Network analysis in psychology · relationships between variables (e.g., symptoms, signs, psychological processes, personality traits, environmental ... mechanisms. Within this context

sychology since its inception has not ceased in itsefforts to improve our understanding of humanbehavior This continuous impulse has driven the

development of different psychological models that inessence aim to further our knowledge of behavior andpsychological processes (in a broad sense) The newtheoretical and psychometric models may allow us toincorporate an alternative prism with which toconceptualize and rethink psychological phenomena Thenetwork model chaos theory and dynamic systems theoryare just some examples that despite being classic themesin some scientific disciplines are just being incorporatedinto the science of human behavior (Nelson McGorry

Wichers Wigman amp Hartmann 2017) The contributionsof the network model for the analysis ofpsycho(patho)logical variables are especially interesting(Borsboom 2017 Borsboom amp Cramer 2013) This newway of understanding and intervening in behavior hasenormous possibilities since it can among other thingsmotivate alternative ways of analyzing data suggestdifferent ways of modeling and analyzing therelationships between variables (eg symptoms signspsychological processes personality traits environmentaltriggers substance use etc) design new forms ofprevention and intervention strategies andor evenimprove the search for etiological mechanismsWithin this context the objective of this work is to providean introduction to network analysis in psychology Ouraim is to present the network model in a briefentertaining and simple way and as far from technicalitiesand complex statistical jargon as possible The goal is for

NETWORK ANALYSIS IN PSYCHOLOGY

Eduardo Fonseca-PedreroUniversidad de La Rioja Centro de Investigacioacuten Biomeacutedica en Red de Salud Mental (CIBERSAM) Oviedo

El objetivo general de este trabajo es introducir un nuevo enfoque denominado anaacutelisis de redes (network analysis)para su aplicacioacuten al campo de la psicologiacutea Baacutesicamente se trata de presentar el modelo de redes de formabreve amena sencilla y en la medida de lo posible alejada de tecnicismos y aparataje estadiacutestico Es un breve bosquejocuya finalidad es por un lado dar los primeros pasos en el anaacutelisis de redes y por otro mostrar las implicaciones teoacutericas ycliacutenicas subyacentes a este modelo En primer lugar se comentan los oriacutegenes de este enfoque y la forma de comprender losfenoacutemenos psicoloacutegicos concretamente las variables de tinte psicopatoloacutegico Se abordan los conceptos de red nodo y aristalos tipos de redes y los procedimientos para su estimacioacuten Seguidamente se explican las medidas de centralidad y semencionan algunas aplicaciones al campo de la psicologiacutea Posteriormente se ejemplifica en un caso concreto estimando yanalizando una red de rasgos de personalidad dentro del modelo de los Big Five Se aporta la sintaxis correspondiente paraque el lector pueda practicar Finalmente a modo de conclusioacuten se realiza una breve recapitulacioacuten se comentan algunasnotas de reflexioacuten y liacuteneas de investigacioacuten futurasPalabras clave Anaacutelisis de redes Salud mental Medicioacuten Psicologiacutea Psicometriacutea Modelo de redes

The main goal of this work is to introduce a new approach called network analysis for its application in the field of psychologyIn this paper we present the network model in a brief entertaining and simple way and as far as possible away fromtechnicalities and the statistical point of view The aim of this outline is on the one hand to take the first steps in networkanalysis and on the other to show the theoretical and clinical implications underlying this model Firstly the roots of thisapproach are discussed as well as its way of understanding psychological phenomena specifically psychopathologicalproblems The concepts of network node and edge the types of networks and the procedures for their estimation are alladdressed Next measures of centrality are explained and some applications in the field of psychology are mentioned Laterthis approach is exemplified with a specific case which estimates and analyzes a network of personality traits within the BigFive model The syntax of this analysis is provided Finally by way of conclusion a brief recapitulation is provided and somecautionary reflections and future research lines are discussedKey words Network analysis Mental health Measurement Psychology Psychometric Network model

Received 4 septiembre 2017 - Accepted 7 noviembre 2017Correspondence Eduardo Fonseca Pedrero Universidad de LaRioja Departamento de Ciencias de la Educacioacuten Calle LuisUlloa 2 26004 La Rioja Espantildea Email eduardofonsecauniriojaes

A r t i c l e sPapeles del Psicoacutelogo Psychologist Papers 2018 Vol 39(1) pp 1-12httpsdoiorg1023923pappsicol20182852httpwwwpapelesdelpsicologoeshttpwwwpsychologistpaperscom

P

1

it to serve as an introductory tutorial for the psychologypractitioner and to allow us on the one hand to take thefirst steps in network analysis and on the other tounderstand the theoretical and clinical implicationsunderlying this model The thread of exposition in thepresent work will be as follows Firstly the origins of thisapproach are discussed as well as its way ofunderstanding psychological phenomena specificallypsychopathological type variables The concepts ofnetwork model node and edge the types of networks andthe procedures for their estimation are addressed Nextcentrality measures are explained and some applicationsto the field of psychology are mentioned Subsequentlythis is exemplified in a specific case estimating andanalyzing a network of personality traits within the BigFive model The corresponding syntax is provided so thatthe reader can practice it Finally by way of conclusiona brief recapitulation is made some notes of generalreflection as well as possible limitations are discussed andfuture lines of investigation are presented

THE ANALYSIS OF NETWORKS IN PSYCHOLOGYNetwork analysis represents a recent theoreticalapproach in psychology although it is not new in thescientific field It has been applied extensively in otherareas under graph theory for example in the study ofsocial relationships (Borgatti Mehra Brass amp Labianca2009 Newman 2010) Professor Denny Borsboom of the University ofAmsterdam and his group of collaborators havepromoted a different vision with which to conceptualizespecifically psychopathological problems (Borsboom ampCramer 2013 Schmittmann et al 2013) It isexpanding to other areas of psychology that go beyondthe study of mental disorders such as intelligence orvoting attitudes (Maas Kan Marsman amp Stevenson2017) Basically the network model is emerging as aresponse to the medical model predominant in the fieldof psychiatry and some areas of psychology which hasbeen promulgated by the main nosological systems Forexample from the Diagnostic and Statistical Manual ofMental Disorders (DSM) (American PsychiatricAssociation 2013) it is considered that the symptomsand signs that patients refer to have their origin in alatent cause called ldquomental disorderrdquo or ldquomental illnessrdquoThe symptoms are mere passive consequences of acommon latent cause This interpretation is known as the

lsquocommon latent disorderrsquo or lsquocommon cause modelrsquo(Borsboom amp Cramer 2013) It is assumed forexample that phenotypic manifestations such ashallucinations delusions or negative symptoms are dueto an underlying disorder that is causing them in thiscase called schizophrenia (see Figure 1) This medicalapproach to the understanding of abnormal behaviorseems to stem from a false premise a common latentcause Obviously this vision is not without limitationsFor example unlike other fields of medicine inpsychopathology it is difficult to identify a commoncause as a condition that exists independently of itssymptoms and that explains their emergence andcovariance (McNally 2016) In addition this approachleads to tautological reasoning (a person hashallucinations because they suffer from a psychoticepisode they are diagnosed with schizophreniapsychosis because they report having hallucinations)and also to reification In response to these possiblelimitations nosological systems have also beencriticized by other international associations with newways of conceptualizing and classifying mentalproblems even being proposed (eg Research DomainCriteria ndashRDoC- of the National Institute of MentalHealth) (Insel et al 2010) As Fonseca-Pedrero (2017) points out the lsquocommonlatent causersquo model is undoubtedly one of the most usefulways of explaining mental disorders however otherinterpretations whether complementary or otherwise thatallow a full understanding of psychopathologicaldisorders as well as other psychological phenomena(eg personality traits) are possible as well as desirableWe wish to exemplify this point with a case Take forexample a person with sleep problems which disturbtheir mood and their reasoning processes making themsuspicious In turn over time these behaviors lead to astate of general malaise and paranoid ideation thatnegatively impact their ability to concentrate and theiracademicwork performance All this ends up unleashinga set of auditory hallucinatory experiences that alter theirsocial functioning generating disability and the need fortreatment The visual representation of this hypotheticalcase is shown in Figure 2 If this model is taken intoaccount an underlying mental disorder namedschizophrenia would not be the common cause of thecovariance between the signs and symptoms Thesymptoms are grouped because they influence each other


2

A r t i c l e s

Article in press

mutually and not because there is a common latent causethat is explaining their emergence and covariation Thesymptoms do not reflect ldquothe causerdquo but are constitutive ofit (McNally 2016) Therefore one might think thatpsychopathological symptoms and signs are not theemerging manifestations of an underlying mental disorderbut rather they are networks of symptoms dynamiccomplex systems or dynamic constellations of symptoms(and signs) that are causally interrelated (Borsboom ampCramer 2013 Fried van Borkulo Cramer et al 2016)Based on the network model psychopathologicaldisorders are conceived as a complex dynamic system(Cramer et al 2016) It is a system because it analyzesdirect relationships between symptoms It is complexbecause the result cannot be predicted by consideringonly one element of the system It is dynamic because itevolves over time For a more detailed analysis of network analysis thereader can consult the previous excellent works both inEnglish (Borsboom 2017 Borsboom amp Cramer 2013Epskamp Maris Waldorp amp Borsboom in pressMcNally 2016 Schmittmann et al 2013) and Spanish(Fonseca-Pedrero 2017) tutorials (Borsboom amp Cramer2013 Costantini et al 2017 Costantini et al 2015)websites (httppsychosystemsorg httppsych-networkscom httpeiko-friedcom) apps foranalyzing and representing the networks(httpsjolandakosshinyappsioNetworkApp ohttpncasemeloopyv11) or sintaxis in the Renvironment (httpsachaepskampcomfilesCookbookhtml) Readers who wish to take their first stepsusing R can consult the excellent manuals andintroductory articles (Elosua 2009 Field Miles amp Field2012 R Core Team 2016 Ruiz-Ruano amp Puga 2016)

BASIC CONCEPTS IN THE ANALYSIS OFPSYCHOLOGICAL NETWORKSNetwork nodes and edgesA network is an abstract model that contains nodes andedges The nodes represent the objects or variables of thestudy while the edges represent the connections betweenthe nodes that is the ldquolinkrdquo that connects them (see Figure3) Nodes can be all sorts of variables such as forexample psychopathological symptoms personalitytraits or environmental triggers (eg traumaticexperiences cannabis use) (Isvoranu et al 2017 Klippelet al 2017) They could also be some other type of

variable from levels of analysis not observable to thehuman eye (eg genetic brain psychophysiologicalneurocognitive) (Santos Jr Fried Asafu-Adjei amp Ruiz2017) The existing graphical representation betweennodes and edges is known as a graph Suchrepresentations can be executed in R (R Core Team2016) and with specific packages such as Qgraph(Epskamp Cramer Waldorp Schmittmann amp Borsboom2012)

Classification of networksThere are different types of networks depending onwhether the edges are weighted or not andor directed ornot Four types result from their combination namelyunweighted not directed unweighted directed weightednot directed and weighted directed Figure 4 shows avisual representation of this taxonomy

EDUARDO FONSECA-PEDRERO

3

A r t i c l e s

Article in press

FIGURE 2POSSIBLE THEORETICAL MODEL OF RELATIONS BETWEEN

SYMPTOMS PROPOSED FOR A PATIENT DIAGNOSED WITH A

FIRST EPISODE PSYCHOSIS

Note This inter-relation between symptoms has to be seen dynamically (not statically) Sinceit is a model it should be seen as a simplification of reality that has been presented herefor expository and didactic purposes Made with httpncasemeloopyv11

Insomnia

Discomfort

Performance

Suspicion

Deliriousideas Hallucinations Incapacity

FIGURE 1EXAMPLE OF REPRESENTATION OF MENTAL DISORDER BASEDON THE MEDICAL MODEL OF lsquoCOMMON LATENT CAUSErsquo

(REFLECTIVE MODEL)

Schizophrenia

Hallucinations Deliriousideas

Disorganizedlanguage

Disorganizedbehavior

Negativesymptoms

First the edges of the networks can be weighted orunweighted In unweighted networks the nodes areconnected without any force or weight whereas in theweighted networks there is a value a coefficient which isindicative of the magnitude of this connection This valueis represented by the thickness of the edge and oscillatesbetween - + 1 The closer to +1 or -1 the value is thegreater the thickness of the edge and the greater thestrength of the association between nodes It follows thatthe association between nodes can be positive ornegative A negative association negative sign of thecoefficient is usually represented with the color red and apositive one with a positive sign of coefficient isrepresented with the color green A value of 0 indicatesthe absence of an edge connecting the nodesSecond the edges of the networks can be non-directedor directed Undirected networks consist of edges orsimple lines connecting pairs of nodes where there is anassociation of a certain magnitude but the direction ofthis relationship is not indicated (eg if node X causes theactivation of node Y or vice versa) Graphically thecolored lines (red and green) that connect the nodeswould not have arrows at their end point For their partdirected networks allow the direction of the prediction

between nodes to go both ways Directed networks consistof edges with arrowheads at one end of the edgepointing in the direction of the prediction and perhapscausal relationships

Network estimationPsychological networks need to be estimated Thisestimation is based on a matrix of correlations that canbasically be of three types a) simple b) partial and c)partially regularized The simple correlations orassociation network are the graphical representationderived from the Pearson correlation matrix The partialcorrelations or concentration network allow us to see thecorrelation between node A and node B controlling theeffect of the rest of the nodes of the network that iscontrolling the spurious correlations that can emerge due tothe multiple comparisons The estimation of the network iscarried out by means of an algorithm called Fruchterman-Reingold Regularized partial correlations implement aregularization procedure which essentially requires fewerparameters to be estimated so it allows us to extract a stableand easy to interpret network In this case the network canbe estimated with the Least Absolute Shrinkage andSelection Operator (LASSO) or with a variation calledGraphical-LASSO (G-LASSO) (Epskamp Borsboom ampFried 2017) The choice of estimation method is not atrivial matter and should not be left to chance because it canhave a great impact both on the resulting structure of the


4

A r t i c l e s

Article in press

FIGURE 3

EXAMPLE OF ESTIMATED NETWORK

Note The circles represent nodes (variables) The edges represent the relationship betweenthe nodes For example node A could represent suicidal ideation node B bullying etc Thegreater the value of the edge coefficient the thicker the line and therefore the strongerassociation between nodes The green color of the edge indicates a positive relationshipbetween nodes (variables) The red color of the edge indicates a negative relationshipbetween nodes (variables)

FIGURE 4TYPES OF NETWORKS

Unweightedand undirected

A

B

CD

E

A

B

CD

E

A

B

CD

E

A

B

CD

E

Unweightedbut directed

Weighted anddirected

Weighted butundirected

estimated network and on the conclusions drawn from thisstructure (Epskamp Kruis amp Marsman 2017)

Analyzing the structure of the network centralitymeasures Based on the estimated network different inferences canbe made that help us to understand its structure as well asexamine the relative importance of the nodes within it Toanalyze the structure of the network there are measuresof a) distance and shortest path length b) centrality andc) connectivity and clustering Only the measures ofcentrality will be presented here so the reader whowishes to learn more about the other measures of networkinference can consult the previous works (Costantini et al2015)Centrality measures ask which is the most importantnode in the network They allow us to analyze the relativeimportance of the node within the network depending onthe pattern of connections In an estimated network not allnodes are equally important A node is central if it hasmany connections A node is peripheral ie it is on theoutside of the network if it has few connections In orderto know if the node is central (important and influential) inthe network the following must be taken into account a)degree and strength b) closeness and c) betweenness Strength centrality refers to the magnitude of theassociation with the other nodes ie it is close to othernodes A node with a high centrality in this parameter isa node that influences many other nodes Closenesscentrality is defined as the inverse of the sum of thedistance from one node to all other nodes in the networkA node with a high closeness centrality index is a nodethat can predict other nodes well Betweenness is definedas the number of times a node is between two othernodes Betweenness is the number of shortest pathsbetween any two nodes (the shortest route from node A tonode B) that passes a specific nodeStatistical programs allow us to extract these centralityindexes (in Z scores) referring to strength closenessandor betweenness as well as generating graphs andtables based on them (see Figures 5 and 6 below)

SOME APPLICATIONS IN THE FIELD OF PSYCHOLOGYIt has not been until relatively recently that thepsychological literature has focused on a networkapproach to model psychological phenomena In thisshort history excellent scientific contributions have been

made a true reflection of the interest it has arousedamong professionals and researchers in psychology andrelated sciences The themes of study under the networkmodel are current topics under great expansion Servingas a sample are works that have analyzed depressivesymptomatology (Bringmann Lemmens HuibersBorsboom amp Tuerlinckx 2015 Cramer et al 2016Fried van Borkulo Epskamp et al 2016) psychosis andits relationship with traumatic experiences orenvironmental impacts (Isvoranu Borsboom van Os ampGuloksuz 2016 Isvoranu et al 2017) negativepsychotic symptoms (Levine amp Leucht 2016) attenuatedpsychotic symptoms (Fonseca-Pedrero 2018) substanceabuse (Rhemtulla et al 2016) quality of life(Kossakowski et al 2016) post-traumatic stresssymptoms (McNally et al 2014) comorbidity (CramerWaldorp van der Maas amp Borsboom 2010)relationship between symptoms and disorders fromtaxonomic systems (Boschloo et al 2015 Tio EpskampNoordhof amp Borsboom 2016) emotional andbehavioral problems (Boschloo Schoevers van BorkuloBorsboom amp Oldehinkel 2016 Fonseca-Pedrero 2017)and intelligence (Maas Kan Marsman amp Stevenson2017) to name but a fewRecently Borsboom (2017) has proposed a theoreticalmodel of mental disorders from this perspective In histheory he posits five theoretical principles in relation to thestructure and dynamics of symptom networks specificallycomplexity symptom-component correspondence directcausal connections mental disorders follow networkstructure and hysteresis First complexity refers to theinteraction that is established between the differentcomponents of the network Second correspondencerefers to the relationship between the components of thenetwork and the symptoms of psychological problemsThird the structure is generated by a pattern of directconnections between the symptoms Fourth thepsychopathological network has a nontrivial topologythat is some symptoms are more strongly connected thanothers (eg a particular symptom within a mentaldisorder is more connected to the symptoms of thatspecific disorder than to the symptoms of other clinicalsyndromes) Fifth hysteresis refers to the phenomenon bywhich a certain event external to the network (egtraumatic experiences) can affect it and the subsequentabsence of such event or external event does notnecessarily deactivate the network In other words the


5

A r t i c l e s

Article in press

symptoms continue to activate each other even when theexternal trigger event has disappeared Finally from thismodel the notion of mental health would correspond to astable state of a weakly connected network whereasmental disorders would correspond to stable states ofnetworks of strongly connected symptoms For exampleunder this approach psychopathological syndromes (seefor example psychotic disorders) are conceived asdynamic causal networks of mental states with increasinglevels of psychopathological severity an aspect totallyconsistent with current staging models (Fonseca-Pedrero2018 McGorry amp van Os 2013 Nelson et al 2017Wigman et al 2013) Network theory has clear implications for our way ofunderstanding the psychological diagnosis and treatmentFor example structural analysis of the psychologicalnetwork and centrality measures have clear clinicalimplications It is possible to judge which symptoms aremost important in the network to use the most centralsymptoms to diagnose and plan the treatment or focus thetreatment on a symptom or the network of symptoms thathave the most connections It is also possible to identifyldquobridgerdquo symptoms that is a symptom that serves as a linkbetween two sets of networks and whose approach andintervention may enable the of controlling the(hypo)activation of other subnetworks For Borsboom(2017) the diagnosis involves identifying networks ofsymptoms while the treatment involves changing ormanipulating the psychopathological network in threeways namely a) interventions on symptoms (modifying thestatus of one or more symptoms) b) interventions in theexternal field (eliminating the triggering cause or causes)and c) network interventions (modifying the connectionsbetween the nodes of the network ie symptom-symptom)For example in the case of a patient with a psychoticspectrum disorder in which an antipsychotic treatment isimplemented a family intervention can be considered tomodify communication patterns or eliminate substance useandor work with cognitive behavioral techniques thatallow us to cope with the delusions of persecution in orderto reduce the associated hallucinatory experiences As thereader can see all of these issues are highly relevant toclinical practice

AN EXAMPLE OF NETWORK ANALYSIS INPERSONALITYIn this section we present briefly and by way of

example a network analysis of personality specifically toanalyze the big five factors of personality (Extraversion ndashE Conscientiousness ndashC Openness ndashO Agreeableness ndashA Neuroticism ndashN) evaluated using 25 items (seeappendix) Each of these dimensions is valued using fiveitems A sample of 2800 participants was used (M =288 years SD = 111 years) which is available in theldquopsychrdquo package (Revelle 2015) of the R environment (RCore Team 2016) The network was estimated usingQgraph (Epskamp et al 2012) The estimated network isweighted and not directed The G-LASSO algorithm wasused The reader can find the corresponding syntax in theappendixThe results of both the estimated psychological networkand the centrality indexes are presented in Figures 5 and6 It was previously noted that a node is central if it hasmany connections and its centrality basically depends onthe strength closeness and betweenness Figure 6 showsthe standardized values referring to these threeparameters The indices are all on the same scale ofmeasurement and they are standardized (z scores)which allows the comparison among them As can beseen the items that had the highest coefficients ofcentrality in strength were C4 (ldquoDoing things by halvesrdquo)and C2 (ldquoContinuing until everything is perfectrdquo) Incloseness items O4 (ldquoTaking time to reflect on thingsrdquo)E5 (ldquoTaking controlrdquo) and E4 (ldquoMaking friends easilyrdquo)had the highest coefficients of centrality And inbetweenness items N4 (ldquoOften feeling sadrdquo) E4(ldquoMaking friends easilyrdquo) and C2The items in the Conscientiousness dimension seem tohave the strongest connections In this case the strengthof centrality reflects the probability with which theactivation of one of these nodes (itemscharacteristics)will be followed by the activation of other nodes in thenetwork The items of the Extraversion dimension and item4 of Openness presented a high closeness centralityindicating that they are nodes that can predict othernodes (itemstraits) of the network well Items N4 E4 andC2 presented a high centrality of betweenness In otherwords they are nodes (itemstraits) that are often locatedbetween (in the middle of) other nodes and passingthrough them are the shortest paths among other nodes ofthe networkIt is worth noting that for a correct interpretation of thenetwork the reader should not only focus their assessmenton a visual inspection A problem to be avoided in


6

A r t i c l e s

Article in press

psychological networks is precisely over-interpretation atthe time of their visualization This aspect refers especiallyto the design and placement of nodes in the graph forexample when the nodes of the network are grouped ina cluster However the reader needs to know that thelocation of the node within a network is only one of manyequally lsquocorrectrsquo ways of placing the nodes in thenetwork ie in the same sample the location of the nodesin the network could be different in a new estimateTherefore we must be cautious when making a visualinterpretation of the nodes and their position in thenetwork Although it is not the subject of this tutorial fora better interpretation of the psychological network onecould analyze the communalities (Golino amp Epskamp2017) andor predictability (Haslbeck amp Fried 2017)From these results we can better understand thestructural relationship established between the big five

personality traits as a complex system of affectivecognitive and behavioral characteristics

RECAPITULATIONThe purpose of this article was to provide an introductionto the analysis of psychological networks In essence theaim was to present this fertile approach to the psychologyprofessional in a completely didactic way Currently the network model is presented in society as apromising approach in the way of conceptualizingpsycho(patho)logy (Fried amp Cramer 2017) In fact someauthors believe that network analysis can transform thefield of psychopathology (McNally 2016) to a certainextent Based on the network model an underlying latentvariable would not be the cause of the covariance of thesymptoms nor would the symptoms be interchangeableindicators of an underlying disorder Consequently thesymptoms do not reflect underlying mental disorders theyare constitutive of them For this reason network analysiscan have a relevant role in the understanding of forexample psychopathological phenomena avoiding thelimitations of the medical model based on a common latentcause It is understood that network analysis can provideclues about the psychological mechanisms that underlie thedevelopment and maintenance of mental health problems It is essential to incorporate different viewpoints andperspectives that help us to rethink human behavior (in abroad sense) There is no doubt that the understanding


7

A r t i c l e s

Article in press

FIGURE 5ESTIMATED NETWORK FOR PERSONALITY TRAITS

FROM THE BIG FIVE MODEL

Note The numbers of the nodes represent the items of the questionnaire (see Annex)Extraversion -E Conscientiousness ndashC Openness -O Agreeableness -A Neuroticism -N

FIGURE 6CENTRALITY MEASURES FOR THE ITEMS OF THE PERSONALITY

QUESTIONNAIRE

Note For a correct interpretation the values of the X axis are standardized (Z scores) Thenumbers correspond to the items in the questionnaire (see Annex) Extraversion ndashEConscientiousness ndashC Openness ndashO Agreeableness ndashA Neuroticism ndashN

AgreeablenessOpenness

ExtraversionNeuroticismConscientiousness

and study of human behavior is a complex task in whichan infinite number of variables operate from multiplelevels of analysis (biological psychological and social)In any case whether or not the network model helpschange the current epistemological and methodologicalapproach to psychology specifically psychopathology atleast this approach is presented as a new approach fromwhich to observe measure analyze understand andintervene in psycho(patho)logical phenomena (Fonseca-Pedrero 2017) In essence it aims to respond to certainproblems that some areas of current psychology sufferfrom such as overcoming the notion of the latent variableand the supposed underlying cause Obviously networkanalysis should not be seen as incompatible with othertheoretical and methodological approaches but rather asa complementary approach Its correct use and itsusefulness depend on the objective of study and theparticular interests of the clinician or researcher as well ason other aspects such as the appropriate use and qualityof the measurement instruments used (Fonseca-Pedrero ampMuntildeiz 2016 2017 Hernaacutendez Ponsoda Muntildeiz Prietoamp Elosua 2016)Research in network analysis is currently in its infancyso it is necessary to continue working on the constructionof a solid and refutable scientific model and toincorporate new scientific evidence (Borsboom 2017)Obviously this model is not exempt from limitations andsome authors have made certain precautionaryreflections (Guloksuz Pries amp van Os 2017 WichersWigman Bringmann amp de Jonge 2017) First studiesunder this perspective have a clear time cost especiallythose that perform longitudinal follow-ups on theparticipants Second psychometric network modelshave not yet been consolidated and are computationallycomplicated even for experts in the field Third we mustdistinguish between the scientific studies that allow ananalysis under this perspective and those that do not Inother words not all studies have to be seen from theprism of networks Fourth the network method with itsimpressive and elegant technology may be detrimentalto qualitative narrative analyses and prototypical ratherthan polythetic classifications Fifth psychologicalnetworks involve (and at the same time have a tendencytowards) homogenizing the symptoms when the samesymptoms could be qualitatively different an aspect thatrequires a phenomenological analysis of theirqualitative differences (Parnas 2015 Peacuterez Aacutelvarez

2012 Peacuterez-Aacutelvarez amp Garciacutea Montes 2018 Sass1992) Sixth one should not engage in a kind ofmethodologicizing In other words the method must beat the service of the psycho(patho)logical issues andproblems and not vice versa Seventh considerationshould be given to the need to incorporate measurementerror in the estimation of the networkMany interesting lines of research will open up in thecoming years First it would be interesting to movetowards models of multilevel networks that allow us tointegrate studies that gather information from multiplelevels of analysis within a translational andinterdisciplinary strategy Second it would be useful tostart analyzing behavior from a perspective that isdynamic (longitudinal) personalized (individual) andstaging (severity levels) (Fusar-Poli McGorry amp Kane2017 Nelson et al 2017 Van Os et al 2013)including the possibility of designing strategies for thediagnosis intervention or even functional analysis ofbehavior For example individualized interventions couldbe designed based on the estimated network structureand connectivity of the signs and symptoms Fourth itwould be interesting to make simpler and ldquomore userfriendlyrdquo statistical programs and packages that could beused by the psychology practitioner to enable amongother things the establishment of relationships betweensymptoms at the scale on which the clinician worksThe network model represents an advance in theapproach understanding and measurement ofpsychological phenomena Naturally future studies willdetermine the true usefulness and depth of the networkmodel in psychology Be that as it may the road ahead isexciting to say the least

ACKNOWLEDGMENTSThe author would like to thank professors Alicia Peacuterez deAlbeacuteniz Joseacute Muntildeiz and Marino Peacuterez for theircomments regarding a preliminary version of this work This research has been funded by the Ministry of Scienceand Innovation of Spain (MICINN) (reference PSI2014-56114-P) by the Carlos III Institute The BiomedicalResearch Center for Mental Health Network (CIBERSAM)and by the BBVA Foundation 2015 Call for Proposals forSupport to Researchers and Cultural Creators

CONFLICT OF INTERESTSThere is no conflict of interest in this article


8

A r t i c l e s

Article in press

REFERENCESAmerican Psychiatric Association (2013) Diagnostic and

Statistical Manual of Mental Disorders (5th ed)Washington DC American Psychiatric Association

Borgatti S P Mehra A Brass D J amp Labianca G(2009) Network Analysis in the Social SciencesScience 323 892ndash896

Borsboom D (2017) A network theory of mentaldisorders World Psychiatry 16 5ndash13

Borsboom D amp Cramer A O (2013) Networkanalysis an integrative approach to the structure ofpsychopathology Annual Review of ClinicalPsychology 9 91ndash121

Boschloo L Schoevers R A van Borkulo C DBorsboom D amp Oldehinkel A J (2016) The networkstructure of psychopathology in a community sample ofpreadolescents Journal of Abnormal Psychology125(4) 599ndash606

Boschloo L van Borkulo C D Rhemtulla M Keyes KM Borsboom D amp Schoevers R A (2015) Thenetwork structure of symptoms of the diagnostic andstatistical manual of mental disorders PLoS One10(9) e0137621

Bringmann L F Lemmens L H J M Huibers M J HBorsboom D amp Tuerlinckx F (2015) Revealing thedynamic network structure of the Beck DepressionInventory-II Psychological Medicine 45 747ndash57

Costantini G Epskamp S Borsboom D Perugini MMotildettus R Waldorp L J amp Cramer A O J (2015)State of the aRt personality research A tutorial onnetwork analysis of personality data in R Journal ofResearch in Personality 54 13ndash29

Costantini G Richetin J Preti E Casini EEpskamp S amp Perugini M (2017) Stability andvariability of personality networks A tutorial onrecent developments in network psychometricsPersonality and Individual Differenceshttpsdoiorg101016jpaid201706011

Cramer A O J van Borkulo C D Giltay E J vander Maas H L J Kendler K S Scheffer M ampBorsboom D (2016) Major depression as a complexdynamic system Plos One 11(12) e0167490

Cramer A O J Waldorp L J van der Maas H L Jamp Borsboom D (2010) Comorbidity a networkperspective The Behavioral and Brain Sciences 33(2ndash3) 137ndash193

Elosua P (2009) iquestExiste vida maacutes allaacute del SPSS

Descubre R [Is there life beyond SPSS Discover R]Psicothema 21(4) 652ndash655

Epskamp S Borsboom D amp Fried E I (2017)Estimating psychological networks and their accuracya tutorial paper Behavior Research Methods 1ndash34

Epskamp S Cramer A O J Waldorp L JSchmittmann V D amp Borsboom D (2012) qgraphNetwork visualizations of relationships in psychometricdata Journal of Statistical Software 48(4) 1ndash18

Epskamp S Kruis J amp Marsman M (2017) Estimatingpsychopathological networks Be careful what you wishfor PLoS ONE 12(6)

Epskamp S Maris G Waldorp L J amp Borsboom D(in press) Network psychometrics In P Irwing DHughes amp T Booth (Eds) Handbook of PsychometricsNew York NY Wiley

Field A Miles J amp Field Z (2012) DiscoveringStatistics Using R London SAGE

Fonseca-Pedrero E (2017) Anaacutelisis de redes iquestunanueva forma de comprender la psicopatologiacutea[Network analysis A new way of understandingpsychopathology] Revista de Psiquiatria y SaludMental 10 206-215

Fonseca-Pedrero E(Coordinador) (2018) Evaluacioacuten delos trastornos del espectro psicoacutetico [Assessment ofpsychotic spectrum disorders] Madrid Piraacutemide

Fonseca-Pedrero E amp Muntildeiz J (2016) Advances inpsychological assessment Papeles del Psicoacutelogo 371-2

Fonseca-Pedrero E amp Muntildeiz J (2017) Quintaevaluacioacuten de tests editados en espantildea mirando haciaatraacutes construyendo el futuro [Fifth review of testspublished in Spain Looking back building the future]Papeles del Psicoacutelogo 38 161ndash168

Fried E amp Cramer A (2017) Moving forward challengesand directions for psychopathological network theory andmethodology Perspectives on Psychological Science doi1011771745691617705892

Fried E I van Borkulo C D Cramer A O JBoschloo L Schoevers R A amp Borsboom D (2016)Mental disorders as networks of problems a review ofrecent insights Social Psychiatry and PsychiatricEpidemiology 58(12) 7250ndash7257

Fried E I van Borkulo C D Epskamp S SchoeversR A Tuerlinckx F amp Borsboom D (2016)Measuring depression over time Or not Lack ofunidimensionality and longitudinal measurement


9

A r t i c l e s

Article in press

invariance in four common rating scales of depressionPsychological Assessment 28 1354-1367

Fusar-Poli P McGorry PD amp Kane JM (2017)Improving outcomes of first-episode psychosis anoverview World Psychiatry 16 251-265

Golino H F amp Epskamp S (2017) Exploratory graphanalysis A new approach for estimating the number ofdimensions in psychological research PLoS ONE12(6)

Guloksuz S Pries L-K amp van Os J (2017)Application of network methods for understandingmental disorders pitfalls and promise PsychologicalMedicine 5 1ndash10

Haslbeck J M B amp Fried E I (2017) How predictableare symptoms in psychopathological networks areanalysis of 17 published datasets PsychologicalMedicine 19 1-10

Hernaacutendez A Ponsoda V Muntildeiz J Prieto G ampElosua P (2016) Revisioacuten del modelo para evaluar lacalidad de los tests utilizados en Espantildea [Assessing thequality of tests in Spain Revision of the Spanish testreview model] Papeles del Psicoacutelogo 37 161ndash168

Insel T Cuthbert B Garvey M Heinssen R Pine DS Quinn K hellip Wang P (2010) Research domaincriteria (RDoC) toward a new classification frameworkfor research on mental disorders The AmericanJournal of Psychiatry 167(7) 748ndash751

Isvoranu A-M Borsboom D van Os J amp GuloksuzS (2016) A network approach to environmentalimpact in psychotic disorder brief theoreticalframework Schizophrenia Bulletin 42(4) 870ndash873

Isvoranu A M van Borkulo C D Boyette L LWigman J T W Vinkers C H amp Borsboom D(2017) A Network approach to psychosis pathwaysbetween childhood trauma and psychotic symptomsSchizophrenia Bulletin 43 187-196

Klippel A Viechtbauer W Reininghaus U WigmanJ T van Borkulo C MERGE hellip Wichers M(2017) The cascade of stress a network approach toexplore differential dynamics in populations varyingin risk for psychosis Schizophrenia Bulletin doi101093schbulsbx037

Kossakowski J J Epskamp S Kieffer J M vanBorkulo C D Rhemtulla M amp Borsboom D (2016)The application of a network approach to health-related quality of life introducing a new method forassessing HRQoL in healthy adults and cancer patients

Quality of Life Research 25 781ndash792Levine S Z amp Leucht S (2016) Identifying a system ofpredominant negative symptoms Network analysis ofthree randomized clinical trials SchizophreniaResearch 178 17-22

Maas H Van Der Kan K Marsman M amp StevensonC E (2017) Network models for cognitivedevelopment and intelligence Journal of Intelligence5 16 doi 103390jintelligence5020016

McGorry P amp van Os J (2013) Redeeming diagnosisin psychiatry timing versus specificity Lancet 381343ndash345

McNally R J (2016) Can network analysis transformpsychopathology Behaviour Research and Therapy86 95ndash104

McNally R J Robinaugh D J Wu G W Y WangL Deserno M K Borsboom D hellip Borsboom D(2014) Mental disorders as causal systems a networkapproach to posttraumatic stress disorder ClinicalPsychological Science 3(6) 1ndash14

Nelson B McGorry P D Wichers M Wigman J TW amp Hartmann J A (2017) Moving from static todynamic models of the onset of mental disorder JAMAPsychiatry 74 528-534

Newman M E J (2010) Networks An IntroductionOxford United Kingdom Oxford University Press

Parnas J (2015) Differential diagnosis and currentpolythetic classification World Psychiatry 14 284ndash287

Peacuterez Aacutelvarez M (2012) Las raiacuteces de la psicopatologiacuteamoderna La melancoliacutea y la esquizofrenia [The rootsof modern psychopathology Melancholy andschizophrenia] Madrid Ediciones Piraacutemide

Peacuterez Aacutelvarez M amp Garciacutea Montes J (2018)Evaluacioacuten fenomenoloacutegica

maacutes allaacute de los siacutentomas [Phenomenological assessmentbeyond the symptoms] In E Fonseca-Pedrero(Coordinador) Evaluacioacuten de los trastornos delespectro psicoacutetico [Assessment of psychotic spectrumdisorders] Madrid Piraacutemide

R Core Team (2016) R A Language and Environment forStatistical Computing Vienna Austria

Revelle W (2015) Package ldquopsychrdquo - Procedures forPsychological Psychometric and Personality ResearchR Package 1ndash358 Retrieved from httppersonality-projectorgrpsych-manualpdf

Rhemtulla M Fried E I Aggen S H Tuerlinckx F


10

A r t i c l e s

Article in press

Kendler K S amp Borsboom D (2016) Networkanalysis of substance abuse and dependencesymptoms Drug and Alcohol Dependence 161 230ndash237

Ruiz-Ruano A M amp Puga J L (2016) R como entornopara el anaacutelisis estadiacutestico en evaluacioacuten psicoloacutegica[R as the environment for data analysis inpsychological assessment] Papeles del Psicoacutelogo 3774ndash79

Santos Jr H Fried E I Asafu-Adjei J amp Ruiz R J(2017) Network structure of perinatal depressivesymptoms in latinas relationship to stress andreproductive biomarkers Research in Nursing ampHealth 40 218-228

Sass LA (1992) Madness and modernism Insanity inthe light of modern art literature and thoughtHarvard University Press

Schmittmann V D Cramer A O J Waldorp L JEpskamp S Kievit R A amp Borsboom D (2013)Deconstructing the construct A network perspective onpsychological phenomena New Ideas in Psychology

31(1) 43ndash53 Tio P Epskamp S Noordhof A amp Borsboom D(2016) Mapping the manuals of madness Comparingthe ICD-10 and DSM-IV-TR using a network approachInternational Journal of Methods in PsychiatricResearch 25 267-276

van Os J Delespaul P Wigman J Myin-Germeys IWichers M (2013) Beyond DSM and ICDintroducing ldquoprecision diagnosisrdquo for psychiatry usingmomentary assessment technology World Psychiatry12 113ndash117

Wichers M Wigman J T W Bringmann L F amp deJonge P (2017) Mental disorders as networks somecautionary reflections on a promising approach SocialPsychiatry and Psychiatric Epidemiology 52 143ndash145

Wigman J T W Collip D Wichers M Delespaul PDerom C Thiery E hellip van Os J (2013) Alteredtransfer of momentary mental states (atoms) as thebasic unit of psychosis liability in interaction withenvironment and emotions PLoS ONE 8(2)


11

A r t i c l e s

Article in press

Appendix

Figure 2httpncasemeloopyv11data=[[[1547236122Malestar224][2315338122Insomnio225][3535487122Suspicacia220][4874357122Alucinaciones221][5698358122Ideas2520delirantes222][61107351122Discapacidad223][777717303322Rendimiento2520221]][[219410][128910][23-5510][15-410][35-2510][31-5910][545610][454810][465410][644810][13-4910][174610][17-2710][75-4910][57-2210]][[123642322a22]]75D

Content of the 25 items used(available at httpswwwpersonality-projectorgrhtmlbfihtml)

AgreeablenessA1 Am indifferent to the feelings of othersA2 Inquire about othersrsquo well-beingA3 Know how to comfort othersA4 Love childrenA5 Make people feel at easeConscientiousnessC1 Am exacting in my workC2 Continue until everything is perfectC3 Do things according to a planC4 Do things in a half-way mannerC5 Waste my timeExtraversionE1 Donrsquot talk a lotE2 Find it difficult to approach othersE3 Know how to captivate peopleE4 Make friends easilyE5 Take chargeNeuroticismN1 Get angry easilyN2 Get irritated easilyN3 Have frequent mood swingsN4 Often feel blueN5 Panic easilyOpenness O1 Am full of ideasO2 Avoid difficult reading materialO3 Carry the conversation to a higher levelO4 Spend time reflecting on thingsO5 Will not probe deeply into a subject


12

A r t i c l e s

Article in press

Appendix (Continuation)

Syntaxis in RInstall R httpscranr-projectorgmirrorshtmlFor consultation httpsachaepskampcomfilesCookbookhtml

installpackages(ldquoqgraphrdquo) install qgraph package

mat2 lt- matrix(c(0 03 0 -03 02 0303 0 -09 0 0 00 -09 0 08 0 0-03 0 08 0 -03 002 0 0 -03 0 003 0 0 0 0 0) ncol = 6 nrow = 6 byrow = TRUE) qgraph(mat2 edgelabels = TRUEesize = 10 labels = LETTERS[16] fade = FALSE) Figure 1

library (ldquopsychrdquo) install the psych packagedata (bfi) load the database called ldquobfirdquoview (bfi) view the database ldquobfirdquosummary (bfi) computer minimum maximum range media etc of the database ldquobfirdquodim (bfi) number of variables and cases of the database ldquobfirdquonames (bfi) name of the variables of the database ldquobfirdquodescribe (bfi) descriptive statistics of the database ldquobfirdquo

bfiSub lt- bfi [ 1 25] selection the first 25 items of the database ldquobfirdquo

corMat lt- cor_auto (bfiSub) compute the correlation between the variables of the database ldquobfirdquo 25 items ordinal measurement scale

Groups lt-c (rep (ldquoKindnessrdquo 5) rep (ldquoResponsibilityrdquo 5) rep (ldquoExtraversionrdquo 5) rep (ldquoNeuroticismrdquo 5) rep (ldquoAperturerdquo 5)) generate groups of items that correspondto the five dimensions each dimension contains 5 items

Graph_lasso lt- qgraph (corMat graph = ldquoglassordquo layout = ldquospringrdquo tuning = 025 sampleSize = nrow (bfiSub) groups = Groups palette = ldquocolorblindrdquo) estimatenetwork with 25 items and 5 dimensions with the GLASSO method Figure 5

centralityPlot (Graph_lasso) estimate the centrality indices Figure 6

it to serve as an introductory tutorial for the psychologypractitioner and to allow us on the one hand to take thefirst steps in network analysis and on the other tounderstand the theoretical and clinical implicationsunderlying this model The thread of exposition in thepresent work will be as follows Firstly the origins of thisapproach are discussed as well as its way ofunderstanding psychological phenomena specificallypsychopathological type variables The concepts ofnetwork model node and edge the types of networks andthe procedures for their estimation are addressed Nextcentrality measures are explained and some applicationsto the field of psychology are mentioned Subsequentlythis is exemplified in a specific case estimating andanalyzing a network of personality traits within the BigFive model The corresponding syntax is provided so thatthe reader can practice it Finally by way of conclusiona brief recapitulation is made some notes of generalreflection as well as possible limitations are discussed andfuture lines of investigation are presented

THE ANALYSIS OF NETWORKS IN PSYCHOLOGYNetwork analysis represents a recent theoreticalapproach in psychology although it is not new in thescientific field It has been applied extensively in otherareas under graph theory for example in the study ofsocial relationships (Borgatti Mehra Brass amp Labianca2009 Newman 2010) Professor Denny Borsboom of the University ofAmsterdam and his group of collaborators havepromoted a different vision with which to conceptualizespecifically psychopathological problems (Borsboom ampCramer 2013 Schmittmann et al 2013) It isexpanding to other areas of psychology that go beyondthe study of mental disorders such as intelligence orvoting attitudes (Maas Kan Marsman amp Stevenson2017) Basically the network model is emerging as aresponse to the medical model predominant in the fieldof psychiatry and some areas of psychology which hasbeen promulgated by the main nosological systems Forexample from the Diagnostic and Statistical Manual ofMental Disorders (DSM) (American PsychiatricAssociation 2013) it is considered that the symptomsand signs that patients refer to have their origin in alatent cause called ldquomental disorderrdquo or ldquomental illnessrdquoThe symptoms are mere passive consequences of acommon latent cause This interpretation is known as the

lsquocommon latent disorderrsquo or lsquocommon cause modelrsquo(Borsboom amp Cramer 2013) It is assumed forexample that phenotypic manifestations such ashallucinations delusions or negative symptoms are dueto an underlying disorder that is causing them in thiscase called schizophrenia (see Figure 1) This medicalapproach to the understanding of abnormal behaviorseems to stem from a false premise a common latentcause Obviously this vision is not without limitationsFor example unlike other fields of medicine inpsychopathology it is difficult to identify a commoncause as a condition that exists independently of itssymptoms and that explains their emergence andcovariance (McNally 2016) In addition this approachleads to tautological reasoning (a person hashallucinations because they suffer from a psychoticepisode they are diagnosed with schizophreniapsychosis because they report having hallucinations)and also to reification In response to these possiblelimitations nosological systems have also beencriticized by other international associations with newways of conceptualizing and classifying mentalproblems even being proposed (eg Research DomainCriteria ndashRDoC- of the National Institute of MentalHealth) (Insel et al 2010) As Fonseca-Pedrero (2017) points out the lsquocommonlatent causersquo model is undoubtedly one of the most usefulways of explaining mental disorders however otherinterpretations whether complementary or otherwise thatallow a full understanding of psychopathologicaldisorders as well as other psychological phenomena(eg personality traits) are possible as well as desirableWe wish to exemplify this point with a case Take forexample a person with sleep problems which disturbtheir mood and their reasoning processes making themsuspicious In turn over time these behaviors lead to astate of general malaise and paranoid ideation thatnegatively impact their ability to concentrate and theiracademicwork performance All this ends up unleashinga set of auditory hallucinatory experiences that alter theirsocial functioning generating disability and the need fortreatment The visual representation of this hypotheticalcase is shown in Figure 2 If this model is taken intoaccount an underlying mental disorder namedschizophrenia would not be the common cause of thecovariance between the signs and symptoms Thesymptoms are grouped because they influence each other


2

A r t i c l e s

Article in press






3

A r t i c l e s

Article in press





Insomnia

Discomfort

Performance

Suspicion



(REFLECTIVE MODEL)

Schizophrenia




Negativesymptoms





4

A r t i c l e s

Article in press

FIGURE 3





A

B

CD

E

A

B

CD

E

A

B

CD

E

A

B

CD

E









5

A r t i c l e s

Article in press





6

A r t i c l e s

Article in press





7

A r t i c l e s

Article in press





QUESTIONNAIRE









8

A r t i c l e s

Article in press




























9

A r t i c l e s

Article in press



























10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press
















3

A r t i c l e s

Article in press





Insomnia

Discomfort

Performance

Suspicion



(REFLECTIVE MODEL)

Schizophrenia




Negativesymptoms





4

A r t i c l e s

Article in press

FIGURE 3





A

B

CD

E

A

B

CD

E

A

B

CD

E

A

B

CD

E









5

A r t i c l e s

Article in press





6

A r t i c l e s

Article in press





7

A r t i c l e s

Article in press





QUESTIONNAIRE









8

A r t i c l e s

Article in press




























9

A r t i c l e s

Article in press



























10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press















4

A r t i c l e s

Article in press

FIGURE 3





A

B

CD

E

A

B

CD

E

A

B

CD

E

A

B

CD

E









5

A r t i c l e s

Article in press





6

A r t i c l e s

Article in press





7

A r t i c l e s

Article in press





QUESTIONNAIRE









8

A r t i c l e s

Article in press




























9

A r t i c l e s

Article in press



























10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press
















5

A r t i c l e s

Article in press





6

A r t i c l e s

Article in press





7

A r t i c l e s

Article in press





QUESTIONNAIRE









8

A r t i c l e s

Article in press




























9

A r t i c l e s

Article in press



























10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press















6

A r t i c l e s

Article in press





7

A r t i c l e s

Article in press





QUESTIONNAIRE









8

A r t i c l e s

Article in press




























9

A r t i c l e s

Article in press



























10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press















7

A r t i c l e s

Article in press





QUESTIONNAIRE









8

A r t i c l e s

Article in press




























9

A r t i c l e s

Article in press



























10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press
















8

A r t i c l e s

Article in press




























9

A r t i c l e s

Article in press



























10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press






































9

A r t i c l e s

Article in press



























10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press





































10

A r t i c l e s

Article in press











11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press





















11

A r t i c l e s

Article in press

Appendix





12

A r t i c l e s

Article in press












12

A r t i c l e s

Article in press











Documents

Network analysis in psychology · relationships between variables (e.g., symptoms, signs, psychological processes, personality traits, environmental ... mechanisms. Within this context