembsingnderacqur is lwledf car

arac

rews (1itione or aelds.pub

ned sHown sci; emunknontriconork wis skore, iacadr of relvesrtlye of tnormly aantic scerthhe cscie

resporenesectoGibb1997of academic results (Lam, 2007; Murray, 2002, 2004;al., 2007; Zucker et al., 2002a,b).all these developments, two areas remain largely unex-t, given the change in the rationale for the demands and engineers and the change in industrial struc-is scope to explore the resulting changes in careers of

hile academia and government may be traditionallythe main sectors for employment for S&E graduates,

rsities andgovernmentorganisationsmight beemploy-E PhDs for non-research tasks such as for developing

r policies. These are some examples of unconventionals within the conventional S&E PhD sector. Similarly, itn for granted that many S&E PhDs will occupy researchindustry, and these positions have traditionally beenith R&D laboratories in large rms in industries such

ls, pharmaceuticals, aerospace, semiconductors, etc. Ithether modern S&E doctorates are more likely to be

n these conventional research positions just mentionedor consultancy rms. Indeed, the UK PPARC case (DTZlting, 2003) showed that for those PPARC sponsoredworked in the private sector, 29% were in softwaretions/management, 24% were in nancial services andbusiness services. Therefore, to address these changes,bearing in mind the academia/non-academia and then-research distinctions, in this paper, we pay specialS&E PhD jobs within and outside the conventional PhD, i.e. academic or public research positions and techni-s in manufacturing, and how they may evolve through, this leads to our rst research question:

are S&E PhDs employed and have they changedtor?

dge and skills acquired from doctoral education

es of doctoral education

K, the ofcial purpose of doctoral education can beto the report by the Committee of Vice-Chancellors andCVCP) (1988). This report, entitled The British PhD,o main purposes of doctoral education: the rst is touates to make original contributions to their respec-nes and the second is to provide professional researchbling them to become independent researchers.o purposes or dimensions of modern doctoral edu-larship and professional training, which are regardedcompeting (Burgess, 1997; Leonard, 2000; Pole, 2000),ual nature of the PhD as a product and as a process). The scholarship dimension is rooted in the generaland requirement that a PhD thesis has to be originale disciplinary knowledge. This implies that the purposeip in doctoral education is assessed by a nal product,hesis has to demonstrate that some original knowledgeroduced. On the other hand, the dimension of profes-ing places emphasis on the process, the developmentdure and the capability to conduct research indepen-hD projects are open-ended scientic investigations inout in-depth understanding of knowledge in the disci-

ke an elegant argument and the ability to frame properestions and then to execute the research, an originaln to knowledge in the discipline is implausible. Hence,

implielearnindures tshouldare subable.

Usideveloa broaual carcompethat reimportemplotype ofemploon therewardedge isorganiand Zatencesanotheof knotypes o

3.2. Chcareer

TheMertonrecognductivtheir qualityrenownities.effect itionatewhiletheir centistslater wbutionTherefyoungnumbethemsthis, pabecausentistssuddensignicacadem

Nevwith tpublicsocialentrepferent1993;Leslie,t the process of a PhD study is a journey of individualth to acquire knowledge in the discipline and proce-

nstruct knowledge, and that successful post-graduatese university with knowledge and skills, some of whichspecic and others that are more general and transfer-

competence-based perspective in discussing careernt, DeFillippi and Arthur (1994) stressed that there isependence of organisational competences on individ-ehaviour and from a career standpoint, an individualses are dened through matched employment settingsise their potential contribution. That is, the relativeof a specic type of knowledge or skills in a specict sector can be dened through the extent to which thewledge or skills are perceived as valuable by individualin the sector, and the employees perception is basedacteristics of how the use of knowledge in the sector isndeed, this resonates with the idea that social knowl-edded in individual relationships that are structured byprinciples, i.e. a shared template (Kogut, 2008; Kogut, 1992). Hence, what distinguishes S&E PhDs compe-ired from doctoral education from one type of career toargely due to the difference in the structures of the usege and how use of knowledge is rewarded in differenteer.

teristics of the use of knowledge by different types of

ard system in academia has been largely based on973)universalismargument, stressing that professionaland rewards are given to those who are the most pro-ble to demonstrate the most signicant contribution toIn the academic setting, professional recognitionmeanslications, peer recognition (especially recognition fromcholars) and reputation within the scientic commu-ever, Merton (1973) further pointed out the Matthewence. That is, recognition in science is often dispropor-inent scientists gain disproportionately greater creditown scientists gain disproportionately little credit forbutions. Another interpretation is that the more a sci-tribution has been recognised, the more the scientistsill be appreciated. The recognition of scientic contri-

ewed in favour of established scientists (Merton, 1988).n order to be recognised at early stage of their career,emics have to establish a sizable lab with a reasonableesearch students to carry out the research and to devoteto more publications in renowned journals. To achievebecause of the need for a convincing track record, partlyhe efciency to carry out further research, academic sci-ally cannot afford to switch subject areas/disciplines

way from their PhD work. This naturally results in aimportance in knowledge in specic subject areas forientists.eless, in recent years, there is increasing concernhanging world of science. There is a consensus thatnce is increasingly assessed by accountability andnsibility and in many public research organisations,

urship and networking with a range of actors from dif-rs are enormously encouraged (Funtowicz and Ravetz,ons et al., 1994; Nowotny et al., 2001; Slaughter and; Ziman, 1996). As a result, researchers in public organ-

872 H.-f. Lee et al. / Research Policy 39 (2010) 869881

isations, in addition to their roles as scientists, are at the sametime becoming project managers and administrators to coordinateactors across sectors. These changes might challenge the tradi-tional reward system in public science and the competences thatresearch scibe expectegeneral andbecome a la

Industritations andthat industrscience andsional scienresearch anorganisatioresults. Inclear, teamBecause mabecause ofnew producand Rosenban industriprojects atling new pthe performities to handimensionskills betweresearchers

Howevetrial scientresearch founplannedproblems. Sbasic reseatages. In somight be cowith their dndings frogood commbasic researndings fromore, in soStephan (19industrial jpublicationneering, nefrom indusreported thandsecondliterature inutation for oin postgradin the UK (JThis impliechemicals atry is strongamount ofessary for ias industriaand race wworks for punderstandtive knowlegeneral intopics).

Many industrial scientists turn into dedicated managers gradu-ally through career progression (Biddle and Roberts, 1994; Lavoieand Finnie, 1998). Such role transformation indicates that thereare career moves for industrial scientists from the conventional

cal otionsearcordis ort devrespor emoblemrt frers,therThe

ss rsatiogadoign eralaccoifferel-depplye situwn to bee ane neocesteraconsulargsim

edgeare

ost opan

t evew thy knt onle th

s ofenc

blemblemwers

ometsidetioncase,whetherlikelthesequireery dabo

e usith Lsatiosatioentists should gain from their doctoral educationmightd to partly shift over time from substantive to moretransferable skills as management and administrationrger part of scientic life.al scientists generally work with very different expec-demands from the academic/public sector. It is arguedial scientists often face tensions between professionalindustrial organisation (Kornhauser, 1962). Profes-

ce concernsmainly contributions to knowledge, qualityd long-term program. On the other hand, industrialn favours prots, cost savings and normally short-termindustry, the key goal (a nal target or product) iswork is essential and deadlines are often very tight.nufacturing industry is highly product-oriented andthe high uncertainty and risks involved in developingts, rms normally adopt parallel strategies (Abernathyloom, 1969) for product development. This implies thatal scientist is likely to be involved in several researchthe same time. As the success or failure in control-roducts time to market will eventually translate intoance of individual scientists, industrial scientists abil-dle research projects are vital. This reveals a crucialdifferentiating the use of more general and transferableen industrial researchers and academic or public sector.r, this is not to say that the competences of indus-ists lie mainly in transferable skills. Firms do basicr many reasons. In some cases, basic research is theby-product of the attempt to solve specic industrialometimes rms such as biotechnology companies dorch that is near market to have rst-mover advan-me other cases, large rms, due to their market power,ndent enough to conduct basic research and expect,iversied products and resources, that at some point,m their basic research activities will eventually haveercial uses (Rosenberg, 1990). Firms might also doch in order to cultivate capabilities to absorb researchmother scientists (Cohen and Levinthal, 1989). Further-me companies, there is a strong culture of publishing.96), based on 1991 data, pointed out that, in the US,

ournal publications accounted for around 16% of totals for both the elds of chemistry and physics. In engi-arly a quarter of scientic and technical articles cametry. Globally, Godin (1996), based on the 1989 data,at chemicals and pharmaceuticals were placed in rstplace in termsofnumbersof industrial publications. Thedicates that in the pharmaceutical industry, rms rep-penness and commitment to publication are importantuate industrial scientists employment decisions, bothones, 1992) and in the US (McMillam and Deeds, 1998).s that the practice of publication in industries such asnd pharmaceuticals, where UK manufacturing indus-ly based, is long established. This shows that a certain

substantive knowledge in related subject areas is nec-ndustrial scientists to be seen as competent. However,l scientists often work in product-oriented projectsith time to launch new products, guring out whatroduct development is normally more important thaning deeply why the solution works. Therefore, substan-dge used by industrial scientists is more likely to be

certain subject areas rather than specic (as are PhD

technioccupatic reand cosuccesproducdirectgreateand pr

Apamanagjobs oturing.busineorganiof Haruct desfor sevmore,play din welity to ain somunknoneed tprovidpropostion prand insome cfrom aout theknowlto prep

Mcomthaknopanthasolvcesyouproproans

In sare ouoccupaIn anyless ofor in ois lessskills;and reserve v

Themay bline worganiorganiccupations to employment outside the conventionals. Dedicated managers very often do not conduct scien-h any longer but are involved with company strategiesnation among internal and external divisions. As thefailure in controlling new products time to market inelopmentmayhavebecome thesededicatedmanagersnsibility, this type of career move is likely to require

phasis on analytical skills, project management skills-solving capability.

om turning from research scientists into dedicatedmany PhD-trained scientists enter private sectors inthan research or technical departments in manufac-y often serve as consultants in knowledge-intensivems. The nature of their jobs is interdisciplinary, cross-nal and international, as demonstrated by the studyn and Sutton (1997), who illustrated how one prod-rm acts as a technology broker serving product design

hundred different rms in over 40 industries. Further-rding to Creplet et al. (2001), experts and consultantsnt roles in consultancy rms. Consultants often workned problems and their know-how lies in their abil-a particular toolbox in well-known contexts. However,ations, consultancy rms encounter problems that are

o their clients as well as to the rms and new solutionsdeveloped. The capability needed is not the ability toalogy between known problems and solutions but tow patterns of interpretation. This knowledge produc-s often involves operation of a new panel of knowledgetion with epistemic community. This capability leadsltants to be regarded as experts. Indeed, a team leadere international engineering consultancy rm pointedilarity between experts and doctoral students in theirproduction process (preliminary interview conductedthe survey):

f the projects come to my team because nobody in myy has a clue of how to solve the problems. It meansry time I look at new problems, I know that I do note answers and I also know that nobody in the com-ows the answers. So you need to go through the processy the PhD training can really teach you in order toese problems. . .Because youhavebeen through thepro-dening a problem and analysing it, next time whenounter a completely different but equally challenging, you are not that scared. You know how to break theinto pieces, to analyse it and come up with some

.

other instances, S&E PhDs might even choose jobs thatthe conventional technical occupations and outside

s such as dedicated managers or consultants/experts.for jobs outside conventional PhD occupations, regard-ther they are in management, in knowledge brokeringnon-research tasks, knowledge in specic subject areasy to be more important than general and transferablejobs are likely to need knowledge that is transferables greater emphasis on the procedural dimensions toiverse clients and situations.

ve discussion suggests some ideas of how knowledgeed in different types of careers. The discussion is inam (2004) typology of use of knowledge in differentnal forms. She argues that the professional bureaucracynal form is based on embrained knowledge, which is

H.-f. Lee et al. / Research Policy 39 (2010) 869881 873

formal and theoretical, while the operating adhocracy (such asprofessional partnerships, software engineering rms and man-agement consultancies) is based on embodied knowledge, whichdraws its capability from the know-how and problem solving skillsembodied inwe suggesting may hacareer typespond to thmore formaside the conto the opeknowledgeness of knothat is morently in difquestion:

(2) To whatoral ed

4. Data an

4.1. Data

We explof graduateof Manchesdesign is todue to theposes, our sthe effectsregions. Thfrom the Unsite universengineeringvides a reascience discwhich repreversity of Mresearch aslent staffs thexcellent).offer attractment wherecareers or strial emploWe also adPhD studenand cohort

The survtory (coverchange in thimize non-reducation ajobs. It wastion samplethat graduaing disciplinaddresses astrategy ismary sampand as all nprobabilityprobability,allows jobs

viduals are independent from each other, while jobs are correlatedwith individuals to whom they belong.

The survey was conducted by post through the Alumni Ofceto preserve condentiality. Our rst wave of survey resulted in

ponswereses.quebtai

ed qual lses).the sses.tionn (Un rehi-sqLawtindicand(X2

0), y2 =1.ed (Tl jobsr of(Armndiciledfrommbe=2.6ed ondentavespearheld

rveyns se adeha lef theum sis isationcan

alys

anais anweig5.2

ch aen, 2ing aistenramstimer v

-basees anproao esthaveindividual experts. Drawing on Lams (2004) typology,that different competences acquired from PhD train-ve different values for S&E PhDs working in differents: the conventional technical occupations, which corre-e professional bureaucracy, may be likely to emphasisel knowledge in subject areas, while employment out-ventional technical occupations, which is more close

rating adhocracy, may be more likely to emphasisethat is general and transferable. As a result, the useful-wledge directly tied to subject areas and of knowledgee general and transferable may be perceived differ-ferent career types. This leads to our second research

t extent are competences developed through S&E doc-ucation relevant for different career types?

d measures

ore the research questions through a complex surveys from a UK research-based university, the Universityter. One of the main considerations of our researchovercome difculties in accessing personal informationUK 1998 Data Protection Act. For exploratory pur-

trategy was to adopt a single university setting to avoidand complexities caused by different universities andere are other benets of studying S&E PhD graduatesiversity of Manchester. Firstly, it is the largest single-ity in the UK and has renowned and well-developedand physical science departments. Practically this pro-sonable size sample from engineering and physicaliplines. Second, it is a member of the UK Russell Group,sents the top 20 leading universities in theUK (theUni-anchester was ranked in the third place in the 2008 UKsessment in terms of the number of full-time equiva-at are judged to be world leading or internationallyIts leading position in research means that it shouldive doctoral training and thus it is an academic environ-students, regardless of whether they aim at academic

implywant to have degrees that are respected by indus-yers, would like to obtain their doctoral degrees from.opt the strategy of selecting home (UK and other EU)ts graduated from specic years to minimise cultureeffects.ey conductedcomprises retrospectiveemploymenthis-ing 710 years employment history to address thee distribution of career types but not too long to min-esponse), types of knowledge acquired from doctoralnd how they are perceived as valuable in differentconducted between April and July 2008. The popula-d for this survey includes all the home PhD studentsted between 1998 and 2001 in science and engineer-es. The sampling frame comprises 512 names with UKnd 84 names with other EU addresses. The samplinga single stage clustered sampling (individuals as pri-ling units [PSUs] and jobs as secondary sampling units),ames in the sampling frame have the same selectionand all jobs from individuals have the same selectionthe sample is self-weighted. Such sampling strategyto be clustered into individuals. It is assumed that indi-

82 resmailsresponsurveywere oreturnindividaddres

Asresponpopulalocatiobetweeusing c1977;1990)dentsgenderp=0.30tion (Xobtainof totanumbewaves1990) itwo-tadentsthe nu(mean

Basresponrentwbias apof jobsour suCochrajobs aran alpation ominimanalysinforminsigni

4.2. An

Theanalyson un-SectionapproaPahkinanalysthe exnon-paance ein highdesignvariablThis apseeks tand toes in 4 weeks just before the response deadline. If e-available, e-mail reminders were sent to encourage

After the deadline, 20 more respondents returned thestionnaires. A total of 91 UK and 11 other EU responsesned. There were 38 UK and 7 other EU undelivereduestionnaires. The overall response rate is 18.51% atevel (19.20% for UK addresses and 15.3% for other EU

ample is self-weighted, bias mainly comes from non-At the individual level, the distributions of surveyaccording to gender, discipline, year of graduation andK or other EU) are known. A characteristic comparisonspondents and non-respondents in these dimensionsuare tests for independence (Armstrong and Overton,on and Parasuraman, 1980; Lambert and Harrington,ates that there is no evidence showing that respon-non-respondents at individual level are different in=0.29; df=1; p=0.590), discipline (X2 =1.073; df=1;ear of graduation (X2 =0.528; df=3; p=0.913) and loca-113; df=1; p=0.291) (Table A1). A total of 282 jobs areable A2). As there is no information about the numberheld by the surveyed PhDs, a comparison of the mean

jobs held by each individual between the concurrentstrong and Overton, 1977; Lambert and Harrington,

ates that there is no signicant difference (t(97)=1.134;p=0.260) between the number of jobs held by respon-the rst wave (mean=2.92; SE=0.130; N=79) and

r of jobs held by respondents from the second wave0; SE=0.245; N=20).the results of the characteristic comparison between

s and non-respondents and the comparison of concur-(between the rst and the secondwaves), non-responses to be insignicant. Therefore, as the average numberby our participants is 2.8, the total number of jobs inpopulation is estimated to be around 1669. Based onample size formula (Cochran, 1977), the obtained 282quate for running regressions for categorical data, withvel of 0.1, 5% margin of error and the standard devi-scale as 0.5 for maximum variability (the estimated

ample size is 234). The nal valid number of jobs for268. There are very few cases of missing data due tonot given. Attrition due to such cases is assumed to be

t.

ing methods

lysis in this paper is based on both individual leveld job level analysis. In Section 5.1, analysis is basedhted descriptive data analysis at individual level. In, when jobs are used as analysing units, the analysingdopted is design-based (Cochran, 1977; Lehtonen and003; Skinner et al., 1989). The design-based survey datapproach takes the complexity of sampling design andce of intra-cluster correlation into account and usesetric variance estimators. Such non-parametric vari-ators are generally unbiased and consistent but resultariances and inefciency (Skinner et al., 1989). Thed approach estimates marginal effects of explanatoryd serves for research aiming at exploratory purpose.ch is different from the model-based approach whichablish precise models, to estimate independent effectspredictive power.

874 H.-f. Lee et al. / Research Policy 39 (2010) 869881

Table 1Career types of S&E PhDs, the University of Manchesters 19982001 home S&E PhDgraduates, 710 years after graduation.

First job (%) Most recent job (%)

Academic/puTechnical poEmploymen

technicalTotal

As the sathat there iadjustmentgraduation-those in thedescriptiveThe analysiysis, by demethod bation for copseudolikelPahkinen,weighted vLemeshow,developedgoodness-osample reualso availaregressionmatrix estimethod.

4.3. Measur

4.3.1. CareeEach res

type of eacalso askedafter PhD trbased on in(details inis restrictedgovernmentions in prto PhDs conin manufacin manufaccareer typein private sconventionemploymenclassicatiocompetencthe increasitechnical otion of ourresearch, 2in employmThe distribacademia/pturing andoccupation

4.3.2. TypesThe disc

tences may

likely that PhD knowledge/skills directly tied to subject areas andPhD knowledge/skills that are more transferable are appreciateddifferently in jobs within and outside the conventional techni-cal occupations. Within the conventional technical occupations, it

likelia oof knto knsts, pby eect at is lesearcen we noboun000)tudeandhniqoulde abthe

singgh clo mememented shhemd thality oexpcraflity.ed odentkillsbs. Tessaluaantowle) spPhDta prd prblem

ob, eiond asTo h

,whid asb (celectal anas ohanknowblic research 42 30sitions in manufacturing 21 12t outside the conventionaloccupations

37 58

100 100

mpling design is self-weighted and although it appearss no signicant non-response bias, a post-straticationis applied to weight the gender-discipline-year oflocation subgroups so that they will be identical topopulation. Analysingmethods comprise design-baseddata analysis and design-based logistic regressions.ng tool is STATA Release 10.1. For survey data anal-fault, the STATA svy command uses the linearisationsed on a rst-order Taylor series linear approxima-variance matrix estimation (Wolter, 1985) and theihood estimation to t the model (Lehtonen and2003). For design-based logistic regression models,ersion of the HosmerLemeshow tests (Hosmer and1980) run through the STATA svylogitgof commandby Archer et al. (2007) are applied to assess thef-t for the models. The jackknife method based onse techniques for covariance matrix estimation is

ble under the STATA svy command. Hence, logisticresults using the linearisation method for covariancemation are compared with results using the jackknife

es

r typespondent was asked to provide information about theh job held after PhD training. Each respondent wasto provide information about tasks in each job heldaining. The variable career type is then constructedformation given by respondents job type and job tasksTable A3). The academic/public research career type

to PhDs conducting research tasks in academia ort/public/non-prot organisations. The technical posi-ivate sector manufacturing career type is restrictedducting research, development, design or productionturing; PhDs who have become dedicated managersturing are not considered as being engaged in this. The academic/public research and technical positionsector manufacturing career types are regarded as theal technical occupations. All other jobs are dened ast outside the conventional technical occupations. Thisn intends to explore thedifference in theuse of S&EPhDes between the conventional technical occupations andngly signicant employment outside the conventionalccupations. According to this measure, the distribu-respondents rst jobs was 42% in academia/public

1% in technical positions in manufacturing and 37%ent outside the conventional technical occupations.

ution of the respondents most recent jobs is 30% inublic research, 12% in technical positions in manufac-

is alsoacadement setregardscienticareerin subjwell, iPhD re

Whthat thhighlyPole (2toral sskillsare tecThey cand thclassifyprocesalthoubility tmanagexperireportin biocrealisecapabifor thesion ofcapabi

Basresponedge/stheir jousefulnmost vimportThe knare: (1edge inand daing an(7) proeach jeducatselecteor not.tenceselectein a jobeen sgenerceivedmore tcialist58% in employment outside the conventional technicals (Table 1).

of knowledge/skills acquired from doctoral educationussion in Section 3.2 implies that different PhD compe-be more or less relevant in different career types. It is

subject aremore thanIn general,topic, genagement sk(Table 2).y that, compared to industrial scientists, scientists inr public research organisations rely on a quite differ-owledge/skills acquired from doctoral education. Withowledge/skills directly tied to subject areas, academicarticularly in science and engineering, often start theirxtending their PhD work, while although knowledgereas normally are important for industrial scientists asss likely that their work will be an extension of theirh.e refer to transferable skills, however, it is argued

tion of transferability remains ambiguous because it isded with the context of application (Craswell, 2007).argued that apart from substantive knowledge, doc-

nts also gain more transferable skills such as technicalcraft knowledge during their study. Technical skillsues that are required to conduct research effectively.be programming skills, the effective use of software

ility to design a research and analyse the results. Wem as application of information technology and dataskills and general analytical skills. Craft knowledge,osely linked to technical skills, emphasises the capa-ake a research project work. This will involve projectnt skills, report writing and presentation skills andation and eldwork. Delamont and Atkinson (2001)ocks and uncertainties encountered by PhD studentsistry, earth science and physical geography when theyt to make an experiment work is far more than thef being able to apply theories and techniques needederiment. We therefore refer to this particular dimen-t knowledge in making things work as problem solving

n these distinctions, in the questionnaire, we askeds to rank the three most valuable types of knowl-that they gained from their PhD and used in each ofhat is, we were interested in measuring the perceivedof a specic type of PhD knowledge/skills in a job. Theble knowledge/skills is given 3 scores; the second mostone is given 2 scores and the third is given 1 score.dge/skills gained from doctoral education to be rankedecialist knowledge in PhD topic; (2) general knowl-subject area; (3) application of information technologyocessing; (4) general analytical skills; (5) report writ-esentation skills; (6) project management skills; and

solving capability. Based on the same measure, forach type of knowledge/skills acquired from doctoralcan be distinguished further by whether it has beenone of the three most valuable PhD skills in the jobighlight the differences, a variable important compe-ch indicateswhethera specicknowledge/skill hasbeenone of the three most variable PhD knowledge/skills

oded as yes if it has been selected and no if noted), was created. At the level of jobs, it appears thatalytical skills and problem solving capability are per-ne of the three most valuable PhD competences inhalf of the survey jobs in all three career types. Spe-ledge in PhD topic and general knowledge in PhD

a are perceived as at least somewhat important inhalf of the survey jobs in academic/public research.perceived usefulness in specialist knowledge in PhDeral knowledge in PhD subject area and project man-ills appears to have greater variation by career types

H.-f. Lee et al. / Research Policy 39 (2010) 869881 875

Table 2Perceived usefulness of PhD competences by career types, the University of Manchesters 19982001 home S&E PhD graduates, design-based descriptive data analysis. Thenumber of observations is 268.

Distribution in score(row percentage)

Selected as among thethree most variable PhDcompetences (%)

Mean score Linearisedstandard error

3 2 1 0

Specialist knowledge in PhD topicAcademic/public research 40 17 6 37 63 1.597 0.199Technical positions in manufacturing 10 10 0 80 20 0.515 0.188Employment outside the conventional technical occupations 6 2 4 88 12 0.254 0.081

General knowledge in PhD subject areaAcademic/public research 18 38 2 42 58 1.310 0.175Technical positions in manufacturing 18 21 4 57 43 1.006 0.255Employment outside the conventional technical occupations 8 8 4 80 20 0.430 0.109

Application of information technology and data processingAcademic/public research 2 7 8 83 17 0.279 0.105Technical positions in manufacturing 6 13 4 77 23 0.482 0.187Employment outside the conventional technical occupations 5 13 11 71 29 0.528 0.111

General anaAcademic/TechnicalEmployme

Report writiAcademic/TechnicalEmployme

Project manAcademic/TechnicalEmployme

Problem solvAcademic/TechnicalEmployme

5. Results

5.1. Domintechnical oc

For the Ugraduates, aular careeramong thosone quarterThe other tpost-doctorlong-termafter graduthis career tare most recontracts. 2moved to em

veracaree

Table 3Distribution of

First job

Academia/pTechnical poEmploymen

technicalN

a Row perceb Column pelytical skillspublic research 6 24 25 45positions in manufacturing 15 25 24 36nt outside the conventional technical occupations 28 34 10 28

ng and presentation skillspublic research 7 11 18 64positions in manufacturing 2 14 27 57nt outside the conventional technical occupations 9 12 20 59

agement skillspublic research 0 5 8 87positions in manufacturing 6 13 15 66nt outside the conventional technical occupations 10 5 29 56

ing capabilitypublic research 13 25 18 44positions in manufacturing 28 33 11 28nt outside the conventional technical occupations 27 38 11 24

tions. Oin thisance of employment outside the conventionalcupations

niversity of Manchesters 19982001 home S&E PhDcademic/public research appears to be the most pop-option for their rst jobs (42%) (Table 3). However,e who were in this career type for their rst jobs, only(27%) secured permanent positions initially (Table 4).

hree quarters were in xed term contracts, mostly inal research positions. Whether this choice is viable forcareer development is uncertain. Indeed, 710 yearsation, only around 67% of those who initially were inype remain in academia/public research. For thosewhocently in this career type, 36% are still in xed term8% of those who initially were in this career type haveployment outside the conventional technical occupa-

those who rhave not being the lengthe alternaseems to liepations (Tacannot be s

Technication) inmanalternativesity of Mancareer type710 yearsoption, 60%positions. Inare due to t

career types, tabulation byrst job andby themost recent job, theUniversity ofManchest

The most recent job

Academia/publicresearch

Technical posiin manufactur

ublic research 26 (67%)a 2 (5%)a

sitions in manufacturing 1 (5%)a 7 (35%)a

t outside the conventionaloccupations

1 (3%)a 2 (6%)a

28 (30%)a 11 (12%)a

ntage.rcentage.55 0.917 0.16564 1.188 0.22172 1.624 0.139

36 0.603 0.13943 0.610 0.15241 0.714 0.121

13 0.186 0.06334 0.587 0.17444 0.690 0.114

56 1.059 0.17572 1.612 0.24476 1.695 0.131

ll, 710 years after graduation, the percentage of PhDsr type has decreased from 42% to 30%. Over one third of

emain in this career type (36%) do so even though theyen able to secure permanent positions, a fact highlight-thening of stages formany academic careers.Moreover,tive for respondents who move out of this career type

in employment outside conventional technical occu-ble 3). Thus, in a long-term perspective, this career typeeen as the dominant one for our survey respondents.l positions (research, development, design or produc-ufacturingwereneither initially nor currently themainof academia/public research. The proportion of Univer-chester 19982001 home S&E PhD graduates in thishas decreased from 21% when rst graduated to 12%after graduation. For those who initially were in thishavemoved topositionsoutside conventional technicala case-by-case investigation, 7 out of 12 of suchmoves

he promotion from researchers to dedicated managers.

ers 19982001homeS&EPhDgraduates, 710years after graduation.

tionsing

Employment outside theconventional technicaloccupations

Total N

11 (28%)a 39 (42%)b

12 (60%)a 20 (21%)b

32 (91%)a 35 (37%)b

55 (58%)a 94 (100%)

876 H.-f. Lee et al. / Research Policy 39 (2010) 869881

Table 4Distribution of career types by employment condition, rst job and the most recent job, the University of Manchesters home 19982001 S&E PhD graduates, 710 yearsafter graduation. Percentage shown is row percentage.

First job The most recent job

Permanent Fixed term Permanent

Academia/p 11 (27%) 10 (36%) 18 (64%)Technical po 20 (91%) 0 (0%) 11 (100%)Employmen 33 (94%) 1 (2%) 50 (98%)N

a The numb er typthe convention are fou

Table 5Employment s uates

al

Academia/pPrivate secto

More themploymenthey rst grof our respostill remainin this carenot only thmany respofor our respaccounts fo

Thereformain careehome S&E PPhDs are acrate R&D laare workingjobs outsideIndeed, althoccupationvey respondthe largestment in thisin this careeappears tocareer typeuation, thisour responemploymennical occupemploymenisations. Taover time wsignicant itime.

To highlcal S&E PhDcase by casfall outsidesector dedicmainly inprare academing or othesector marknical writer

iffere

rall,knolistancecal pn PhDlimi

cal oal edscortionnica

pondal anant iign-btionrentidenh typtic rele ans catrencFixed term

ublic research 29 (73%)sitions in manufacturing 2 (9%)t outside the conventional technical occupationsa 2 (6%)

34 (35%)

er used in the distribution excludes cases of those who are self-employed; the careal technical occupations; there is one self-employed case in terms of rst job and

ectors of S&E PhDs, the University of Manchesters home 19982001 S&E PhD grad

First job

Sector percentage Percentage of unconventionjobs within sector

ublic organisations 47 15r 53 59

an one third of our respondents (37%) initially tookt outside the conventional technical occupations whenaduated. 710 years after graduation, there is little signndents in this career type moving out, as 91% of themin this career type; that is, those who initially were

er type continue to stay (Table 3). This career type ise most stable one, but also the main destination forndentsmoving from the other two career types. Indeed,ondents, 710 years after graduation, this career typer 58% of all employment.e, academia/public research cannot be regarded as ther type for the University of Manchesters 19982001hD graduates. Similarly, very few of our surveyed S&Etually working as industrial scientists in large corpo-boratories in manufacturing, although many of themin industry. These results highlight the signicance ofthe conventional technical occupations for S&E PhDs.ough employment outside the conventional technical

smight not account for the largest proportion of the sur-ents rst employment, it was however only 5% behindemployment sector. Moreover, 94% of initial employ-career typewas permanent and 91% of thosewhowerer type remain in this sector. Furthermore, over time, it

be the main destination for movers from the other twos. Thus, it is not surprising that 710 years after grad-career type accounts for 58% of total employment of

5.2. D

Ovedents,speciaimporttechniedge iIt is oftechnidoctorhigherconvenin techby resgenerimport

Despercepindiffeable toForeaca logisvariabtypes aas refedents and has become the dominant career type. Thet dynamics inside and outside the conventional tech-ations is invisible if the discussion mainly focuses ont dynamics inside and outside academia/public organ-ble 5 shows the stable career patterns of our S&E PhDshen the analysis is based on the latter case and the

ncrease in unconventional jobs within the sectors over

ight the heterogeneity of the unconventional techni-jobs, a detailed investigation looking into our sample

e shows that among individuals most recent jobs thatthe conventional technical occupations, 29% are privateated managers, 34% are technical positions in services,ogramming, softwaredevelopmentor consultancy, 20%ic/public non-research positions, 11% are school teach-r types of lecturing positions, and the rest are privateeting positions, patent attorneys, sales positions, tech-s, business analysts, etc.

regressionranking of avaluable Phsity to rankvaluable Phtypes. Thevalid 268 jdents are frperceptionvariable enis used as cocontrol var(UK or othe

2 An alternacareer types (spresented in t64 (65%) 11 (12%) 79 (88%)

e of those who are self-employed is classied as employment outsider self-employed cases in terms of the most recent job.

, 710 years after graduation.

The most recent job

Sector percentage Percentage of unconventionaljobs within sector

41 2859 80

nt competences mix for different career types

based on scores (Table 2) given by the survey respon-wledge directly tied to subject areas, particularlyknowledge in the PhD topic, is regarded as of greatin academia/public research. It is less important in

ositions in manufacturing, although general knowl-subject area is quite important in this career type.

ted signicance in employment outside conventionalccupations. In general, knowledge/skills acquired fromucation related to general and transferable skills receivees by respondents working in employment outside theal technical occupations, lower scores by respondentsl positions in manufacturing and even lower scoresents in academia/public research positions. However,alytical skills and problem solving capability aren all career types, but to different degrees.ased logistic regressions are applied to test whetherof the relative importance of each specic competencecareer types is signicantly different. In thisway,we aretify specic PhD competences for different career types.eofknowledge/skills acquired fromdoctoral education,gression using important competence as dependentd career type (comprising the three possible careeregories and the career type of academic/public researche category) as explanatory variable is applied.2 The

aims at evaluating how the propensity of S&E PhDsspecic type of knowledge as among the three most

D knowledge/skills in a job, compared to the propen-this type of knowledge as not among the three mostD knowledge/skills in a job, varies in different careeranalysing units are individual jobs, and thus the totalobs are all used in the analysis. Whether the respon-om engineering or science disciplines might affect theirof usefulness of knowledge in jobs and therefore, thegineering (science disciplines as reference category)ntrol variable. Results are shown in Table 6. Additional

iables such as gender, year of graduation and locationr EU) are explored, but they do not change the impres-

tive approach is to compare several means of the original scores byuch as Tukeys test). Using this approach does not change the resultshis paper.

H.-f. Lee et al. / Research Policy 39 (2010) 869881 877

Table 6Relative perceived usefulness of PhD competences by career types. Comparison uses academic/public research as reference category. Odds ratio measures the likelihood ofeach type of knowledge/skills been selected as among the three most valuable types of PhD knowledge/skills in a job rather than not been selected at all by career typesusing design-based logistic regressions based on the University of Manchesters 19982001 home S&E PhD graduates with 710 years job histories.

thod

Specialist knCareer typ

TechnicEmploym

Engineerin

General knoCareer typ

TechnicEmploym

Engineerin

ApplicationCareer typ

TechnicEmploym

Engineerin

General anaCareer typ

TechnicEmploym

Engineerin

Report writiCareer typ

TechnicEmploym

Engineerin

Project manCareer typ

TechnicEmploym

Engineerin

Problem solvCareer typ

TechnicEmploym

Engineerin

N observation* Signicanc

** Signicanc*** Signicanc

sion of theusefulnessThe resultsare very simcondencegoodness-o

Comparepublic reseaingaremorPhDknowleless likely toPhDknowledentsworkoutside theselect geneproblem stheir jobs raspecialistPhD subjecIt appears twith whichcessing anas valuableThe linearised me

Odds ratio

owledge in PhD topiceal positions in manufacturing 0.130***

ent outside the conventional technical occupations 0.071***

g 2.968**

wledge in PhD subject areaeal positions in manufacturing 0.540

ent outside the conventional technical occupations 0.171***

g 1.639

of information technology and data processingeal positions in manufacturing 1.623

ent outside the conventional technical occupations 2.139g 0.442

lytical skillseal positions in manufacturing 1.369

ent outside the conventional technical occupations 2.091*

g 1.498

ng and presentation skillse

al positions in manufacturing 1.369

ent outside the conventional technical occupations 1.258g 0.795

agement skillseal positions in manufacturing 3.501**

ent outside the conventional technical occupations 5.173***

g 0.527

ing capabilityeal positions in manufacturing 2.151

ent outside the conventional technical occupations 2.672**

g 0.613

s: 268.e (two tailed): .1.e (two tailed): .05.e (two tailed): .01.

association between career types and the perceivedof each specic type of PhD knowledge/skills in a job.from the linearised methods and the jackknife methodsilar but the jackknife methods result in wider range of

intervals (CI). All regressions pass the STATA svylogitgoff-t tests.d to the survey respondents working in academia/rch, respondents in technical positions in manufactur-

e likely to select projectmanagement skills as valuabledge/skills in their jobs rather thannot select it at all, butselect specialist knowledge in PhD topic as valuable

dge/skills in their jobs. Compared to the survey respon-ing in academia/public research, respondents employedconventional technical occupations are more likely toral analytical skills, project management skills and

olving capability as valuable PhD knowledge/skills inther than not select them at all, but less likely to selectknowledge in PhD topic and general knowledge int area as valuable PhD knowledge/skills in their jobs.hat there is no signicant difference in the propensitiesapplication of information technology and data pro-

d report writing and presentation skills are perceivedin different career types; this indicates that these two

particular ttion are lesmay be use

A furthefacturing (conventionregressionsbetween thPhD subjeccal positionthe convenin the percearea, thereof all other(Table A4).positions intied to subjetional technthe subjectPhD topic, w

Thus, foPhD graduaatively lie iThe jackknife method

90% CI Odds ratio 90% CI

0.0520.322 0.130*** 0.0490.3440.0310.159 0.071*** 0.0290.1701.3156.699 2.969** 1.2387.116

0.2381.227 0.541 0.2261.2920.0900.326 0.171*** 0.0870.3350.7993.364 1.693 0.7643.520

0.5574.730 1.623 0.4815.4840.9464.833 2.139 0.8945.1140.1741.122 0.442 0.1571.264

0.5573.363 1.369 0.5293.5451.0564.140 2.091* 1.0294.2470.7562.969 1.498 0.7243.0980.5913.171 1.369 0.5653.3190.6562.410 1.258 0.6422.4640.3951.599 0.795 0.3811.659

1.3349.189 3.502* 1.21610.0852.46210.871 5.173*** 2.34111.4320.2261.226 0.527 0.2051.356

0.8615.374 2.151 0.8005.7801.3665.266 2.672** 1.3305.3700.3081.221 0.613 0.2961.269

ypes of knowledge/skills acquired from doctoral educa-s relevant in differentiating the PhD competences thatful in different career types.r comparison between technical positions in manu-as reference category) and employment outside theal technical occupations using design-based logistic(Table 7) shows that it is possible to distinguish

e two career types in terms of general knowledge int area, which is perceived as more valuable for techni-s in manufacturing but is less in employment outsidetional technical occupations. Apart from the differenceived usefulness of general knowledge in PhD subjectis no signicant difference in the perceived usefulnessPhD knowledge/skills between the two career types

This implies that although PhD competence in technicalmanufacturing also relies on knowledge that is directlyct areas, compared to employment outside the conven-ical occupations, it is the general type of knowledge inarea, rather then the specic type of knowledge in thehere the competence resides.

r the University of Manchesters 19982001 home S&Etes, PhD competences in academia/public research rel-n knowledge that is directly tied to subject areas. In

878 H.-f. Lee et al. / Research Policy 39 (2010) 869881

Table 7Relative perceived usefulness of general knowledge in PhD subject areas between technical positions in manufacturing and employment outside the conventional technicaloccupations. Comparison uses technical positions in manufacturing as reference category. Data source and other explanations are the same as in Table 5.

The linearised method The jackknife method

General knoCareer typ

EmploymEngineerin

N observation** Signicanc

contrast, Phtechnical ocPhD compeboth knowlgeneral forspecialist kof the genement outsidterms, genacquired froof career ty

We alsoa specic tytion in a spemploymenthere is no

6. Discussi

This papof Manchesknowledgeperceived abroad resutions havePhDs in a lcharacterisetracts, boththe labourtract researtype. Fromvate sectorEven if theydedicated memploymenacademic oufacturing tretaining itstypes.

Second,black box oemploymenis otherwisedynamics inpointed ounon-researcjobs in manindustry.

Third, thdoctoral edjobs differsdoctoral edent compet

typedemeas bore gnufackillstioncare

nt sty. It

o woempjor ean c

se cothatrial Rmatipropatedositioprivathanearstechnand

e thhe dpattto thncesdomupatingshowrolesng toin t

cal ocre dr sosearepthreerOdds ratio

wledge in PhD subject areaeent outside the conventional technical occupations 0.317**

g 1.683

s: 185.e (two tailed): .05.

D competences in employment outside conventionalcupations lie in themoregeneral and transferable skills.tences in technical positions in manufacturing lie inedge that is directly tied to subject areas but in a morem of knowledge in the PhD subject area (rather thennowledge in PhD topic) and in a less intensive levelral and transferable skills than it is used in employ-e the conventional technical occupations. In absoluteeral analytical skills and problem solving capabilitym doctoral education are valuable for jobs regardless

pes.explored whether the perception of the usefulness ofpe of knowledge/skills acquired from doctoral educa-ecic career type is affected by respondents previoust in different career types. The results indicate that

signicant difference.

on and conclusions

er has examined the career patterns of the Universityters 19982001 home S&E PhD graduates and whichand skills developed through doctoral education ares useful in the jobs they have held. We derive threelts. First, in our case, academic/public research posi-become a secondary career type for the surveyed S&Eong run. The academia/public research career type isd by a high level of employees with xed term con-in terms of rst jobs, and in jobs after 710 years in

market. It shows that there is a large number of con-chers struggling but determined to pursue this careerthe very beginning, most of the PhDs who enter the pri-do not become industrial scientists in manufacturing.were industrial scientists initially, they transferred toanagers gradually. The majority of the PhDs work int outside the conventional technical occupations, i.e.

r public non-research or private sector outside theman-echnical jobs. This career type is not only successful atmembers, but is also the destination of the other career

the study represents our rst attempt to unpack thef S&E PhD jobs. We revealed the dynamics of S&E PhDst in conventional and unconventional occupations thatinvisible in traditional analyses based on employmentside and outside the academia/public sector. We have

careerfor acaject arthe min maable soccupa

ThedifferetronomPhDs tondarythemaAmericour ca2003)industMangelarger(gradunent pin theto less710 y12% inthe UKindicatwith tcareerlook indiffere

Thecal occincreassectorwidertributisectorstechnithem aming onon-rean in-dthis cat the increasing signicance of S&E PhDs working inh academic/public research jobs and the dominance ofagerial activities, business services or consultancy in

e way in which knowledge and skills acquired fromucation are perceived as useful by respondents in theirdepending upon career types. Our study shows thatucation in science and engineering provides differ-ences that are relatively more valuable for different

managers afurther advlic funded2007; Mart1991).

The studusefulnesseducation ition of the90% CI Odds ratio 90% CI

0.1420.703 0.317** 0.1340.7460.7524.017 1.683 0.6594.297

s. These are knowledge directly tied to subject areasia/public research, both knowledge directly tied to sub-ut the more general type rather than in PhD topic andeneral and transferable skills for technical positionsturing, and mainly the more general and transfer-for employment outside the conventional technicals.er patterns that emerge from our survey show quite aory from Martin and Irvines (1981) UK case in radioas-conrms that over time, there has been a shift for S&Erk outside academia and academia has become a sec-loyment sector or S&E PhDswhile industry has become

mployment sector in theUK. This trend is in linewith theases (Stephan, 1996; Stephan et al., 2004). Furthermore,rresponds to the PPARC survey (DTZ Pieda Consulting,indicates the dominance of employment outside the&D laboratories. However, our case is different fromns (2000) French case, where in engineering science, aortion of PhDs secured permanent academic positionsbetween 1984 and 1996, in 1997, 44% secured perma-ns in academia) and most of the French PhDs workingte sector were in research positions (37%), compared20% permanent positions in academia/public researchafter graduation in our UK case (based on Table 4) andical positions inmanufacturing. The similaritybetweenthe US cases and the difference in the French case

at although scholars in many countries are concernedecrease in academic jobs, international differences inerns of S&E PhDs remain and further research maye underlining institutional mechanisms that shape the.inanceof employmentoutside the conventional techni-ions for S&E PhDs present in the private sector and thesignicance of this career type in the academic/public

the diversied career options for S&E PhDs and theirin the economy; it suggests that S&E PhDs are con-knowledge production and absorption across many

he economy. Although jobs outside the conventionalcupations range from sales to school teaching, most ofedicated managerial positions, consultancy, program-ftware developing positions in business services, andch positions in academia or public organisations. Hence,examination to further untangle the heterogeneity oftype, particularly the roles of S&E PhDs as dedicated

nd experts or consultants in business services, willance the notion of manpower training effect of pub-basic science through S&E doctoral education (Lardo,in and Irvine, 1981; Mowery and Sampat, 2005; Pavitt,

y uses subjectivemeasures to investigate the perceivedof different knowledge/skills acquired from doctoraln different career types. The diversity in the percep-usefulness of different knowledge/skills acquired from

H.-f. Lee et al. / Research Policy 39 (2010) 869881 879

doctoral education in our case may be interpreted as the effective-ness of the modern doctoral education in providing an adequateknowledge base for employment across different career types. Thisinterpretation is not only in line with the manpower training effectof public funded basic science, but also reveals how and whattypesof knowledgeproduced inacademia is transferred todifferentsectors through PhDs careermobility. However, asmost of our sur-veyed PhDs are working in employment outside the conventionaltechnical occupations and as PhD competences in this career typemainly lie in more general and transferable skills, this may raise

the question of the uniqueness of the PhD path to acquire suchskills and how exactly a doctoral qualication may enhance onesemployability if the person intends to enter employment outsidethe conventional technical occupations. These questions are openfor debate and further research.

Appendix A.

See Tables A1A4.

Table A1Assessing non-response bias using the characteristic comparison method.

Respondent Non-respondent Survey population at individual level

GenderMale 77 (75%) 385 (78%) 463 (78%)Female 25 (25%) 109 (22%) 134 (22%)Total 102 (100%) 494 (100%) 596 (100%)X2 =0.29; df=1; p=0.590

DisciplineEngineering 26 (25%) 103 (21%) 129 (22%)Science 76 (75%) 391 (79%) 467 (78%)Total 102 (100%) 494 (100%) 596 (100%)X2 =1.073; df=1; p=0.300

Year of graduation1998 22 (22%) 128 (21%)1999 22 (22%) 147 (25%)2000 30 (30%) 182 (31%)2001 26 (26%) 139 (23%)Total 100 (100%) 596 (100%)X2 =0.528; df=3; p=0.913

LocationUK 91 (89%) 421 (85%) 512 (86%)Other EU 11 (11%) 73 (15%) 84 (14%)Total 102 (100%) 494 (100%) 596 (100%)X2 =1.113; df=1; p=0.291

Table A2Denition of a job.

Include any job (including self-employment), full-time or part-time, which you did for at least 6 months (or which you expect to last for at least 6 months).Dont count jobs or work experience that you did while registered as a full-time PhD student.If you changed the kind of work you did, rank or job title while working for the same employer, count it as a change of job.If you have worked in a Government Department, school or hospital, count any move from one Government Department, school or hospital to another, as a change ofjob.Contract researchers in academic institutions or other employment on short-term contracts: if your contract was renewed count this as an extension of the same job.If you had a period of temping, free-lancing, consultancy or self-employed contract work, count the whole period as one job.If you went on maternity leave or sick leave and went back to the same employer for the same kind of work, rank and job title, count the whole period as one job.

Table A3The construction of variable career type.

Each respondent was asked to provide informationabout employment code of each job after PhD trainingamong the following options

Each respondent was also askedto provide information abouttasks in each job after PhDtraining among

Combination: employmentcode+ job tasks

Career type

(1) University faculty position (a) Managerial (1) Academic/public research(2) University research position (b) Research/development (2)(3) Other university post (c) Other, specify (7)(4) Private sector company service (5) + (b) Technical positions in manufacturing(5) Private sector company manufacturing All other combinations Employment outside the conventional

technical occupations(6) Private sector company other(7) Researchorganisation(8) Other poorganisation(9) Running(10) Freelan(11) Other typost in a government/public/voluntary

sition in a government/public/voluntary

own companyce workerpe of employment

880 H.-f. Lee et al. / Research Policy 39 (2010) 869881

Table A4Relative perceived usefulness of general knowledge in PhD subject areas between technical positions in manufacturing and employment outside the conventional technicaloccupations. Comparison uses technical positions in manufacturing as reference category. Data source and other explanations are the same as in Table 5.

The linearised method The jackknife method

Specialist knCareer typ

EmploymEngineerin

General knoCareer typ

EmploymEngineerin

ApplicationCareer typ

EmploymEngineerin

General anaCareer typ

EmploymEngineerin

Report writiCareer typ

EmploymEngineerin

Project manCareer typ

EmploymEngineerin

Problem solvCareer typ

EmploymEngineerin

N observation* Signicanc

** Signicanc

References

Abernathy,W.Managem

Almeida, P., Koneers in re

Archer, K.J., LeregressionComputat

Armstrong, J.SJournal of

Biddle, J., Robeto manage

Braddock, D.J.Labor Revi

Burgess, R.G.,in the UK.Lifelong Le

Cervantes, M.,Backgroun

Cochran, W.G.Cohen, W.M.,

The EconoCommittee ofCraswell, G., 2

Higher EduCreplet, F., Dup

experts inDany, F., Mang

and job inscation Poli

David, P.A., ForInternatio

DeFillippi, R.J.perspectiv

Delamont, S., ASocial Stud

Department foLondon (ation natioOdds ratio

owledge in PhD topiceent outside the conventional technical occupations 0.543

g 2.827*

wledge in PhD subject areaeent outside the conventional technical occupations 0.317**

g 1.683

of information technology and data processingeent outside the conventional technical occupations 1.320

g 0.481

lytical skillseent outside the conventional technical occupations 1.567

g 2.435*

ng and presentation skillseent outside the conventional technical occupations 0.910

g 0.642

agement skillseent outside the conventional technical occupations 1.479

g 0.562

ing capabilityeent outside the conventional technical occupations 1.243

g 0.621

s: 185.e (two tailed): .1.e (two tailed): .05.J., Rosenbloom, R.S., 1969. Parallel strategies in development projects.ent Science 15 (10), 486505.gut, B., 1999. The localization of knowledge and the mobility of engi-gional networks. Management Science 45 (7), 905917.meshow, S., Hosmer, D.W., 2007. Goodness-of-t tests for logisticmodels when data are collected using a complex sampling design.

ional Statistics & Data Analysis 51, 44504464.., Overton, T.S., 1977. Estimating nonresponse bias in mail surveys.Marketing Research 14 (3), 396402.rts, K., 1994. Private sector scientists and engineers and the transitionment. The Journal of Human Resources 29 (1), 82107., 1992. Scientic and technical employment, 19902005. Monthlyew 115 (2), 2841.1997. The changing context of postgraduate education and trainingIn: Burgess, R.G. (Ed.), Beyond the First Degree: Graduate Educationarning and Careers. SRHE/Open University Press, Buckingham.2001. Mobilising human resource for innovation. In: OECD Growthd Papers No. 2. OECD, Paris., 1977. Sampling Techniques, 3rd ed. Wiley, New York.Levinthal, D.A., 1989. Innovation and learning: the two faces of R&D.mic Journal 99 (397), 569596.Vice-Chancellors and Principles, 1988. The British PhD. CVCP, London.007. Deconstructing the skills training debate in doctoral education.cation Research & Development 26 (4), 377391.ouet, O., Kern, F., Mehmanpazir, B., Munier, F., 2001. Consultants andmanagement consulting rms. Research Policy 30, 15171535.ematin, V., 2004. Beyond the dualism between lifelong employmentecurity: some new career promises for young scientists. Higher Edu-cy 17, 201219.ay,D., 2002.An introduction to theeconomyof theknowledge society.nal Social Science Journal 54 (171), 923., Arthur, M.B., 1994. The boundaryless career: a competency-basede. Journal of Organizational Behavior 15 (4), 307324.tkinson, P., 2001. Doctoral uncertainty: mastering craft knowledge.ies of Science 31 (1), 87107.r Innovation, Universities & Skills, 2008. Innovation Nation. HMSO,ccessed on 20.10.09) http://www.dius.gov.uk/innovation/ innova-n.

Departmentothe Challe

Dosi, G., Freemducing somIndustrial

DTZ Pieda Con(accessedCohort.pd

Enders, J., 2002need of in493517.

Enders, J., 200the transfo

Finn,M.G., Bakor surplus

Foray, D., LundknowledgKnowledg

Fox,M.F., Steprealities b

Freeman,C., 19of innovata Theory o

Funtowicz, S.OSchomberDecision-M

Gibbons, M., LProductionrary Socie

Giret, J., RecotigraduatesEuropean

Godin, B., 199Policy 25,

Hargadon, A.,developm

Harris Report(accessed

Hosmer, D.W.regression90% CI Odds ratio 90% CI

0.1991.477 0.543 0.1821.6211.0347.728 2.827 0.9368.537

0.1420.703 0.317** 0.1340.7460.7524.017 1.683 0.6594.297

0.5073.438 1.320 0.4453.9130.1651.401 0.481 0.1431.611

0.7113.456 1.567 0.6793.6191.0195.817 2.435 0.9446.278

0.4291.929 0.910 0.4122.0090.2771.489 0.642 0.2621.574

0.6493.371 1.479 0.6003.6480.2211.434 0.562 0.1961.611

0.5492.817 1.243 0.5103.0280.2661.448 0.621 0.2521.532f EducationandScienceWhitePaper, 1987.Higher Education:Meetingnge. HMSO, London.an, C., Fabiani, S., 1994. The process of economic development: intro-e stylised facts and theories on technologies, rms and institutions.

and Corporate Change 3 (1), 145.sulting, 2003. A Study of the Career Paths of PPARC PhD Studentson 17.05.09) http://www.so.stfc.ac.uk/publications/pdf/ PiedaNew-f.. Servingmanymasters: thePhDon the labourmarket, theeverlasting

equality, and the premature death of Humboldt. Higher Education 44,

5. Broader crossing: research training, knowledge dissemination andrmation of academic work. Higher Education 49, 119133.er, J.G., 1993. Further jobs innatural scienceandengineering: shortage? Monthly Labor Review 116 (2), 5461.vall, B., 1996. The knowledge-based economy: from the economics ofe to the learning economy. In: OECD Employment and Growth in thee-based Economy. OECD, Paris.han, P.E., 2001. Careers of young scientists: preferences, prospects andy gender and eld. Social Studies of Science 31 (1), 109122.92. Formal scienticand technical institutions in thenational systemsion. In: Lundvall, B.. (Ed.), National Systems of Innovation: Towardsf Innovation and Interactive Learning. Pinter, London.., Ravetz, J.R., 1993. The emergence of post-normal science. In: van

g, R. (Ed.), Science, Politics, and Mobility. Scientic Uncertainty andaking. Kluwer, Dordrecht.

imoges, C., Nowotny, H., Schwartman, S., Scott, P., Trow, M., 1994. Theof Knowledge: The Dynamics of Science and Research in Contempo-

ties. Sage, London.llet, I., 2004. The impact of CIFRE programme into early careers of PhDin France. In: Paper Presented to the 16th Annual Conference of theAssociation of Labour Economics, Lisbon.6. Research and the practice of publication in industries. Research587606.Sutton, R.I., 1997. Technology brokering and innovation in a productent rm. Administrative Science Quarterly 42 (4), 716749., 1996. Review of Postgraduate Education. HEFEC, CVCP and SCOP08.03.09) http://www.hefce.ac.uk/pubs/hefce/1996/m14 96.htm., Lemeshow, S., 1980. Goodness-of-t tests for the multiple logisticmodel. Communications in Statistics A9 (10), 10431069.

H.-f. Lee et al. / Research Policy 39 (2010) 869881 881

Jones, O., 1992. Postgraduate scientists and R&D: the role of reputation in organisa-tional choice. R&D Management 22 (4), 349358.

Kogut, B., Zander, U., 1992. Knowledge of the rm, combination capability, and thereplication of technology. Organization Science 3, 383397.

Kogut, B., 2008. Knowledge, Options, and Institutions. Oxford University Press,New York.

Kornhauser, W., 1962. Scientists in Industry: Conict and Accommodation. Univer-sity of California Press, Berkeley and Los Angeles.

Lam, A., 2004. Societal institutions, learning organisations and innovation in theknowledge economy. Research on Technological Innovation and ManagementPolicy 8, 4367.

Lam, A., 2007. Knowledge networks and careers: academic scientists inindustryuniversity links. Journal of Management Studies 44 (6), 9931015.

Lambert, D.M., Harrington, T.C., 1990. Measuring nonresponse bias in costumer ser-vice mail surveys. Journal of Business Logistics 11 (2), 525.

Lardo, P., 2007. Revisiting the third mission of universities: toward a renewedcategorization of university activities? Higher Education Policy 20, 441456.

Lavoie,M., Finnie, R., 1998. The occupational dynamics of recent Canadian engineer-ing graduates inside and outside the bounds of technology. Research Policy 27,143158.

Lavoie, M., Roy, R., Therrien, P., 2003. A growing trend toward knowledge work inCanada. Research Policy 32, 827844.

Lawton, L., Parasuraman, A., 1980. The impact of the marketing concept on newproduct planning. The Journal of Marketing 44 (1), 1925.

Lehtonen, R., Pahkinen, E.J., 2003. PracticalMethods for Design and Analysis of Com-plex Surveys, 2nd ed. John Wiley & Sons, Chichester.

Leonard, D., 2000. Transforming doctoral studies: competencies and artistry. HigherEducation in Europe XXV (2), 181192.

Madsen, T., Mosakowski, E., Zaheer, S., 2003. Knowledge retention and personnelmobility: the nondistructive effects of inows of experience. Organization Sci-ence 14 (2), 173191.

Mangematin, V., 2000. PhD job market: professional trajectories and incentivesduring the PhD. Research Policy 29, 741756.

Mangematin, V., 2001. Individual careers and collective research: is there aparadox?International Journal of Technology Management 22 (7/8), 670675.

Martin, B., Irvine, J., 1981. Spin-off from basic science: the case of radioastronomy.Physics in Technology 12 (5), 204212.

Martinelli, D., 1999. Labour market performance of French PhDs: a statistical anal-ysis. In: OECD Mobilising Human Resources for Innovation. OECD, Paris.

McMillam, G.S., Deeds, D.L., 1998. The role of reputation in the recruitment of sci-entists. R&D Management 28 (4), 349358.

Merton, R.K., 1973. The Sociology of Science. University of Chicago Press, Chicago.Merton, R.K., 1

symbolismMiozzo,M.,Gr

nizationalMowery, D.C.,

In: FagerbInnovation

Murray, F., 2002. Innovation as co-evolution of scientic and technologi-cal networks: exploring tissue engineering. Research Policy 31, 13891403.

Murray, F., 2004. The role of academic inventors in entrepreneurial rms: sharingthe laboratory life. Research Policy 33, 643659.

Nowotny, H., Scott, P., Gibbons, M., 2001. Re-Thinking Science: Knowledge and thePublic in an Age of Uncertainty. Polity Press, Cambridge.

OECD, 1991. Science and Technology Policy: Review and Outlook. OECD,Paris.

OECD, 1992. Technology-Economy Programme. Technology and The Economy: TheKey Relationship. OECD, Paris.

OECD, 2000. A New Economy? The Changing Role of Innovation and InformationTechnology in Growth. OECD, Paris.

Park, C., 2005. New variant PhD: the changing nature of the doctorate in the UK.Journal of Higher Education Policy and Management 27 (2), 189207.

Pavitt, K., 1991. What makes basic research economically useful? Research Policy20, 109119.

Pelz, D.C., Andrews, F.M., 1966. Scientists in Organizations. John Wiley and Sons,New York.

Pole, C., 2000. Technicians and scholars in pursuit of the PhD: some reections ondoctoral study. Research Papers in Education 15 (1), 95111.

Robin, S., Cahuzac, E., 2003. Knocking on academias doors: an inquiry into the earlycareers of doctors in life science. Labour 17 (1), 123.

Rosenberg, N., 1990. Why do rms do basic research (with their own money)?Research Policy 19 (2), 165174.

Rosenkopf, L., Almeida, P., 2003. Overcoming local search through alliances andmobility. Management Science 49 (6), 751766.

Skinner, C.J., Holt, D., Smith, T.M.F., 1989. Analysis of Complex Surveys. Wiley, NewYork.

Slaughter, S., Leslie, L.L., 1997. Academic Capitalism. The Johns Hopkins UniversityPress, Baltimore.

Stephan, P.A., 1996. The economics of science. Journal of Economic Literature 34,11991235.

Stephan, P.E., Gurmu, S., Sumell, A.J., Black, G., 2007. Whos patenting in the univer-sity? Evidence from the survey of doctorate recipients. Economics of Innovationand New Technology 16, 7199.

Stephan, P.E., Sumell, A.J., Black, G.C., Adams, J.D., 2004. Doctoral education and eco-nomic development: the owof newPh.D.s to industry. EconomicDevelopmentQuarterly 18, 151167.

The Dearing Report, 1997. Higher Education in the Learning Society: The NationalCommittee of Inquiry into Higher Education. HMSO, London.

Wolter, K.M., 1985. Introduction to Variance Estimation. Springer, New York.., 1996s. Sci

L.G., Dity sciagemL.G., Dce. Jou988. The Matthew effect in science, II cumulative advantage and theof intellectual property. ISIS 79, 606623.

imshaw,D. (Eds.), 2006.Knowledge IntensiveBusiness Services:Orga-Forms and National Institutions. Edward Elgar, Cheltenham.Sampat, B.N., 2005. Universities in the national innovation systems.erg, J., Mowery, D.C., Nelson, R.R. (Eds.), The Oxford Handbook of. Oxford University Press, New York.

Ziman, Jnorm

Zucker,versMan

Zucker,mer. Postacademic science: constructing knowledge with networks andence Studies 9 (1), 6780.arby, M.R., Armstrong, J.S., 2002a. Commercializing knowledge: uni-ence, knowledge capture, and rm performance in biotechnology.ent Science 48 (10), 138153.arby, M.R., Torero, M., 2002b. Labor mobility from academe to com-rnal of Labor Economics 20 (3), 629660.

Career patterns and competences of PhDs in science and engineering in the knowledge economy: The case of graduates from a ...IntroductionCareers of S&E PhDsKnowledge and skills acquired from doctoral educationPurposes of doctoral educationCharacteristics of the use of knowledge by different types of career

Data and measuresDataAnalysing methodsMeasuresCareer typesTypes of knowledge/skills acquired from doctoral education

ResultsDominance of employment outside the conventional technical occupationsDifferent competences mix for different career types

Discussion and conclusionsReferencesReferences