Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
1
SupervisingthePhD:identifyingcommonmismatchesinexpectationsbetween
candidateandsupervisortoimproveresearchtrainingoutcomes
AdamP.A.Cardilini1*†,AliceRisley2*andMarkF.Richardson3,4
*Co-firstauthors
1DeakinUniversity,Geelong,Australia,SchoolofLifeandEnvironmentalScience,VIC,Australia
2InstituteofEvolutionaryEcologyandConservationGenomics,UniversityofUlm,Albert-Einstein-
Allee1,89081,Ulm,Germany
3DeakinUniversity,Geelong,Australia,CentreforIntegrativeEcology,SchoolofLifeand
EnvironmentalSciences,WaurnPondsCampus,VIC,Australia
4DeakinUniversity,Geelong,Australia,DeakinGenomicsCentre,SchoolofLifeandEnvironmental
Sciences,BurwoodandWaurnPondsCampuses,VIC,Australia
†CorrespondingAuthor:[email protected]
AuthorORCID:APAC-0000-0002-1032-3466;AR-0000-0002-0731-2934;MFR-0000-0002-1650-
0064
Keywords:Researchtraining,PhDsupervision,PhDcandidate,PhDmentalwell-being,research
expectations
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
2
Abstract
TherelationshipbetweenaPhDcandidateandtheirsupervisorisinfluentialinnotonlysuccessful
candidatecompletion,butmaintainingcandidatesatisfactionandmentalhealth.Wequantified
potentialmismatchesbetweenthePhDcandidatesandsupervisorsexpectationsasapotential
mechanismthatfacilitatespoorcandidateexperiencesandresearchtrainingoutcomes.114PhD
candidatesand52supervisorsrankedtheimportanceofstudentattributesandoutcomesatthe
beginningandendofcandidature.Inrelationtospecificattributes,supervisorsindicatedthelevelof
guidancetheyexpectedtogivethecandidateandcandidatesindicatedthelevelofguidancethey
expectedtoreceive.Candidatesalsoreportonwhetherdifferentaspectsofcandidatureinfluenced
theirmentalwell-being.Weidentifieddifferencesbetweencandidatesandsupervisorsperceived
supervisorteachingresponsibilityandinfluencesonmentalwell-being.Ourresultsindicatethatthe
majorityofcandidatesweresatisfiedoverallwiththeirsupervision,andfindalignmentofmany
expectationsbetweenbothparties.Yet,wefindthatcandidateshavemuchhigherexpectationsof
achievingquantitativeoutcomesthansupervisors.Supervisorsbelievedtheygivemoreguidanceto
candidatesthancandidatesperceivetheyreceived,andsupervisorsoftenonlyprovidedguidance
whenthecandidateexplicitlyasked.Personalexpectationsandresearchprogresssignificantlyand
negativelyinfluencedover50%ofcandidate’smentalwell-being.Ourresultshighlightthe
importanceofcandidatesandsupervisorsexplicitlycommunicatingtheresponsibilitiesand
expectationsoftherolestheyplayinhelpingcandidatesdevelopresearchskills.Weprovidefour
suggestionstosupervisorsthatmaybeparticularlyeffectiveatincreasingcommunication,avoiding
potentialconflictandpromotingcandidatesuccessandwellbeing.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
3
Introduction
AnycandidatewilltellyoudoingaPhDishard.However,guidingaPhDcandidatethroughthe
bumpyprocesscanbejustastrickyforthesupervisor.Developingagoodrelationshipbetweenthe
candidateandsupervisorisoneofthemostimportantcomponentsofasuccessfulPhD.Research
hasconsistentlyshownthattherelationshipbetweenthecandidateandtheirsupervisorisoneof
themostimportantpredictorsofcandidatedissatisfaction(Ives&Rowley,2005;KovachClarketal.,
2009;Tompkinsetal.,2016),PhDdiscontinuation(Bair&Haworth,2004;Buckley&Hooley1,1988;
Golde,2000;Kiley,2011;Lovitts&Nelson,2000),anddepression(Pelusoetal.,2011).InAustralia,
20%ofcandidatesreportnotbeingsatisfiedwiththeirsupervision(McGaghetal.,2016),and20-
35%areestimatedtodropoutoftheirprogram(Jiranek,2010;McGaghetal.,2016).Similarly,
otherstudieshavefoundover30%ofPhDcandidatesareatriskofhavingordevelopingcommon
psychiatricdisorders(Levecqueetal.,2017;Pelusoetal.,2011).Thisequatestolarge-scalenegative
mentalhealthandcareerconsequencesacrossstudents,aswellasasignificantlossofresearch
output(Larivière,2011).Importantly,thesenegativeoutcomescannotbesolelyattributedtothe
candidate.Supervisorengagement,amongstotherfactorssuchasfacultysupport(Golde,2000),
havebeenincreasinglylinkedtocandidateresearchproductivity(Guetal.,2011),student
completionrates(Buckley&Hooley1,1988;Kiley,2011),andstudentmentalhealth(Levecqueet
al.,2017;Pelusoetal.,2011).Assuch,therehasbeenanincreasingemphasisontheimportanceof
supervisortraininginimprovingoutcomesforcandidates(Delamontetal.,2004;Halse,2011;Halse
&Bansel,2012),aswellascognitive-behaviouralcoachingforcandidates(Kearns,Gardiner,etal.,
2008).However,althoughsupervisortrainingemphasizestheimportanceofclearcommunicationof
expectationsbetweenbothparties(Delamontetal.,2004;Moxhametal.,2013),thereiscurrently
littlequantitativeunderstandingofhowcommonexpectationsareeitheralignedormismatched
withinAustralianinstitutionsandhowthesecanbeaddressed.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
4
Mismatchedexpectationsbetweencandidatesandsupervisorshavebeenshowntohaveanegative
effectoncandidatecompletionratesandtimeliness(Holbrooketal.,2014;McCormack,2004).Clear
communicationofexpectationsbetweencandidatesandsupervisorsisconsideredtobeparamount
tofosteringasuccessfulworkingrelationship(Delamontetal.,2004;Moxhametal.,2013).
However,expectationsarechallengingtoaddressandquantifybecausetheycoverahugearrayof
responsibilitiesandoutcomesthatarecriticaltoPhDcompletion(Mowbray&Halse,2010),and
thesebynecessitychangeovertime(e.g.supervisorexpectationsofthestudentwilldifferbetween
thebeginningofcandidatureandatthetimeofcompletion).Thereisalsoanincreasingdistinction
betweenexpectationsrevolvingaroundquantifiablestudentoutcomes(e.g.numberofpapers
publishedorgrantsattained)andqualitativeoutcomes(e.g.criticalthinkingortechnicalskills
specifictothefield),andthesemaydifferbetweencandidateandsupervisoriftheinstitutiondoes
notrequireaPhDbypublication(Lindénetal.,2013;Vilkinas,2008).Moreover,theextenttowhich
thesupervisorisexpectedtoguidethecandidateindevelopingthenecessarycriticalskillsto
completeaPhDandpreparethecandidateforacareerintheirchosenfieldmaybeanothersource
ofconflict.Critically,supervisorsareincreasinglytimedeficientandjugglingmanyresponsibilities
andobligations,givingthemlittlefreedomtodedicatetimetothechallengesthatindividual
candidatesmayface.Althoughthedynamicsofsuchrelationshipsareuniquetothecandidateand
supervisorinquestion,trendsincandidateandsupervisorexpectationsmayexistthatallow
supervisorstofocuseffortsoncertainareasinordertopromotemoreeffectivecommunication
betweenthetwo.
Inthisstudy,weaimedtoidentifytheimportanceofcommonexpectationssurroundingthe
candidate–supervisorrelationship.Byidentifyingexpectationsmisalignedincandidatesand
supervisors,wecanunderstandwhereeffortsshouldbedirectedtoavoidconflictandpromote
candidatesuccess.114PhDcandidatesand52supervisorsansweredaquestionnairethatasked
themtorankcommonexpectationsandoutcomesofstudentsbyimportanceforboththebeginning
andendofcandidature.Inaddition,wealsoassessedthelevelofguidanceexpectedtobegivento
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
5
thestudentbythesupervisorforeachoutcome,inordertoidentifydiscrepanciesbetweenhow
studentsandsupervisorsperceivesupervisorresponsibility.Wethencomparedanswersbetween
studentsandsupervisorsforbothanalysestoidentifywherethebiggestdiscrepanciesin
expectationslay.Finally,weassessedwhethersupervisorguidancewasperceivedtobelinkedto
studentwell-being,inordertoassessanynegativeimpactsofsupervisiononstudentmentalhealth.
Wesummarisetheresultsbyprovidingfoursuggestionstosupervisorsthatmaybeparticularly
effectiveatincreasingcommunication,avoidingconflictandpromotingcandidatesuccess.
Methods
Thesurvey
AsurveywasdevelopedtodeterminethedifferenceinPhDcandidateandsupervisorexpectations
ofthecandidates’attributesandsupervisorguidance(supplementaryfile).Thefirstquestionhad
participantsself-identifyasaPhDsupervisor,PhDcandidate,recentlygraduatedPhD(<2yr)ora
discontinuedcandidate.Candidatesandsupervisorsweregivenadifferentsurveysettocomplete.
Thequestionsaskedineachsurveysetwereequivalent,butthequestionswerewrittenforthe
perspectiveofeachgroup(seesupplementaryfile).Forexample,bothcandidatesandsupervisors
receivedthefollowingequivalentquestionrespectively,‘PleaserankthequalityofthePhD
supervisionyoureceived’or‘Inyouropinion,rankthequalityofthePhDsupervisionyouprovide’.
Eachsurveysetincludedfivedemographicquestionsandsixquestionsregardingcandidate
attributes,supervisorguidanceandthequalityofsupervision.Thecandidatesurveysetincluded
threeprivatequestions,tworelatingtocandidatementalwell-beingduringcandidatureandoneon
thelikelihoodofpursuingacareerinresearchacademia.
Twoquestionswereaskedtodeterminethecandidateattributesthatweremostimportantto
candidatesandsupervisors.Participantswereaskedtoselectandrankthetopfiveattributesofa
candidatestartingaPhDandthetopfiveattributesoroutcomesofacandidateatthetimeofthesis
submission(seesupplementaryfileforattributeandoutcomeoptionssets).Twoquestionswere
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
6
askedtodeterminetheattributesthatasupervisorhasthemostresponsibilityforhelpinga
candidatedevelopduringaPhD.Thefirstsupervisorresponsibilitiesquestionrequiredparticipants
toselectandrankthetopthreeattributesthatasupervisorismostresponsibleforhelpinga
candidatedevelopduringtheirPhD(seesupplementarymaterialforattributeoptions).Thesecond
questionaskedcandidatesandsupervisorstoindicatethelevelofguidancetheyrespectively
received,orprovideduringcandidatureinrelationtoeachattribute.Thelevelofguidancewas
indicatedonafourpointscaleincluding,‘none’,‘onlywhenasked’,‘whenseenasneeded’and‘at
everyopportunity’.Todeterminethemismatchbetweencandidateandsupervisorexperienceof
PhDsupervision,candidatesandsupervisorswereaskedtoindicatedthequalityofsupervisionthey
respectivelyreceivedorprovidedduringcandidature.
PhDcandidateswereaskedtwoquestionsregardingtheirmentalwell-beingduringcandidature.The
firstquestionaskedcandidatestoindicatewhetherexperiencesduringtheircandidaturehad
negativelyaffectedtheirmentalwell-being.Responseswereindicatedonafive-pointscalefrom
stronglydisagreetostronglyagree.Thesecondquestionaskedcandidatestoindicatehow
significantlyalongafive-pointscalefivedifferentaspectsofcandidaturenegativelyinfluencedtheir
mentalwell-being.Thefiveaspectsofcandidaturethatwereaskedaboutwere,‘supervisor
relationship’,‘researchenvironment’,‘researchprogress’,‘personalexpectations’,and‘supervisor
expectations’.Theoptionsforthefive-pointscaleincluded,notatallsignificantlytovery
significantly.Finally,candidatesandsupervisorswereaskedtoratethequalityofthesupervision
theyhadreceived/providedalonga5pointscaleincludingoptionsfromverylowtoveryhigh.
Recruitment
Anemailinvitingpeopletoparticipateinthesurvey,whichincludedalinktothesurvey,wassentto
PhDcandidatesandsupervisorsinScienceandHealthfacultiesatDeakinUniversity,anAustralian
highereducationinstitution.SchoolsthatwereexpectedtohavealargenumberofPhDcandidates
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
7
completingresearchthesisweretargeted,includinglifesciences,medicineandpsychology.The
surveywasopenedforcompletionfrom5thAugust2016to14thNovember2016.
Analysis
Questionresponseswhereparticipantsrankedasubsetofattributescreatedapartiallyranked
dataset.Eachattributethatwasrankedbyaparticipantwasgivenanumericalvalueequivalenttoits
rank,e.g.rank1=1,rank2=2,etc.Ifparticipantsrankedthesameattributetwiceinasingle
questionthesecondinstanceoftheattributewasreplacedwith‘NA’.Tocalculatetherankofall
attributes,unrankedattributevaluesmustbefilledinthedatamatrix.Wepopulatedunfilled
attributevalueswiththeaverageofthemeanunrankedvaluepossibilities,e.g.unranked
possibilities6-16haveameanof11.Weusedthe‘rank’functioninRstatisticalpackage‘base’
package(RCoreTeam,2017)torankeachattribute.Anytiedattributesweresettothemaximum
numberofthosetiedattributes.Wecalculatedseparaterankvaluesforcandidateandsupervisor
subsetsandscaledvaluestoallowforcomparison(‘scale()’,RCoreTeam,2017).
Wetestedwhethertherewasasignificantdifferenceintheaverageranksetsbetweencandidates
andsupervisorsusingaKendallrankcorrelation,tau,fromthe‘Kendall’packageinR(McLeod,
2011).Correlationp-values≤0.05wereconsideredsignificant.Afterscaling,rankvaluescouldbe
negativeorpositive.Tomakeplotcomparisonseasiertointerprettheminimumscaledrankvalue
wasaddedtoeachrankscore,thismadetheminimumrankvalueequal‘0’andallothervalues
positive.Thehighestscaledrankvaluecorrespondstothetoprankingoption.
WeusedGeneralisedLinearModels(GLM)todeterminedifferencesbetweenthelevelsofguidance
candidatesandsupervisorsindicatedthattheyreceived/provided.Wereducedthefourguidance
optionresponsesintotwovaluesforanalysis;‘none’and‘onlywhenasked’responsesweregrouped
aspassiveguidanceandgivenavalueof0,while‘whenseenasneeded’and‘ateveryopportunity’
weregroupedasactiveguidanceandgivenavalueof1.Thiscreatedabinarydatasetforanalysis.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
8
TheGLMincludedthebinaryguidanceresponseastheresponsevariablewithcandidate/supervisor
statusasthepredictorvariable.Themodelspecifiedabinomialdistribution.
WeusedKendallnon-parametrictests(‘cor.test’,RCoreTeam,2017)todeterminecorrelations
betweencandidatesreportedsupervisorquality,withexperiencesandimpactsofnegativewell-
beingandcareeraspirations.Tauandp-valuesarereported.WeusedWilcoxnon-parametrictests
(‘wilcox.test’,RCoreTeam,2017)toanalysethegenderdifferencesinreportedsupervisionquality,
well-beingmeasuresandcareeraspirations.
p-valueswereFDRcorrectedforeachsetoftests(Benjamini&Hochberg,1995)andFDRcorrected
p-values<0.05wereconsideredsignificant.
Results
166academicsansweredthesurveyonPhDexpectations.Ofthese,55%werecurrentPhD
candidates,31%weresupervisors,and13%wererecentgraduates,whileonepersonwasa
discontinuedcandidate(Table1).Wefoundasignificantdifferencebetweenhowcandidatesand
supervisorsrankedtheimportanceofattributesinallthreeofthefollowingtests;1)therankof
attributescandidatesareexpectedtohaveatthebeginning,and2)theendofcandidature,and3)
therankofattributessupervisorsareexpectedtoprovideguidanceon(Table2).
Beginningofcandidature
Expectationsofbothcandidatesandsupervisorswerefairlywellalignedatthebeginningof
candidature,withbothgroupsplacinghighimportanceoncandidatemotivation,enthusiasm,and
writtencommunication(Fig.1a).However,candidatesplacedmuchhigherimportanceongood
academicgradesthansupervisors,whilstsupervisorsplacedmuchhigherimportanceonthe
candidate’sabilitytothinkcriticallythandidthecandidatesthemselves.Previouspublications,
industryexperience,self-confidence,andgoodverbalcommunicationwereallmoreimportantfor
candidatesthantheyweretosupervisorsatthebeginningofcandidature.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
9
Endofcandidature
Overall,candidateshadmuchhigherexpectationsofachievingquantitativeoutcomesbytheendof
candidaturethansupervisors,withcandidatesplacinghighimportanceonpublishingatleastfour
papers,andfairlyhighimportanceonwinningawardsandgrants(Fig.1b).Incontrast,supervisors
expectedthecandidatetopublishjustoneortwoarticlesbytheendofcandidature,andplacedlow
importanceonwinningawardsandgrants.Bothgroupsplacedhighimportanceondiscipline
knowledge,criticalthinkingskillsandwrittencommunication,althoughsupervisorsplacedallthese
qualitativeoutcomeshigherinimportancethanthecandidatesthemselves.
Responsibilitytoprovideguidance
Bothcandidatesandsupervisorsconsideredthatsupervisorshaveanimportantresponsibilityto
provideguidancedevelopingthecandidate’swrittencommunicationskills,criticalthinkingskillsand
disciplineknowledge.Butcandidatesplacethehighestimportanceondevelopingwritten
communicationskills,whilesupervisorsoncriticalthinking(Fig.2).However,candidatesconsidered
thatsupervisorsshouldhaveaslightlystrongerroleinencouragingordevelopingtheiracademic
independence,motivationandteamworkskills.
Guidancegiven/received
Acrossattributes,candidatesgenerallyconsideredtheywerebeinggivenlessguidancethan
indicatedbythesupervisor(Table3).Forwrittencommunication,criticalthinking,anddiscipline
knowledge(thethreeattributesidentifiedbythecandidatesasbeingthemostimportanttoreceive
supervisorguidanceon),candidatesconsideredtheygotnoguidanceoronlyreceivedguidanceif
theydirectlyaskedin23%,20%and29%ofcases,respectively,whilstsupervisorsconsideredthis
wasalmostneverthecase(2%,0%and12%respectively)(Fig.3).Overall,candidatesconsidered
theyreceivedguidanceateveryopportunityorwhentheirsupervisorobservedtheyneededitin
64%ofcases,whistsupervisorsperceivedtheyweregivingthislevelofguidancein82%ofcases.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
10
Qualityofsupervision,mentalwell-being
Candidatesreportedreceiving‘high’(34.5%)or‘veryhigh’(38.1%)qualityPhDsupervision,while
7.1%reportedreceiving‘verylow’qualitysupervision,4%‘low’and15.9%‘average’.Nosupervisor
reportedproviding‘low’or‘verylow’qualitysupervision,18%reported‘average’,68%‘high’and
14%‘veryhigh’.
Morethan35%ofcandidatesreportedthatexperiencesduringtheirPhDhadnegativelyimpacted
theiroverallwell-beinginasignificantway(Fig.4).Candidatesindicatedthattheirmentalwell-being
wassignificantlyinfluencedinanegativewayfrompersonalexpectations(54%),researchprogress
(53%),researchenvironment(32%),supervisorexpectations(31%)andrelationshipwithsupervisor
(29%)(Table4).
Candidateswhoreportedlowersupervisionqualitywerealsomorelikelytoreportexperiencing
higherlevelsofnegativewell-being,whereashigherqualitysupervisionwasrelatedtomorepositive
researchcareeraspirations(Table4).Femalecandidatesreportedreceivinghigherquality
supervisionthanmalecandidates(Table5),yetfemalecandidatesreportedhigherlevelsofnegative
well-beingcomparedtomalecandidates,withsupervisorrelationshipandsupervisorexpectations
beingahighersourceofnegativewell-beinginparticularthanreportedbymales(Table5).
Discussion
InthisstudyweexploredhowtheexpectationsofPhDcandidatesandsupervisorsdifferinrespect
tocandidatePhDgoalsandsupervisorguidance,withanaimtopromoteeffectivesupervisory
strategiesthatincreaseunderstandingandcollaborationbetweencandidateandsupervisor.Our
resultssuggestthatthemajorityofcandidates(72.6%)feltthattheyreceivedbetterthanaverage
supervisionquality,andindeedthereweremanyexpectationsthatareinalignmentbetweenboth
parties.Forexample,atthebeginningofcandidature,bothconsideredthatcandidatemotivation,
enthusiasmforthefieldandacertaindegreeofindependenceareimportantdriversforcandidate
success.However,11.1%ofcandidatesreportedreceivinglowerthanaveragesupervision,whichis
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
11
reflectedbysimilarstudiesinAustralianuniversities(Heath,2002;McGaghetal.,2016).This
dissatisfactionmaybedrivenbyimportantdifferencesinexpectationsthatwereobservedacross
respondents.Inparticular,supervisorsstronglyconsideredthatcandidatesshoulddemonstrate
goodcriticalthinkingskillsfromthestartofcandidature.However,candidatesconsideredthisless
important,insteadbelievingthatgoodacademicgradeswereamoresignificantdemonstrationof
theirabilitytodoaPhD,despitelittleevidenceforthis(Bair&Haworth,2004).Thismayleadto
conflictifsupervisorsexpectcandidatestodemonstratecriticalthinkingwhencandidatesareunsure
ofwhatthisentails(forexample,criticalthinkingmaybeconfusedwith‘booksmart’),ordonot
realisethatthisiswhattheirsupervisorexpectsfromthem.Inthiscase,reflectionbythesupervisor
regarding:i)whetherthisisarealisticorfairexpectationtohaveforanewcandidate(Ellisetal.,
2015),ii)howbesttodeveloptheircriticalthinkingskills(e.g.givingthemrelevantarticlestoreview
andcritique),andiii)effectivecommunicationtoensurethatthecandidateisawarethatthisisan
importantskilltodevelop,mayallbeusefulatavoidingmisunderstandingsandconflictsdownthe
line.
Outcomesattheendofcandidature
Bytheendofcandidature,bothcandidatesandsupervisorsagreedthatthecandidateshould
demonstrategoodcriticalthinkingskills,ahighlevelofdisciplineknowledgeandexcelatwritten
communication.However,candidateshadfarhigherexpectationssurroundingthenumberof
publicationstheywouldachieve,andplacedhigherimportanceonwinningawardsandgrantsthan
didsupervisors.Thismayreflectdifferencesinthinkingbetweencandidatesandsupervisors
regardingultimategoalsofthePhDcandidature.Supervisorsmaybeprimarilyconcernedwith
buildingtherequisiteskillsetrequiredforthecandidatetobesuccessfulinacademiaorindustry
(Gilbertetal.,2004),withtheassumptionthatpublishedpapers(andgrants)willbeanimportant
consequenceoftheseskills(p171;Delamontetal.,2004).Conversely,candidatesmaybemore
focussedonquantitativeoutcomesfromtheirPhDthatprovideavaluableskillsetforfuturejobs
(Roach&Sauermann,2010),evenattheexpenseofqualitativeskills.Forexample,whencandidates
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
12
aregivenalotofhelpinordertopublishquickly,andtherebydonotfullydevelopthewritten
communicationskillsthemselves.Althoughthispotentialconflictiswellknownanddiscussed
amongstacademics,withsomeinstitutionsnowpromotingPhDsbypublicationtoavoidthisconflict
(Jackson,2013),suchpotentiallyminordifferencesinthinkingcanhaveprofoundnegative
consequencesforthecandidate.Atworst,thecandidatemayperceivethatthesupervisorisnot
workingintheirbestinterestsiftheyareworkingtowardsdifferentgoals.Althoughthecomplexity
ofthisscenarioincreaseswhensupervisorshavevestedinterestsincandidatesuccess(MacDonald&
Williams-Jones,2009),itisstilllikelythatinvestedsupervisorsstillconceptuallyprioritizecandidate
skillasanecessarymechanismtowardspublication,ratherthanseepublicationsastheoverallgoal
ofthePhD.Assuch,supervisorsshouldbeexplicitlyawareofwhatthecandidatewishestoachieve
duringtheirPhD,andguidethemtowardsachievablegoalsthatbothpromotesuccessintheir
desiredcareerandmeettherequirementsofthesupervisorandtheinstitution.Critically,
communicatingthatbothcandidateandsupervisorareaimingtowardsthesamespecificgoalswill
avoidconflictsneartheendofcandidature.
Supervisorguidance
Bothcandidatesandsupervisorsagreedthatsupervisorshaveastrongresponsibilitytogive
guidanceandfeedbackoncriticalthinking,writtencommunication,andrelevantdiscipline
knowledge.Interestingly,supervisorsconsideredtheirresponsibilitytoguidecriticalthinkingand
problemsolvinggreaterthanwasexpectedbycandidates.Similartootherstudies,candidates
insteadexpectedmoreguidanceondevelopingtheiracademicindependence,theircollaboration
skills,andmaintainingmotivation(Mowbray&Halse,2010).Yet,supervisorsconsideredtheyhad
littleornoresponsibilityinguidingtheseless‘academic’attributes(Craswell,2007).Thismayhave
disproportionallynegativeeffectsonthecandidate,withstudiesconsistentlyshowingsignificant
emotionalcostswhenindependenceislow(DeLangeetal.,2004;Levecqueetal.,2017;Vanroelen
etal.,2009),andwhenthereislittlesocialorcollaborativeintegrationwithinanacademicgroupor
institution(Gardner,2009;Golde,2000;KovachClarketal.,2009;Pyhältöetal.,2009),bothof
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
13
whichdecreasemotivation(Gagné&Deci,2005;Mason,2012).Critically,attributessuchas
independence,self-confidence,collaborationskills,andsustainedmotivationareallcrucial
attributesforasuccessfulPhDandcareer(Mowbray&Halse,2010;Wuchtyetal.,2007),andare
qualitiesthatmostmustlearnanddevelopviamentoring,regularinteractionwiththeirresearch
group,andcarefulself-reflection.Activelysupportingandguidingthecandidatetoincreasetheir
autonomyandcollaboratewithothers(eitherwithintheresearchgrouporwithoutside
collaborators)islikelytohavebeneficialeffectsoncandidatemotivationandproductivity(Larivière,
2011),whilsthelpingthemdevelopindependentideasandteamworkskills(Sinclairetal.,2014).
Supervisorsmaythereforeincreasethechancesofcandidatesatisfactionandsuccessbyactively
encouragingthecandidatetoreflectonhowtheycandevelopthesecrucialqualities.
Candidatemental-well-being
WeshowthatcandidaturehasnegativelyinfluencedoverathirdofPhDcandidates’surveyed
mentalwell-being,thisisinlinewithfindingselsewhere(Evansetal.,2018;Levecqueetal.,2017).
Ourfindingsthatpersonalexpectations,researchprogress,researchenvironmentandsupervision
impactarelativelylargeproportionofcandidatesalsoechoespreviouslypublishedword(Barryet
al.,2018;Evansetal.,2018).Relationshipbetweencandidateandsupervisorcomesuptimeand
againasasignificantcontributortonegativewell-being.Though,asshownhere,thedynamicsof
howthatrelationshipimpactswell-beingissometimescomplicated.Forinstance,despiteonaverage
femalecandidatesself-reportinghigherlevelsofsupervisionqualitythanmalecandidates,
supervisorexpectationsandrelationshiphadagreaternegativeimpactontheirwell-being.There
areseveralpossiblyexplanationsforthisdynamicthatwarrantsfurtherinvestigation.
Effectivecommunicationandhiddenpowerdynamics
Ourresultssuggestthatthereisadisparitybetweenhowmuchguidancesupervisorsthinktheyare
givingtothecandidate,andhowmuchcandidatesperceivetheyarereceiving.Inparticular,
candidatesfelttheyreceivednoguidanceatallin20%ofcases,whilstthiswasveryrareamongst
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
14
supervisors(3%).Worryingly,over20%ofcandidatesreportlittleguidance(noneoronlyifasked)
forcriticalthinking,writtencommunicationanddisciplineknowledge,despitebothparties
identifyingthesesubjectsasimportantareasforsupervisorguidance.Thisdisparityislikelytostem
fromthefactthatsupervisorsaretimedeficientandjugglemultipleobligations,whilstcandidatesin
contrastoftenfocusalmostsolelyonobligationssurroundingtheirPhD.Consequently,supervisors
mayfeelthattheyonlyhavetimetogiveguidanceifthecandidatespecificallyasksforitor
demonstratesthatitisneeded.Thismaybeperceivednegativelybythecandidate,whomay
interpretthesituationasthesupervisornotmeetingtheirsupervisoryobligations.Onecommon
suggestiontominimisethechancesofthisconflictistospendtimediscussingandoutliningthe
numerousseparateresponsibilitiesandexpectationsofbothcandidateandsupervisor(Moxhamet
al.,2013).Althoughtimeconsumingandrequiringagreatdealofthought,thisisanincredibly
worthwhileinvestmentonbothsides.Moreover,thereareanumberofstrategiesthatcanhelp
candidatesself-reflectandbemoreproductive(Kearns,Forbes,etal.,2008;Kearns,Gardiner,etal.,
2008),andsupervisorsshouldpointstudentstowardstheseresourceswhentheyfeeltheyarenot
abletomeettheneedsofthecandidate.
However,inadditiontothis,supervisorsshouldbeexplicitlyawareofpowerdynamicsandits
consequencesforcandidatedevelopmentandengagement(Grant,2003;Manathunga,2007).An
inherentdiscrepancyexistsinthemajorassumptionsrelatingtoacademicsupervision:firstisthe
assumptionthatbothpartiesareautonomousandrational,andthereforeonequaltermstodebate
anddiscussthedirectionofthecandidate.Thesecondassumptionisthatthesupervisorisawise
andknowingauthoritythatalsoplaystheroleofexaminer,andthatthestudentisawillingdisciple
inneedofguidance(Grant,2003;Johnsonetal.,2000;Manathunga,2007).Theseassumptionsare
inherentlyinconflictwitheachother,becauseinthefirstscenariobothpartiesareequals,whilstin
thesecondadistinctpowerimbalanceexists.Thisdiscrepancyislikelytoleadtoconflictswhen
supervisorsexpectcandidates(asrationalequals)tobringupproblemstheymaybeexperiencing,
whilstcandidatesarelikelytobesubmissiveordeferentialtotheauthorityofthesupervisor,and
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
15
mayfeelthatdisagreeingoraskingforhelpisinappropriateorwillcauseconflict.Assuch,
supervisorsshouldbeawarethatpowerdynamicsplayanimportantroleinmediatingcandidate
behaviour,andthatthismaydiscouragecandidatesfromactivelypursuingautonomyor,conversely,
askingforhelpifrequired(e.g.Diamandis,2017).Althoughitischallengingforsupervisorstowalk
thelinebetweengivingtoolittleguidance(so-calledlaissez-fairesupervision)andtoomuch
(autocraticsupervision)(Delamontetal.,1998;Deuchar,2008;Gardner,2008;VanVugtetal.,2004),
awarenessthatcandidatesmayalsobeconflictedbetweendutifullyfollowingadvice/ordersand
demonstratingacademicindependencemayhelpresolvetherootofsuchconflictswhentheyarise.
Thisconflictmaybeavoidedbyconsistentlyapplyingademocraticleadershipstyle(or‘participative’
leadership),wherebycandidatesareencouragedtotakeamoreparticipativeroleinthedecision
makingprocessfromthestartofcandidature.Ademocraticleadershipstylehasbeenconsistently
showntobethemosteffectivestyleofleadershiptoincreaseperformanceandsatisfactionamongst
groupmembers(Eaglyetal.,2003;Foelsetal.,2000;VanVugtetal.,2004),andallowsthe
candidatetoslowlygrowtheirconfidenceintheirownautonomywithinasupportiveframework.
Summary
ThisstudyattemptstoidentifyconflictingexpectationsbetweenPhDcandidatesandsupervisors
thatarelikelytoleadtoconflict,allowingsupervisorstofocuseffortsonkeyareastoencouragea
successfulworkingrelationship.Ourresultscanbesummarisedbythreesuggestionstosupervisors
thatmayacttoreduceconflictandpromotepositiveoutcomesforbothcandidateandsupervisor:
1. Spendtimeearlyonincandidaturetodiscusstheimportanceof,andaligneachother’s
expectations.Forexample,supervisorsshouldassiststudentsindevelopingcriticalthinking
skillsfromtheoutset,asourresultssuggestthatcandidatesarenotawareof,ordonot
placethesamelevelofimportanceonitassupervisorsdoearlyonintheircandidature.
2. Supervisorsandcandidatesshouldagreetoachievablegoalsthattheyworktowards.
Theseshouldincludebothqualitative(skillsetsthecandidateshouldlearn)andquantitative
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
16
(numberofpapersorgrantstobewon)outcome.Althoughthesedonotneedtobeoverly
specificandmayevolveovertime,thiscommunicatestothecandidatethattheyareboth
workingtowardsthesamesetofgoals.
3. Supervisorsshouldplayastrongerroleinguidingthedevelopmentofcandidateacademic
independenceandcollaborationskills.BotharecriticaltoasuccessfulPhDandcareer.
Supervisorsmayfindthatbroadeningthescopeoftheirsupervisoryroletoactivelyguide
thecandidateindevelopingthesequalitieswillhelpthecandidatemaintainmotivationand
satisfactionoverthecourseoftheirPhD,andleadtomoreproductiveandcollaborative
researchbythecandidate.
4. Maintaineffectivecommunicationanddialoguethroughout.Supervisorsandstudents
shouldagreeonacommunicationstylethatbestfitsboththeirneeds,andregularly
evaluateanddiscusstheircommunicationseffectiveness.
Wefoundthatsupervisorsconsideredthattheygivemoreguidancethancandidatesperceivedthey
receive,andthatsupervisorsoftenonlyprovidedguidancewhenthecandidateasks.Wesuggest
thatcandidatesandsupervisorsexplicitlycommunicatetheirseparateresponsibilitiesand
expectationsregardingthespectrumofskillsneededtosuccessfullycompletelyaPhD.Inaddition,
supervisorsmustbecognisantofinherentpowerdynamicsinthestudent-supervisorrelationship,in
ordertounderstandandremediatecommonmisalignmentofexpectations/goalsthatcouldleadto
dissatisfactionandpotentialconflict.Applyingademocraticleadershipstylefromtheoutsetof
candidaturemayhelpdecreasetheeffectsofthispowerimbalance.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
17
References
A.I.McLeod.(2011).Kendall:KendallrankcorrelationandMann-Kendalltrendtest.https://CRAN.R-
project.org/package=Kendall
Bair,C.R.,&Haworth,J.G.(2004).Doctoralstudentattritionandpersistence:Ameta-synthesisof
research.InHighereducation:Handbookoftheoryandresearch(pp.481–534).Springer.
Barry,K.M.,Woods,M.,Warnecke,E.,Stirling,C.,&Martin,A.(2018).Psychologicalhealthof
doctoralcandidates,study-relatedchallengesandperceivedperformance.HigherEducation
Research&Development,37(3),468–483.https://doi.org/10.1080/07294360.2018.1425979
Benjamini,Y.,&Hochberg,Y.(1995).ControllingtheFalseDiscoveryRate:APracticalandPowerful
ApproachtoMultipleTesting.JournaloftheRoyalStatisticalSociety.SeriesB
(Methodological),57(1),289–300.JSTOR.
Buckley,P.J.,&Hooley1,G.J.(1988).Thenon-completionofdoctoralresearchinmanagement:
symptoms,causesandcures.EducationalResearch,30(2),110–120.
Craswell,G.(2007).Deconstructingtheskillstrainingdebateindoctoraleducation.HigherEducation
Research&Development,26(4),377–391.
DeLange,A.H.,Taris,T.W.,Kompier,M.A.J.,Houtman,I.L.D.,&Bongers,P.M.(2004).Work
characteristicsandpsychologicalwell-being.Testingnormal,reversedandreciprocal
relationshipswithinthe4-waveSMASHstudy.
Delamont,S.,Atkinson,P.,&Parry,O.(2004).Supervisingthedoctorate.McGraw-HillEducation
(UK).
Delamont,S.,Parry,O.,&Atkinson,P.(1998).Creatingadelicatebalance:thedoctoralsupervisor’s
dilemmas.TeachinginHigherEducation,3(2),157–172.
Deuchar,R.(2008).Facilitator,directororcriticalfriend?:Contradictionandcongruenceindoctoral
supervisionstyles.TeachinginHigherEducation,13(4),489–500.
Diamandis,E.(2017).Agrowingphobia.Nature,544(7648),129–129.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
18
Eagly,A.H.,Johannesen-Schmidt,M.C.,&VanEngen,M.L.(2003).Transformational,transactional,
andlaissez-faireleadershipstyles:ameta-analysiscomparingwomenandmen.American
PsychologicalAssociation.
Ellis,A.M.,Bauer,T.N.,Mansfield,L.R.,Erdogan,B.,Truxillo,D.M.,&Simon,L.S.(2015).
NavigatingUnchartedWatersNewcomerSocializationThroughtheLensofStressTheory.
JournalofManagement,41(1),203–235.
Evans,T.M.,Bira,L.,Gastelum,J.B.,Weiss,L.T.,&Vanderford,N.L.(2018).Evidenceforamental
healthcrisisingraduateeducation.NatureBiotechnology,36(3),282–284.
https://doi.org/10.1038/nbt.4089
Foels,R.,Driskell,J.E.,Mullen,B.,&Salas,E.(2000).Theeffectsofdemocraticleadershipongroup
membersatisfactionanintegration.SmallGroupResearch,31(6),676–701.
Gagné,M.,&Deci,E.L.(2005).Self-determinationtheoryandworkmotivation.Journalof
OrganizationalBehavior,26(4),331–362.
Gardner,S.K.(2008).“What’stoomuchandwhat’stoolittle?”:Theprocessofbecomingan
independentresearcherindoctoraleducation.TheJournalofHigherEducation,79(3),326–
350.
Gardner,S.K.(2009).Conceptualizingsuccessindoctoraleducation:Perspectivesoffacultyinseven
disciplines.TheReviewofHigherEducation,32(3),383–406.
Gilbert,R.,Balatti,J.,Turner,P.,&Whitehouse,H.(2004).Thegenericskillsdebateinresearch
higherdegrees.HigherEducationResearch&Development,23(3),375–388.
Golde,C.M.(2000).ShouldIstayorshouldIgo?Studentdescriptionsofthedoctoralattrition
process.TheReviewofHigherEducation,23(2),199–227.
Grant,B.(2003).Mappingthepleasuresandrisksofsupervision.Discourse:StudiesintheCultural
PoliticsofEducation,24(2),175–190.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
19
Gu,J.,Lin,Y.,Vogel,D.,&Tian,W.(2011).Whatarethemajorimpactfactorsonresearch
performanceofyoungdoctorateholdersinscienceinChina:aUSTCsurvey.Higher
Education,62(4),483–502.
Halse,C.(2011).‘Becomingasupervisor’:theimpactofdoctoralsupervisiononsupervisors’
learning.StudiesinHigherEducation,36(5),557–570.
Halse,C.,&Bansel,P.(2012).Thelearningalliance:ethicsindoctoralsupervision.OxfordReviewof
Education,38(4),377–392.
Heath,T.(2002).AquantitativeanalysisofPhDstudents’viewsofsupervision.HigherEducation
Research&Development,21(1),41–53.
Holbrook,A.,Shaw,K.,Scevak,J.,Bourke,S.,Cantwell,R.,&Budd,J.(2014).PhDcandidate
expectations:Exploringmismatchwithexperience.InternationalJournalofDoctoralStudies,
9,329–346.
Ives,G.,&Rowley,G.(2005).Supervisorselectionorallocationandcontinuityofsupervision:Ph.D.
students’progressandoutcomes.StudiesinHigherEducation,30(5),535–555.
Jackson,D.(2013).CompletingaPhDbypublication:AreviewofAustralianpolicyandimplications
forpractice.HigherEducationResearch&Development,32(3),355–368.
Jiranek,V.(2010).Potentialpredictorsoftimelycompletionamongdissertationresearchstudentsat
anAustralianfacultyofsciences.InternationalJournalofDoctoralStudies,5(1),1–13.
Johnson,L.,Lee,A.,&Green,B.(2000).ThePhDandtheautonomousself:Gender,rationalityand
postgraduatepedagogy.StudiesinHigherEducation,25(2),135–147.
Kearns,H.,Forbes,A.,Gardiner,M.,&Marshall,K.(2008).WhenaHighDistinctionIsn’tGood
Enough:AReviewofPerfectionismandSelf-Handicapping.AustralianEducational
Researcher,35(3),21–36.
Kearns,H.,Gardiner,M.,&Marshall,K.(2008).InnovationinPhDcompletion:Thehardyshall
succeed(andbehappy!).HigherEducationResearch&Development,27(1),77–89.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
20
Kiley,M.(2011).Developmentsinresearchsupervisortraining:causesandresponses.Studiesin
HigherEducation,36(5),585–599.
KovachClark,H.,Murdock,N.L.,&Koetting,K.(2009).PredictingBurnoutandCareerChoice
SatisfactioninCounselingPsychologyGraduateStudents.TheCounselingPsychologist,37(4),
580–606.https://doi.org/10.1177/0011000008319985
Larivière,V.(2011).Ontheshouldersofstudents?ThecontributionofPhDstudentstothe
advancementofknowledge.Scientometrics,90(2),463–481.
Levecque,K.,Anseel,F.,DeBeuckelaer,A.,VanderHeyden,J.,&Gisle,L.(2017).Workorganization
andmentalhealthproblemsinPhDstudents.ResearchPolicy,46(4),868–879.
https://doi.org/10.1016/j.respol.2017.02.008
Lindén,J.,Ohlin,M.,&Brodin,E.M.(2013).Mentorship,supervisionandlearningexperienceinPhD
education.StudiesinHigherEducation,38(5),639–662.
Lovitts,B.E.,&Nelson,C.(2000).Thehiddencrisisingraduateeducation:AttritionfromPh.D.
programs.Academe,86(6),44.
MacDonald,C.,&Williams-Jones,B.(2009).Supervisor–studentrelations:Examiningthespectrum
ofconflictsofinterestinbiosciencelaboratories.AccountabilityinResearch,16(2),106–126.
Manathunga,C.(2007).Supervisionasmentoring:Theroleofpowerandboundarycrossing.Studies
inContinuingEducation,29(2),207–221.
Mason,M.M.(2012).Motivation,Satisfaction,andInnatePsychologicalNeeds.InternationalJournal
ofDoctoralStudies,7,259–277.https://doi.org/10.28945/1596
McCormack,C.(2004).Tensionsbetweenstudentandinstitutionalconceptionsofpostgraduate
research.StudiesinHigherEducation,29(3),319–334.
McGagh,J.,Marsh,H.,Western,M.,Thomas,P.,Hastings,A.,Mihailova,M.,&Wenham,M.(2016).
ReviewofAustralia’sResearchTrainingSystem.ReportfortheAustralianCouncilofLearned
Academies,www.acola.org.au.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
21
Mowbray,S.,&Halse,C.(2010).ThepurposeofthePhD:Theorisingtheskillsacquiredbystudents.
HigherEducationResearch&Development,29(6),653–664.
Moxham,L.,Dwyer,T.,&Reid-Searl,K.(2013).ArticulatingexpectationsforPhDcandidatureupon
commencement:ensuringsupervisor/student‘bestfit’.JournalofHigherEducationPolicy
andManagement,35(4),345–354.
Peluso,D.L.,Carleton,R.N.,&Asmundson,G.J.(2011).DepressionsymptomsinCanadian
psychologygraduatestudents:doresearchproductivity,funding,andtheacademicadvisory
relationshipplayarole?CanadianJournalofBehaviouralScience/RevueCanadienneDes
SciencesDuComportement,43(2),119.
Pyhältö,K.,Stubb,J.,&Lonka,K.(2009).Developingscholarlycommunitiesaslearningenvironments
fordoctoralstudents.InternationalJournalforAcademicDevelopment,14(3),221–232.
RCoreTeam.(2017).R:Alanguageandenvironmentforstatisticalcomputing.RFoundationfor
StatisticalComputing.https://www.R-project.org/
Roach,M.,&Sauermann,H.(2010).Atasteforscience?PhDscientists’academicorientationand
self-selectionintoresearchcareersinindustry.ResearchPolicy,39(3),422–434.
Sinclair,J.,Barnacle,R.,&Cuthbert,D.(2014).Howthedoctoratecontributestotheformationof
activeresearchers:Whattheresearchtellsus.StudiesinHigherEducation,39(10),1972–
1986.
Tompkins,K.A.,Brecht,K.,Tucker,B.,Neander,L.L.,&Swift,J.K.(2016).Whomattersmost?The
contributionoffaculty,student-peers,andoutsidesupportinpredictinggraduatestudent
satisfaction.TrainingandEducationinProfessionalPsychology,10(2),102.
VanVugt,M.,Jepson,S.F.,Hart,C.M.,&DeCremer,D.(2004).Autocraticleadershipinsocial
dilemmas:Athreattogroupstability.JournalofExperimentalSocialPsychology,40(1),1–13.
Vanroelen,C.,Levecque,K.,&Louckx,F.(2009).Psychosocialworkingconditionsandself-reported
healthinarepresentativesampleofwage-earners:Atestofthedifferenthypothesesofthe
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
22
demand–control–support–model.InternationalArchivesofOccupationalandEnvironmental
Health,82(3),329–342.
Vilkinas,T.(2008).AnexploratorystudyofthesupervisionofPh.D./researchstudents’theses.
InnovativeHigherEducation,32(5),297–311.
Wuchty,S.,Jones,B.F.,&Uzzi,B.(2007).Theincreasingdominanceofteamsinproductionof
knowledge.Science,316(5827),1036–1039.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
23
Figure1.Attributeranksasindicatedbycandidates(darkgrey)andsupervisors(lightgrey).PanelA)andB)showtherankofthemostimportantattributes
foracadidatestartingaPhDandfinishingaPhDrespectively.Therelativelengthoftheattributebarsrelatestothescaledrankvaluethattheattribute
receivedinrelationtootherattributes.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
24
Figure2.Averagedrankofattributesthatsupervisorshavearesponsibilityfordevelopingina
candidateasindicatedbycandidates(darkgrey)andsupervisors(lightgrey).
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
25
Figure3.Percentageofrespondentswhoreportedlevelsofguidancereceived/provided.Guidancelevelsinclude‘none’,‘onlywhenaskeddirectly’,‘whenI
seeitisneeded’,‘ateveryopportunity’fromgreytoblackrespectively.
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
26
Figure4.ProportionofcandidatesindicatingwhethertheirPhDexperienceshavenegativelyimpactedtheirwell-being(overallwell-being),andwhetherspecificaspectsofcandidaturehasnegativelyimpactedtheirwell-being.Fromstronglydisagreeingthattheirwell-beingwasnegativelyimpactedontheleft(lightgrey=hasnotexperiencednegativewell-being)tostronglyagreeingontheright(darkgrey=hasexperiencedsignificantnegativewell-being).
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
27
Table1.Participantdetails
Group Type N GenderF/M(Unspecified)
Age(range)
Candidate Current 92 55/35(1) 32±8(22-65) Graduate 21 10/11 28±10(27-61) Discontinued 1 1 -Supervisor 52 27/25 46±9(31–66)
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
28
Table2.Kendallrankcorrelationbetweentheaveragedattributeranksofcandidateandsupervisor.
Ranktestresponses:candidatevs.supervisor tau p-valueAveragerankedattributesofnewcandidates 0.660 <0.001
Averagerankedattributesofendcandidates 0.623 <0.001
Averagerankedresponsibilitytoprovideguidanceforattribute 0.500 0.008
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
29
Table3.Relationshipbetweenperceivedlevelsofguidancereceived/providedbycandidatesand
supervisor.Positivezvaluesindicatesupervisorsreportinghigherlevelsofguidance.p-valueisFDR
corrected.‘Other’attributeisnotincludedhereandwasrankedasC=14andS=16.
Attribute Predictor Estimate Std.error zvalue p-valueProp.agreeactive
guidance(C/S;%)
Rankresponsibility
(C/S)Criticalthinking Intercept 1.398 0.238 5.870 0.000
df=159 Cand./sup. 18.168 1520.847 0.012 0.990 80/100 2/1
Writtencommunication
Intercept
1.185 0.224 5.286 0.000
df=159 Cand./sup. 2.707 1.035 2.617 0.014 77/98 1/2
Problemsolving Intercept 1.135 0.221 5.130 0.000
df=159 Cand./sup. 18.431 1520.847 0.012 0.990 76/1 10/4
Independence Intercept 1.062 0.219 4.845 0.000
df=156 Cand./sup. 2.095 0.754 2.777 0.010 74/96 5/9
Disciplineknowledge Intercept 0.904 0.210 4.313 0.000
df=159 Cand./sup. 1.089 0.483 2.254 0.033 71/88 3/3
Motivation Intercept 0.817 0.206 3.970 0.000
df=159 Cand./sup. 1.380 0.514 2.682 0.013 69/90 6/15
Understandingofscientificethics
Intercept
0.775 0.204 3.796 0.000
df=158 Cand./sup. 1.194 0.481 2.481 0.020 68/88 16/11
Self-confidence Intercept 0.680 0.202 3.367 0.002
df=158 Cand./sup. 2.072 0.629 3.295 0.002 66/94 8/7
Verbalcommunication Intercept 0.521 0.197 2.640 0.013
df=158 Cand./sup. 2.658 0.748 3.553 0.000 63/96 15/10
Teamworkandcollaborativeskills
Intercept
0.521 0.197 2.640 0.013
df=158 Cand./sup. 0.996 0.418 2.385 0.024 63/82 7/12
Projectmanagementskills
Intercept
0.383 0.193 1.981 0.063
df=159 Cand./sup. 1.609 0.476 3.380 0.002 59/88 4/5
Timemanagementskills
Intercept
0.368 0.194 1.896 0.073
df=158 Cand./sup. 1.291 0.432 2.989 0.006 59/84 11/8
Disciplinespecifictechnicalskills
Intercept
0.219 0.192 1.142 0.293
df=157 Cand./sup. -0.342 0.345 -0.991 0.357 55/47 9/6
Self-reflectionskills Intercept -0.277 0.193 -1.432 0.182
df=156 Cand./sup. 1.095 0.365 2.998 0.006 43/69 12/13
Industryexperience Intercept -0.639 0.201 -3.187 0.002
df=157 Cand./sup. -0.085 0.365 -0.233 0.874 35/33 13/14
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
30
Table4.Kendallnon-parametriccorrelationbetweenhowcandidatesreportedsupervisionquality&
aspectsofwell-being,andsupervisionquality&careeraspirations.p-valuesareFDRcorrected.
Comparisonwithreportedsupervisionquality p-value tau
Well-beingimpactedbycandidatureexperiences <0.001 -0.407
Supervisorrelationshipinfluenceswell-being <0.001 -0.509
Researchenvironmentinfluenceswell-being 0.003 -0.251
Researchprogressinfluenceswell-being 0.068 -0.148
Personalexpectationsinfluenceswell-being 0.028 -0.183
Supervisorexpectationsinfluenceswell-being <0.001 -0.338
Likelihoodofpursuingacareerinresearchacademia 0.026 0.186
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint
31
Table5.Wilcoxtestscomparinggenderdifferencesinexperiencesofsupervisorquality,impactson
negativewell-beingandcareeraspirations.p-valuesareFDRcorrected.
Comparisonwithgender p-value Differenceinlocation Description
Reportedsupervisionquality<0.001 4
Femalereporthigher
qualitysupervision
Negativewell-being <0.001 2
Femalereporthigher
levelsofnegativewell-
being
Supervisorrelationshipimpactsnegativewell-being <0.001 1
Femalereporthigher
levelsofnegativewell-
beingrelatedto
supervisorrelationships
Researchenvironmentimpactsnegativewell-being <0.001 2
Femalesreportbimodal
distribution
Researchprogressimpactsnegativewell-being <0.001 3
Malesreporthigher
levelsofnegativewell-
beingrelatedto
researchprogress
Personalexpectationsimpactnegativewell-being <0.001 3
Femalesskewed
towardsreporting
higherlevels,males
havebimodalreporting
Supervisorexpectationsimpactnegativewell-being <0.001 2
Femalesreporthigher
levelsofnegativewell-
beingrelatedto
supervisorexpectations
Likelihoodofpursuingacareerinresearchacademia <0.001 3
Femalecareer
aspirationsaremore
highlydistributedat
theendoftheoptions
.CC-BY-NC-ND 4.0 International licensepreprint (which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for thisthis version posted February 23, 2020. . https://doi.org/10.1101/2020.02.20.958520doi: bioRxiv preprint