Personality and Social Networks

MISSING IMAGE

Material Information

Title:
Personality and Social Networks Trait Antecedents of Cognitive and Electronic Social Structures
Physical Description:
1 online resource (44 p.)
Language:
english
Creator:
Crosier, Benjamin S
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Psychology
Committee Chair:
Webster, Gregory Daniel
Committee Members:
Chambers, John
Mccarty, Christopher

Subjects

Subjects / Keywords:
networks -- personality
Psychology -- Dissertations, Academic -- UF
Genre:
Psychology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Who we are sculpts the structure of the social network in which we are embedded. Reliable patterns in behavior rooted in psychological traits influence how we socialize with others. Best framed in terms of the Five Factor Model, personality shapes how we interact with others. Once summated, these interactions form our network. Across three samples, this thesis examined the relationship between Big Five personality components and social network structure in university students (Study 1), in school-level network data from nearly 6,500 high school students from the National Longitudinal Study of Adolescent Health (Study 2), and in nearly 10,000 Facebook users from the myPersonality project (Study 3). Overall, personality influences the position people occupy in their egocentric social networks. Specifically, extraversion and age emerged as powerful predictors of brokerage (connecting different cliques), density, network size and centrality (importance or influence) across three studies. The primacy of individual-level over group-level variables relating to network structure is highlighted and the extent to which online social networks reflect real-world ones is explored in depth.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Benjamin S Crosier.
Thesis:
Thesis (M.S.)--University of Florida, 2012.
Local:
Adviser: Webster, Gregory Daniel.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-05-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0044310:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 PERSONALITY AND SOCIAL NETWORKS: TRAIT ANTECEDENTS OF COGNITIVE AND ELECTRONIC SOCIAL STRUCTURES By Benjamin Crosier A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Benjamin Crosier

PAGE 3

3 For Wayne, Diane, Jesse, Bill and Glenn

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank Greg Webster for his support throughout this project and for letting me do whatever I want Additionally, thanks go out to John Chambers and Chris McCarty for their helpful feedback and to Michal Kosinski and David Stillwell for access to myPersonality. Brian Collison and Jenny Howell have been incredibly motivating, never ceasing to scare me in to doing work. I would also like to thank the women who have grounded me throughout the years.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 ABSTRACT ................................ ................................ ................................ ..................... 8 CHAPTER 1 INTRODUCTI ON ................................ ................................ ................................ .... 10 Dimensions of the Big Five ................................ ................................ ..................... 10 Origins of Personality ................................ ................................ .............................. 13 Social Network Analysis ................................ ................................ .......................... 14 Social Network Analysis and Psychology: A Brief History ................................ ....... 17 Cognitive Social Structures ................................ ................................ ..................... 19 Electronic Social Structures ................................ ................................ .................... 20 The Problem of Technological Literacy ................................ ................................ ... 21 2 STUDY 1 ................................ ................................ ................................ ................. 22 Purpose ................................ ................................ ................................ .................. 22 Procedure ................................ ................................ ................................ ............... 22 Participants ................................ ................................ ................................ ............. 22 Measures ................................ ................................ ................................ ................ 23 Results and Discussi on ................................ ................................ ........................... 24 3 STUDY 2 ................................ ................................ ................................ ................. 25 Purpose ................................ ................................ ................................ .................. 25 Procedure ................................ ................................ ................................ ............... 25 Participants ................................ ................................ ................................ ............. 26 Measures ................................ ................................ ................................ ................ 26 Results and Discussion ................................ ................................ ........................... 26 Discussion ................................ ................................ ................................ .............. 29 4 STUDY 3A ................................ ................................ ................................ .............. 30 Purpose ................................ ................................ ................................ .................. 30 Procedure ................................ ................................ ................................ ............... 30 Participants ................................ ................................ ................................ ............. 31 Results and Discussion ................................ ................................ ........................... 31 Purpose ................................ ................................ ................................ .................. 32 Participants ................................ ................................ ................................ ............. 32 Results and Discussion ................................ ................................ ........................... 32

PAGE 6

6 5 GENERAL DISCUSSION ................................ ................................ ....................... 35 APPENDIX: ADDPHES ITEMS ................................ ................................ ..................... 40 LIST OF REFERENCES ................................ ................................ ............................... 41 BIOGRAPHIC AL SKETCH ................................ ................................ ............................ 44

PAGE 7

7 LIST OF TABLES Table page 3 1 Personality predicting degree. ................................ ................................ ............ 27 3 2 Personality predicting betweenness centrality. ................................ ................... 27 3 3 Personality predicting density ................................ ................................ ............. 28 3 4 Personality predicting brokerage ................................ ................................ ........ 28

PAGE 8

8 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science PERSONALITY AND SOCIAL NETWORKS: TRAIT ANTECEDENTS OF COGNITIVE AND ELECTRONIC SOCIAL STRUCTURES By Benjamin Crosier May 2012 Chair: Gregory D. Webster Major: Psychology Who we are sculpts the structure of the social network in which we are embedded. Reliable patterns in behavior rooted in psychological traits influence how we socialize with other s. Best framed in terms of the Five Factor Model, personality shapes how we interact with others. Once summated, these interactions form our network. Across three samples, this thesis examined the relationship b etw een Big Five personality components and social network structure in university students (Study 1), in school level network data from nearly 6,500 high school students from the National Longitudinal Study of Adol escent Health (Study 2), and in nearly 10,000 Facebook users from the m yPersonality project (Study 3). Overall, personality influ ences the position people occupy in their egocentric social networks. Specifically, ext raversion and age emerged as powe rful predictors of brokerage (connecting different c liques), density, network size and centrality (importance or influence) across three studies. T he primacy of in dividual level over group level variables relating to network str ucture is highlighted

PAGE 9

9 and the extent to which online social networks reflect re al world ones is explored in depth.

PAGE 10

10 CHAPTER 1 INTRODUCTION Who we are determines how we relate to the world. More specifically, stable and heritable psychological traits influence the topology of egocentric social networks. This thesis seeks to explore the extent to which personality, best conceptualized using t he Five Factor Model (FFM), influences a set of structural metrics that mathematically capture the shape of personal social networks. Of Big Five components, extraversion is hypothesized to be the primary predictor of network structure. Further, to addre ss methodological and theoretical problems regarding the best way to define and collect these networks, the FFM will be evaluated predictively in relation to networks that are built from cognitive recall tasks and data that are recorded from online behavio r. The relationship between psychological traits and network structure is hypothesized be similar across both types of networks. All too often is there a dismissive distinction drawn between online and offline behavior, one that insinuates electronic int eraction is somehow less genuine. The present investigation seeks to challenge this notion while simultaneously looking at the variables that shape the social world around us. Dimensions of the Big Five The Big Five is comprised of five components: openne ss, conscientiousness, extraversion, agreeableness and neuroticism Openness is a trait that categorizes the degree to which people are oriented towards experiencing novel stimuli, whether it be other people, new situations, art, music and culture rather t han being more focused on all things down to earth and avoiding the complex, abstract and strange (Pervin, 1999) Someone that is very open might be more prone to try exotic cuisine while backpacking in a foreign country, while a person on the other side of the spectrum might have a daily

PAGE 11

11 diet consisting of the same mundane items for years on end. Both behavioral strategies have risks and benefits; neither is optimal in all situations. For instance, be ing very open can reveal many new pleasurable and fitness enhancing opportunities; conversely, there is a considerable protective buffer that accompanies a lack of openness. It may be quite a gift in a threat laden environment to avoid new things. Openne ss can factor into social interaction, and thus, network structure. In light of the potential costs and benefits of the trait, it may or may not be beneficial to be linked to a very open person. From an egocentric perspective, it would be reasonable to a ssume that openness would free one to be connected with a large number of diverse individuals. Openness may loosen prejudicial attitudes, encouraging the formation of Conscientiousness captures h ow drives and impulses are handled (Pervin, 1999) Those on the low end of the trait might chase short term benefit, capriciously following every desire, tending to not see things through until they are f inished. Those on the other side of the spectrum are always considering the long term, perfectly finishing every task in a disciplined manner always considering the eventual payout. Again, it is not necessarily the case that one strategy is superior both may lead to adaptive outcomes in different scenarios. People who are exceedingly conscientious may not be able to recognize a lost cause while those who are not very conscientious may never achieve lofty goals and face potentially negative outcomes (Soldz & Vaillant, 1999) On average, one draws benefits from being conscientious and associating with conscientious people. The trait is usually framed in a positive light and should be r elated to keeping existing ties active in a social network.

PAGE 12

12 Extraversion classifies the tendency for individuals to draw positive emotions either from social interactions with others or from internal sources from within themselves (Pervin, 1999) Those high on extraversion attend many social events and have a great number of friends. Introverts are quite different and are content with being alone and generally have a small number of close friends. While both types can potentially lead healthy and happy lives, there are some typically negative consequences that occur when an individual is exceedingly introverted. Extraversion has been linked to increased happiness and subjective well being, longer life s pan, successful peer relationships, volunteerism, leadership, and a buffer against anxiety disorders (Ozer & Benet Martinez, 2005). This highly socially relevant trait should be the driving force between making new ties and keeping old ones in a social ne twork. In terms of personality, extraversion should be the driving force behind network topology. Agreeableness is simply the propensity to be nice. Those who are agreeable are suspicious of others and are willing to make compromises and comply with social norms. focused on personal desires and sometimes unfriendly (Pervin, 1999) Like all other aspects of personality, agreeableness also has tangible outcomes. Low agreeableness has been linked to poor health while high agreeableness translates into better socioemotional competence, which is lin ked to personal and social benefits (Sjoberg 2001). In terms of social network structure, agreeableness could potentially fuel the presence of ties.

PAGE 13

13 Completing the five factor taxonomy is the single negatively framed trait of neuroticism. Also frequently called emotional instability, neuroticism captures the psychological state, neuroticism is most likely related (Pervin, 1999) This dimension is strongly associated with a plethora of negative outcomes in the domains of health (both physical and mental), relationships and general life quality (Ozer & Benet Martinez, 2005). Neuroticism should serve as a repellant force in social networks as it should foster social isolation. Origins of Personality Like nearly all psychological attributes, personality is shaped by both evolution and the environment in which development occurs. Averaging across the dimensions of the Big Five, pers onality is 48% heritable (openness = 61%, conscientiousness = 44%, extraversion = 53%, agreeableness = 41% and neuroticism = 41%) (Jang, Livesley, & Vemon, 1996) While t his does not mean that half of our personality is determined by our genes, it is indicative of a very strong genetic influence. Given that nature nurture is a false dichotomy, it is evident that personality has a rich causal background founded in both ev olutionary heritage and personal development. Using 123 pairs of identical twins and 127 pair of fraternal twins, Jang and colleagues (1996) determined that a substantial amount of personality as captured by the Big Five was genetically grounded. This is no surprise: a collection of traits that are so central to the essence of what it means to be a human must be related to behaviors that are tied to some selection pressure. Each trait can convey a survival advantage, especially in ancestral environments. Openness fosters willingness for exploration that reveals new opportunities and resources. Conscientiousness focuses efforts on long

PAGE 14

14 term goals that enhance status and provide access to resources. Extraversion puts individuals in contact with other indiv iduals (again opening up avenues for resources exchange). Agreeableness is a sign of a good ally and paternal potential. Neuroticism provides some protective elements and is piggybacked on other traits (Buss, 201 0) The status of personality as a genetically based human universal is further bolstered by evidence that involved traits are smoothly distributed across cultures. Schmitt and colleagues (2007) demonstrated that the structure of the five factor model as measured by the Big Five Inventory was confirmed in a sample of 17,837 participants from 56 different nations speaking 28 different languages. It is important to note that there was significant variation between cultures. Although genes lay the foundat ion for personality, cultural idiosyncrasies and potent local influences in the immediate environment alter the development of personality through learning in a reliable way. Social Network Analysis One of the best ways to conceptualize social environment s is through the relatively new method and theory of social network analysis (Wasserman & Faust, 1994) This approach can adopt one of two perspectives of a social network: sociocentric or egocent ric. Sociocentric or whole network perspectives focus holistically on an entire population defined by a meaningful boundary. These boundaries range from students within a school, prisoners within a jail or citizens within a country. Alternatively, egoce ntric perspectives focus on a particular person (called an ego ), capturing who they know (called an alter ) and then how these people know each other. Both perspectives consider people within the network as nodes and the relationships between them ties Graph and network theory provide a concrete mathematical

PAGE 15

15 foundation for the analysis of both types of networks, allowing for the calculation of a set of socially relevant and richly informative metrics. Metrics from egocentric networks can be conceptualiz ed as individual difference variables that arise as a combination of both evolved and learned psychological traits along with environmental and ecological pressures (Crosier, Webster & Dillon, 2012). The metrics degree number of isolates, density broker age and centrality are highly pertinent to the proposed research due to their potential causal antecedents and the outcomes they are hypothesized to be tied to. Degree the number of people that they are t ied to (Wasserman & Faust, 1994) The size of a social network has been linked to many psychological and physical health outcomes (Berkman, Glass, Brissette, & Seeman, 2000; Bowling & Browne, 1991; DeLongis, Folkman, & Lazarus, 1988; Pressman et al., 2005; Shumaker & Hill, 1991; Teresa E., 1996) A social isolate is a node that remains completely unconnected to others (except for the ego in an egocentric network). In a sociocentric network, this individual is disconnected from all others. Social isolation poses a set of highly negative risks, both in terms of me ntal and physical health. Density captures the ratio of existing relationships in a network to the number of possible relationships (Wasserman & Faust, 1994) In conjunction with a large enough de gree, density is strongly related to social support. Inversely related to density, both brokerage and centrality are measures of social status, power and influence within a social network (Wasserman & Faust, 1994)

PAGE 16

16 Brokerage captures the extent to which an ego ties disparate groups of people. Consider a person that works the ticket booth of a popular musical venue. This individual links the music and arts community to their friends, strongly encou raging these acquaintances to try to get the limited resource of tickets popular events. Alters must go through this broker, dependent upon them for access. A broker occupies a special position in a network, one that conveys social power. The same is tru e for centrality for which there are multiple metrics (e.g., betweenness, closeness and eigenvector; all capture approximately the same concept yet do so through different computational techniques). Instead of connecting different clusters of people, hig h centrality individuals lie at the focus of a cluster. Alters need to go through this person to access others within this cluster. A secretary is a perfect illustration of this. While the position is not as prestigious as CEO, secretaries set the tone of the workplace because they connect most employees, controlling the flow of information (Grosser, Kidwell, Labianca, & Ellwardt, 2010) Metrics that cap ture power are inversely related to density because dense networks provide multiple pathways to get what one wants or needs if a particular person is not willing to help, one can simply go to the next. This is not necessarily a negative. While the person may have less influence on the network, the social context that they are embedded in provides redundant safety nets should trouble arise. Like all physical systems, human social networks are bound by causality. Network structure is not spontaneously c reated or shaped solely by environmental factors. It is at least partially caused by the psychological characteristics of the actors in the network and the environment in which the network is embedded. The exploration of this

PAGE 17

17 hypothesis is the central ai m of this thesis. While the environment can alter network topology (e.g., varying resource levels can increase or decrease population size, geography can impede the exchange of information and resources), the internal traits of the actors should be a power ful sculptor of the resource and information pathways between the nodes of a human network. Of central importance is personality because it way that a person approa ches the world and the way the world affects that person. Our behavioral tendencies and the reaction of others to these inclinations determine patterns of social interaction the arrays of iterated interactions that define human connectedness. My hypoth esis is that personality shapes social networks. Social Network Analysis and Psychology: A Brief History Despite that psychology has been slow to adopt social network analysis, the two have a long, albeit sparse, history. The eminent social psychologist S tanley Milgram was one of the first psychologists to conduct a true social network investigation. small world experiment (1967) sought to establish the average path length between any two randomly selected individuals in a population. Us ing a simple design, Milgram demonstrated that through an average of 5.5 of intermediary people geographically or socially separated they may be. Sociologists were the first to formalize the method of social network analysis proper ly by utilizing mathematics from network theory, which is a specialized application of graph theory. Graph theory provides a framework by which to calculate a collection of powerful struct ural metrics based on a set of connected dyads. It need not matter what the dyads are: switchboards in a telecommunication system, computer routers,

PAGE 18

18 lions on the savannah or humans are all fair game, as long as they are link in some way or engage in some exchange of information. The method allows researchers to extract numerous concrete descriptions of how individuals are connected, along with providing an array of convincing visualizations. The crucial importance of social network analysis grows every da y. The growth of global telecommunications networks makes it all but necessary to adopt this methodology in our modern hyper connected age (Crosier, Webster & Dillon, 2012) to explore new questions within psychology. The World Wide Web has provided a fert ile platform on which electronic social networks can grow, as made apparent by the fact that Facebook now accounts for the second largest amount of traffic and market share on the internet, outpaced only by Google (Facebook, 2011). While different in natu re, online interaction encompass a great deal of day to day information and resource exchange. This is especially true when considering that 1 in 13 people have a Facebook profile, spending an average of approximately one hour each day using it (Facebook, 2011). Network function and structure is fundamental; while there are surely differences between traditional and online networks, they work and can be investigated in the same way. If we are to understand the meteoric rise of the online social network a nd its consequences (e.g., inciting revolutions, staying connected with family and friends and knowing the exact second when Johnny and Sally make their relationship official), it is first necessary to understand the function of organic, real world social networks from the viewpoint of psychology. Most literature searches regarding personality and social network structure return

PAGE 19

19 cial network analysis (e.g. Casciaro, 1998) A vast majority of current investigations do not look at structure beyond degree. There are many structural metrics available and few are explored in relat ion to personality. Personality was shown to vary with the presence of structural holes (Burt, Jannotta, & Mahoney, 1998) and with network closure (Kalish & Robins, 2006) The link between personality disorders and network metrics has also been superficially explored (Clifton, Turkheimer, & Oltmanns, 2009) Although not framed in personality, propensity to connect with others (PCO) was shown to be related to network size (Totterdell, Holman, & Hukin, 2008) Presumably PCO would possess a great deal of overlapping variance with extraversion. Encouragingly, a single (locatable) paper used the Big Five in conjunction with true social network analysis (Wehrli, 20 08), finding links between the Big Five and network growth. However, the scope of the paper was limited to growth and not the intricacies of structure. Cognitive Social Structures There are multiple ways to define a network. Often, the most convenient wa y to do this is based on a cognitive recall task that has a person list the members of their network then describe the relationships between them (Krackhardt, 1987) This method has its advantages because it capt ures how one cognitively defines the social world around them. The relationships that they detail in such a task are representative of their conception of their social space (Krackhardt, 1987). These methods are easy to implement in the field and have had success across many scientific disciplines. However, ample success is not indicative of a lack of problems. How much does social network? As it turns out, recall can be inaccurate. Beyond the failings of memory

PAGE 20

20 and personal biases, it is important to consider that a large amount of social activity is clandestine. Sometimes the relationships that define us are purposely hidden, both to ourselves and to others (Lehmiller, 2009; Wegner Lane, & Dimitri, 1994) Electronic Social Structures Online social networks can be based off email communication, telephone use, participation in online forums and so on (Harasim, 1993). Notably important are the online venues created explicitly for n etworked connectivity. While multiple options exist, Facebook is the current juggernaut of social networking. Facebook data can be tapped to build egocentric or sociocentric networks. It is ubiquitous across demographic groups and cultures, used by memb ers of most every nation in the world. The average user, of which there are 800 million has 130 friends. Half of these individuals log in every day, interacting with more than 900 million online objects, connected to an average of 80. This massive commu nity uploads more than 250 million photos per day communicating in over 70 different languages. Over 350 million of these active users connect through their mobile devices These powerful statistics point to the true scale of human connectivity Facebook takes advantage of our innate propensity to connect (Crosier, Webster & Dillon, 2012). The same rules apply to online and real world social networks. Who we are contributes to our behavior, even if it is gargantuan size. This network is shaped by the personalities of its constituents, just as the network of a secluded Amazonian tribe is shaped by the psychological traits of its members.

PAGE 21

21 The Problem of Technological Literacy Technological literacy (TL) presents a unique problem to research on social networks and to technology related s ocial science as a whole. Unlike cognitive social struct ures, nodes and ties are behaviorally recorded. However, in the case of electronic social networks, TL is a potentially major confound. The ability to use technology, especially computers, tablets, and smart phones, is a huge factor in determining the muted. Massive po tential methodological shortcomings due to generational differences in the ability to use technology force the use of age as a statistical control. Age and TL have been linked in terms of decreasing comfort, efficacy and control with an increase in age (Czaja & Sharit, 1998) People who have grown up with it and rely on it, while those who saw the rise of the home computer later in their lives may ultimately shun its novelty. Therefore, it is necessary to attempt to account for individual differences in TL.

PAGE 22

22 CHAPTER 2 STUDY 1 Purpose Study 1 seeks to establish a link between components of the FFM and egocentric social network structure as reported by participants. Personality influences how we relate to others. These relationships, in turn, form our social networks. Networks should be imbued with a dispositional signature reflective of how people behave on a day to day basis. Extraversion is hypothesized to have special importance because it is directly social and captures the extent to which one draws energy for interpersonal relationships. Increased extraversion should lead to more tie formation and maintenance. Outside of extraversion based hypotheses, this study is largely exploratory in nature Procedure Data was collected from the University of Florida (UF) student population with a For these participants, each was awarded one research credit point for volunteering. Others received no compensation. After arriving at the lab space, participants completed two online tasks guided by a research assistant. First, they completed a battery of survey measures and finished with an EgoNet interview. The data streams were associated by a unique and anonymous participant created identifier. Participants were given the choice to view a graph of their network if they desired. Participants This largely undergraduate sample ( N = 185) ranged from 18 to 55 years of a ge ( M = 20.17), and was comprised of 52 males (28.1 %), 132 females (71.4), and one

PAGE 23

23 person who reported being transgender (0.5 %). Unfortunately, the transgender individual was dropped from the analysis because of insufficient statistical power. The samp le was 8.8% Black ( n = 19), 63% White ( n = 117), 10.2 % Hispanic ( n = 22), 4.9 % other ( n = 9), 2.2 % Filipino ( n = 4), 0.5 % Alaska Native ( n = 1), 0.5 % American Indian ( n = 1), and 0.5 % Chinese ( n = 1). Measures The battery consisted of the 44 item Big Five Inventory (John, Donahue & Kentle, 1991), the 9 item Sociosexuality Inventory Revised (Penke, 2010), the 20 item Mini K component of the Arizona Life History Battery (Figueredo et al., 2006), a single item self esteem scale (SISE; Robins, Hendin, & T rzesniewski, 2001), a single item narcissism scale (Konrath, Meier & Bushman, under review), the 12 item dark triad scale (Jonason & Webster, 2010), the 10 item Positive Affect/Negative Affect schedule (PANAS; Thompson, 2007), a 5 item measure of subjectiv e well being (Diener, 1985), and the 17 item Emotional Expressivity Scale (Kring, Smith & Neale, 1994). There were 120 items in total. Only the Mini IPIP was used in the following analyses. The Mini IPIP was on average reliable. All five personality fa ctors showed acceptable reliability, especially considering that each trait had only four items and that alpha is confounded 6. The EgoNet interview captured egocentric social networks as defined by means of list 35 people you know. Knowing means you know them by sight or by name, you have ha d some contact with them over the past two years and you could contact them if you

PAGE 24

24 Person A a nd Person B talk to each other independently of you? If they would only talk in your presence then your response should be Not at all likely If you know that they talk when you are not around them then your answer should be Very likely If it is possib le, but you are not sure then your answer should be Maybe closeness centrality, betweenness centrality and number of isolates were computed and merged with the personality data. Results and Discussion A series of 25 hierar chical multiple regressions were run to evaluate the hypothesized relationship between personality and egocentric network structure. Beginning with openness, each metric (degree centrality, closeness centrality, betweenness centrality and isolates) was ru n with age and gender controls in the first step. Only agreeableness yielded a significant finding, with the trait being negatively related to number of isolates, b = .15, t = 2.07, p < .05. The more agreeable one was the few isolates that they had in their network. Age and gender were controlled for in this analysis with only gender being important, b = .15, t = 2.07, p < .05. Males tended to have less isolates than females. Due to the atheoretical nature of this exploratory set of analyses and th e high probability of a Type 1 error in a series of 25 regressions, the meaningfulness of this finding is questionable in light of the fact that it is difficult to explain this relationship, even after the fact.

PAGE 25

25 CHAPTER 3 STUDY 2 Purpose Study 2 has an identical aim to study 1, yet seeks to explore the personality structure link in a large geographically diverse online sample with a broader age range. Although behavior varies appropriate to context (Mischel & Shoda, 2008) online social networks can collect trait data in a variety of contexts, much like cognitive social structures. Most social domains (e.g. family, work and friends) should be adequately represented on Facebook. If personality could be predictive of tradi tional social structures, it is reasonable to assume the same is true for electronic social structures. Procedure Run by David Stillwell and Michal Kosinski at the University of Cambridge, my Personality is a highly collaborative web based project that ai massive potential as a data collection tool. The initial goal was to vet the psychometric properties of the International Personality Item Pool (IPIP NEO) personality measure. The project has expanded its scope to include 25 survey i nstruments and a trove of Facebook user data. To date, over 5 million people have participated. Participants grant consent through a Facebook application and then complete a set of randomly distributed measures. Registered collaborators (approved by the principle investigators) Network metrics are then computed with Statnet in R and merged with the main dataset, which includes all measures along with rich demographic data.

PAGE 26

26 Participants Participants ( N = 9,690) for whom network metrics had been calculated ranged from 18 to 60 years ( M = 24.68,) and hailed from 86 countries. Of these, 3,592 were female (53.6 %) and 3,106 were male (46.4 %), with 2,992 people not making th ese data accessible on Facebook (demographic information was culled from the site directly and not in survey form). Participants were compensated by getting instantaneous feedback on their personality scores. This is common outside of research endeavors with a multitude of facebook users seeking out similar activities for enjoyment. Measures The IPIP NEO Short Form used in the my Personality project is a 100 item Big conscientiousn the extracted egocentric networks. All network metrics were successfully log transformed to achi eve normality. Age and sex were used as controls and were entered in the first step of all regression analyses. Results and Discussion In the following analyses, statistical significance is not informative in light of the large sample siz e. Although age, extraversion and conscientiousness reach ed signific ance, only age and extraversion displayed meaningful effect size s. E xtraversion significantly predicted network degree, after controlling for age The more extraverted one was, the larger degree there ne twork tended to be, with young er people tending to have bigger networks (see Table 1).

PAGE 27

27 Table 3 1 Personality predicting degree Predictor b t R 2 Age .34 19.37 *** .12 Gender .00 .01 .00 Openness .03 1.56 .00 Conscientiousness .05 2.8 ** .00 Extraversion .25 13.4 *** .06 Agreeableness .03 1.6 .00 Neuroticism .01 .54 .00 Significant to the p < .05 level Significant to the p < .01 level ** S ignificant to the p < .001 level Looking at the link between personality and betweenness centrality, extraversion emerged a s the only noteworthy predictor. Again, regardless of the network metric that was being used as the dependent variable, it proved useful to control for age, as younger people scored higher on this metric as well. The mor e extraverted a person was the increasing amount of social influence and power they tended to have (see Table 2). Table 3 2 P ersonality predicting betweenness centrality. Predictor b t R 2 Age .33 18.93 *** 0.11 Gender .01 0.43 0.00 Openness .03 1.97 0.00 Conscientiousness .05 2.59 ** 0.00 Extraversion .25 13.72 ** 0.06 Agreeableness .03 1.75 0.00 Neuroticism .08 0.34 0.00 Significant to the p < .05 level Significant to the p < .01 level ** S ignificant to the p < .001 level

PAGE 28

28 Extraversion displayed an inverse relationship with density. Highly dense networks typically belonged to introverted peop le. Although agreeableness, openness and gender also contribu ted, age was again a strong player in fostering density, with older indi viduals tending to have more dense networks (see Table 3). Table 3 3 P ersonality predicting density Predictor b t R 2 Age .24 13.25 *** 0.06 Gender .07 3.5 *** 0.00 Openness .06 3.47 ** 0.00 Conscientiousness .02 1.19 0.00 Extraversion .28 14.42 *** 0.07 Agreeableness .04 2.1 0.00 Neuroticism .04 1.76 0.00 Significant to the p < .05 level Significant to the p < .01 level ** S ignificant to the p < .001 level Finally, brokerage was scu lpted by extraversion, even after considering the potent control of age Extraverts also convey social power by linking together disparate groups, and by controlling the flow of information and resources among them (see Table 4). Table 3 4 P ersonality predicting brokerage Predictor b t R 2 Age .3 4 19.21 *** 0.11 Gender .00 0.13 0.00 Openness .03 1.64 0.00 Conscientiousness .05 2.76 ** 0.00 Extraversion .25 13.49 *** 0.06 Agreeableness .03 1.62 0.00 Neuroticism .01 0.467 0.00 Significant to the p < .05 level Significant to the p < .01 level ** S ignificant to the p < .001 level

PAGE 29

2 9 Discussion Within online social networks, extraversion is the most influential of the Big Five components in shaping network topology. This study suggests that extraversio n is a reliable force in tie forma tion and maintenance, at least in online networks Due to the effects of technological competence, age emerged as an important control. Those in younger generations grew up in a digital world much different than their parents or grandparents. Electronic media has become a primary way to connect, and those who are less comfortable, knowledgeable, or skilled with technology do not use it in the same way. This difference leaves its signature in network structures. In these analyses, extraversion and age together accounted for between 13% and 18% of the variance in structural metrics. This suggests that a vast majority, approximately 85%, of the variance is left unexplained. However, considering all of the involved complexity in the millions of relations hips that these thousands of people had, just two variables explaining 15% of the variance in structure is encouraging.

PAGE 30

30 CHAPTER 4 STUDY 3A Purpose This two part study attempts to replicate the extraversion based findings attained in the Study 2 in anoth er large and diverse sample that had important qualitative distinctions. Instead of online networks, in school networks of adolescents were assessed. Despite the range restriction of age in this sample, Add Health (described below) covered every geograph ical and social base, drawing data from students in every part of the country. Participants were asked many questions pertaining to distributed was not framed in terms of t he FFM. Thankfully, many extraversion like questions were asked. First, I sought to create a post hoc measure of the Big Five with existing items by testing a candidate measure in an online study against existing and highly used measures. Procedure The National Longitudinal Study of Adolescent Health (Add Health) began in 1 994 and is currently in its fif th wave of data collection (four waves are currently available for analysis amassing over 90,000 participants in a selection of all American high schools. This ambitious project sought to identify current health problems along with contributing social factors. Schools were combined into 80 geographical clusters and were included if they had at least 30 students atten ding. During the first wave of the project, students who participated completed lengthy in home and in school questionnaires. From the large number of existing items, a post hoc Big Five measure was constructed. The resulting extraversion scale (see app endix) was comprised of 10 items, including the

PAGE 31

31 I The post hoc measure was distributed on Indicator (BFI) (John, Naumann, & Soto, 2008) the Ten Item Personality Inventory (TIPI) (Gosling, Rentfrow, & Swann, 2003) and the Mini International Personality Item Pool (Mini IPIP) (Donnellan, Oswald, Baird, & Lucas, 2006) P articipants Participants ( N = 400) were recruited on Mechanical Turk and were compensated US $0.30 for completing the study. Age ( M = 30.59) ranged from 18 to 79 years. The sample was split evenly between the men (52.1%) and women (47.9%), and was compri sed mostly of White and Asian Indian participants. Results and Discussion The 10 item Add Health Post Hoc Extraversion Scale (AddPHES, see appendix with existing popular measures of extraversion, AddPHES and the BFI r = 0.64, p < .01, AddPHE S and the TIPI r = 0.50, p < .01 and AddPHES with the Mini IPIP r = 0.47, p < .01. These results suggest that the post hoc measure is tapping extraversion. These encouraging results not only allow for confidence to be placed in Study 3b, but provide addi tional avenues for exploration within the Add Health dataset.

PAGE 32

32 CHAPTER 5 STUDY 3B Purpose Using the scale created in Study 3A, the present study has two goals. First, it sought to replicate the extraversion centrality link established in Study 2. Se cond, taking advantages of the benefits that multilevel modeling provides, it aimed to investigate the source of the variance in network centrality. It is hypothesized that variance will be due to level 1 units (people) instead of level 2 units (schools). Participants There were 4,379 participants and 72 clusters of schools included in the analysis. Students were in grades 7 12 (middle and high school) with 51.6% of these were female. Results and Discussion Within Add Health the AddPHES was reliable = .90 Two multilevel models were run. First a random ANOVA model was used to obtain the interclass correlation coefficient (ICC). The model is as follows ( CENT corres ponds to Bonacich centrality): Level 1 Model: CENT ij = 0j + r ij Level 2 Model: 0j = 00 + u 0j Combined Model: CENT ij = 00 + u 0j + r ij

PAGE 33

33 Neither the fixed effects nor variance components produced any significant results. The overall grand mean for the sample was 0j = 0.0001, due to the prior standardization within level 1 units of bo th the outcome and predictor variables. Both the between school 00 2 = .99956 were used to compute the intraclass correlation, ICC = 0.0005. Therefore, because only an absolutely miniscule amount and ro unded conservatively to 0%, almost no variance was due to level 2 units. This suggests that a means as outcomes model was unnecessary. A small ICC confirms the hypothesis that individual differences contribute more to network structure than local group le vel phenomena. Next, to test the main hypothesis, a random regression coefficients model using group mean centering was implemented. The model was as follows ( CENT and EXTRA refer to Bonacich centrality and extraversion): Level 1 Model: CENT ij = 0j + 1j *( EXTRA ij ) + r ij Level 2 Model: 0j = 00 + u 0j 1j = 10 Combined Model: CENT ij = 00 + 10 *( EXTRA ij ) + u 0j + r ij Extraversion predicted a small amount of variance in centrality, with the average relationship across all schools being 10 = 0.39, t (4324) = 12.116, p < .001, r 2 = .03. The higher on the trait of extraversion students were, the more central they tended t o be in

PAGE 34

34 their network, resulting in increased social power. The within cluster variance after controlling for extraversion was 2 = 0.96. The average of the cluster specific intercepts 00 = .0002, which again was due to prior standardization. Consi dering the random 10 2 (71) = 78.49, p = 0.253. In sum, the extraversion centrality relationship was replicated. This relationship could be attenuated for an important re ason. The constructed measure did not load perfectly on to existing measures. This new measure captures extraversion, although not perfectly.

PAGE 35

35 CHAPTER 6 GENERAL DISCUSSION The present studies offer encouraging avenues for future exploration of how psychological traits shape the structure of social networks. Study 1 did not yield any meaningful results while Study 2 suggests that extraversion is a driving force behind the sh ape of electronic social worlds. Study 3 offered a modest replication of Study 2, along with providing a new successful measure of extraversion and illustrating that variance in network structure, at least in centrality, is most likely due to individual d ifferences and not local culture. Why is extraversion so important over the four other traits, especially in terms of network centrality? Centrality was captured in four different ways in the present studies: degree, closeness, betweenness and by Bonaci exhaustive list and alternatives exist. Although each employs a different computational tactic, they all are a measure of social power, the importance of the individual within that network. This trait captures the extent to which people draw positive emotions from social interaction thus, extraverts tend to know most everyone in their social group. They often lie on the paths between others, meaning that people need them to be connected to others. This is the very defini tion of social power. Another way power is conferred is brokerage. Instead of commanding influence by being at the very center of a network, a broker connects separate clusters of people. Extraversion drives brokerage because it coaxes individuals to befriend individuals from different social spheres, making it possible to become a middleman between groups that exchange information and resources. Brokers are the gatekeepers of these intergroup relationships and reap the benefits in terms of influence.

PAGE 36

36 Further, why should individual differences play a predominant role in network topology, surpassing the role of environmental and ecological influences? Simply, they came first. A baby is born with a strong propensity to be extraverted or not. A lifetime of being magnetically drawn or repulsed to social interaction should overpower local cultural differences. Only rare and severe situations should alter these predispositions in substantial ways (e.g. remarkably strict parents or as seen in cases of feral children). In Study 3, a primary limitation is that level 2 units are defined as schools or geographically close groups of schools. This may be too fine and narrow of a distinction. Although this level of grouping would be adequate to capture differenc es in phenomena like the dissimilarity between high and low socioeconomic areas, between southern and northern American culture or the differences between red and blue states, it is definitely insufficient to capture everything. It would be wise to pursue this on a more global level and look at East versus Western differences (i.e. individualistic versus collectivistic). It would be reasonable to hypothesize that collectivistic cultures promote smaller, denser networks with a focus on the family. Add Hea lth offers a priceless asset in the multilevel data it makes freely available for public analysis. The next step is to conduct increasingly comprehensive studies that explore cultural differences in the way that we socialize under the guiding principle th at patterns of socialization are reflected in network structure. Study 2 was the most successful in that age, which captures technological competency, and extraversion ate up a respectable chunk of variance in structural metrics. Study 3 suggests that mos t of the variance in these metrics exists at the level of the person and not in local culture. So what variables encompass the remaining

PAGE 37

37 variance? This is a question that social and personality psychologists are well equipped to pursue. First, scientist s exploring this question have to identify the candidate variables that determine how we approach our social lives, while recognizing that sociality is complex and highly componential. Mischel is changing the face of personality psychology by addressing the context specificity of traits. In a networks specific manner, the social domains (e.g. family, romantic, work, etc.) have to be delineated and drawn apart. Only when they have been separated can the unique and shared factors that shape them be illumi nated. What is the best way to represent a social network? Is there one true social space? How might cognitive and electronic social spaces be related to true social space? To begin to answer these questions, it will first be helpful to understand the an tecedents of these social spaces, along with their relationships. Currently, it is impossible to capture true social space. This would necessitate a technology that monitored every social interaction of a person, along with the relationships between thes e interactants, requiring a device that captured detailed information of every communication online and face to face. While currently possible with an existing device like the iPhone, this would still require a feat of engineering. Of course, long befor e online platforms like Facebook, people were forming social networks. This has been happening since the first social organism (Crosier, Webster & Dillon, 2012 ). Once one becomes a member of a social network service, they usually set out to replicate the friends, eventually branching off to the kid that they sat across the room from in second

PAGE 38

38 grade (the conscientious extraverts, at least). Thus, our traditional social structures partly determine our online structures. More recently, our online structures help connect us to new people, influencing social connectivity. This allows us to reconnect with individuals that may have fallen out of touch because of physical limitations, like livi ng on the other side of the country. Further, venues like Facebook have expended a tremendous amount of resources on developing algorithms to connect people based on shared affiliations, interests and mutual acquaintances. This encourages people who woul d never otherwise meet to become connected, forming a tie that would not have been developed only a decade ago. This tie has the potential to become a bona fide social connection, even if the two involved individuals never meet. Our increasingly digital world provides an astonishing opportunity to collect data. This does not mean the modern social scientist needs to generate data they simply need to mine it. Web servers are the new laboratory, and despite the potential limits of such methods such as range restrictions in age and limitations regarding socioeconomic status, these capabilities mark a transition into a new era of social research across all disciplines. The hypothesized link betwee n online and cognitive social spaces hint at the rapidly closing gap between these domains and it is no longer accurate or fair to downplay the importance of online life. Online sociality is a valid, important and now crucial domain of the global lifestyl e. Online networks provide yet another way to measure network structure. Methods for capturing network structure should not be considered to be in competition. Due to current technological hurdles, it is impossible to arrive at one true social structu re. In this light, metrics derived from different

PAGE 39

39 methodological techniques (e.g. cognitive and electronic social structures) can be conceptualized as manifest indicators of latent social structures. Much as personality cannot directly be measured, curre nt ways to capture network structure are approximations, each with strengths and weaknesses. Mixed method multiplex networks and structural equation modeling are two approaches that are well suited to explore the problem of true social structure. Only wh en this dilemma has been addressed can the complex trait antecedent structure relationship be understood.

PAGE 40

40 APPENDIX ADDPHES ITEMS 1. I have a lot of energy. 2. I like myself just the way I am. 3. I feel socially accepted. 4. During the past week, how many times did you just hang out with friends? 5. You feel close to people at your school or work. 6. You are happy. 7. You like yourself just the way you are. 8. You feel like you are doing everything just about right. 9. You feel that your friends care about you. 10. Do you know most of the people in your neighborhood?

PAGE 41

41 LIS T OF REFERENCES Berkman, L. F., Glass, T., Brissette, I., & Seeman, T. E. (2000). From social integration to health: Durkheim in the new millennium. Social Science & Medicine 51 (6), 843 857. doi:10.1016/S0277 9536(00)00065 4 Bernard, H. R., Killworth, P., Kronenfeld, D., & Sailer, L. (1984). The Problem of Informant Accuracy: The Validity of Retrospective Data. Annual Review of Anthropology 13 495 517. Bowling, A., & Browne, P. D. (1991 ). Social Networks, Health, and Emotional Well being Among the Oldest Old in London. Journal of Gerontology 46 (1), S20 S32. doi:10.1093/geronj/46.1.S20 Burt, R. S., Jannotta, J. E., & Mahoney, J. T. (1998). Personality correlates of structural holes. Soc ial Networks 20 (1), 63 87. doi:doi: DOI: 10.1016/S0378 8733(97)00005 1 Buss, D. (n.d.). The evolution of personality and individual differences New York: Oxford University Press. Casciaro, T. (1998). Seeing things clearly: social structure, personality, and accuracy in social network perception. Social Networks 20 (4), 331 351. doi:doi: DOI: 10.1016/S0378 8733(98)00008 2 Clifton, A., Turkheimer, E., & Oltmanns, T. F. (2009). Person ality disorder in social networks: Network position as a marker of interpersonal dysfunction. Social Networks 31 (1), 26 32. doi:doi: DOI: 10.1016/j.socnet.2008.08.003 Czaja, S. J., & Sharit, J. (1998). Age Differences in Attitudes Toward Computers. The Jo urnals of Gerontology Series B: Psychological Sciences and Social Sciences 53B (5), P329 P340. doi:10.1093/geronb/53B.5.P329 DeLongis, A., Folkman, S., & Lazarus, R. S. (1988). The impact of daily stress on health and mood: Psychological and social resour ces as mediators. Journal of Personality and Social Psychology 54 (3), 486 495. doi:10.1037/0022 3514.54.3.486 Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The Mini IPIP Scales: Tiny yet effective measures of the Big Five Factors o f Personality. Psychological Assessment 18 (2), 192 203. Facebook Statistics. (2011). Retrieved July 1, 2011, from http://www.facebook.com/press/info.php?statistics Gosling, S. D., Rentfrow, P. J., & Swann, W. B. (2003). A very brief measure of the Big F ive personality domains. Journal of Research in Personality 37 (6), 504 528. doi:10.1016/S0092 6566(03)00046 1

PAGE 42

42 Grosser, T., Kidwell, V., Labianca, G., & Ellwardt, l. (2010). Hearing It Through the Grapevine: Positive and Negative Workplace Gossip. Organiz ational Dynamics Jang, K. L., Livesley, W. J., & Vemon, P. A. (1996). Heritability of the Big Five Personality Dimensions and Their Facets: A Twin Study. Journal of Personality 64 (3), 577 592. John, O. P., Naumann, L. P., & Soto, C. J. (2008). Paradigm s hift to the integrative Big Five trait taxonomy: History, measurement, and conceptual issues. Handbook of personality: Theory and research (3rd ed.). (pp. 114 158). New York, NY, US: Guilford Press. Kalish, Y., & Robins, G. (2006). Psychological predispo sitions and network structure: The relationship between individual predispositions, structural holes and network closure. Social Networks 28 (1), 56 84. doi:doi: DOI: 10.1016/j.socnet.2005.04.004 Krackhardt, D. (1987). Cognitive social structures. Social N etworks 9 (2), 109 134. doi:10.1016/0378 8733(87)90009 8 Lehmiller, J. J. (2009). Secret Romantic Relationships: Consequences for Personal and Relational Well Being. Personality and Social Psychology Bulletin 35 (11), 1452 1466. doi:10.1177/01461672093425 94 Mischel, W., & Shoda, Y. (2008). Toward a unified theory of personality: Integrating dispositions and processing dynamics within the cognitive affective processing system. Handbook of personality: Theory and research (3rd ed.). (pp. 208 241). New York, NY, US: Guilford Press. Pervin, L. (1999). (2nd ed.). New York: Guilford Press. Pressman, S. D., Cohen, S., Miller, G. E., Barkin, A., Rabin, B. S., & Treanor, J. J. (2005). Loneliness, Social Network Size, a nd Immune Response to Influenza Vaccination in College Freshmen. Health Psychology 24 (3), 297 306. doi:10.1037/0278 6133.24.3.297 Shumaker, S. A., & Hill, D. R. (1991). Gender differences in social support and physical health. Health Psychology 10 (2), 10 2 111. doi:10.1037/0278 6133.10.2.102 Soldz, S., & Vaillant, G. E. (1999). The Big Five Personality Traits and the Life Course: A 45 Year Longitudinal Study,. Journal of Research in Personality 33 (2), 208 232. doi:doi: DOI: 10.1006/jrpe.1999.2243 Teresa E ., S. (1996). Social ties and health: The benefits of social integration. Annals of Epidemiology 6 (5), 442 451. doi:10.1016/S1047 2797(96)00095 6

PAGE 43

43 Totterdell, P., Holman, D., & Hukin, A. (2008). Social networkers: Measuring and examining individual differe nces in propensity to connect with others. Social Networks 30 (4), 283 296. doi:doi: DOI: 10.1016/j.socnet.2008.04.003 Wasserman, S., & Faust, K. (1994). Social Network Analysis: Methods and Applications Cambridge: Cambridge University Press. Wegner, D. M ., Lane, J. D., & Dimitri, S. (1994). The allure of secret relationships. Journal of Personality and Social Psychology 66 (2), 287 300. doi:10.1037/0022 3514.66.2.287

PAGE 44

44 BIOGRAPHICAL SKETCH Ben Crosier is a Florida transplant, growing up in upstate New York. Ben completed a B achelor of A rts in p syc hology from SUNY Plattsburgh, a M aster of A rts in p sychology from SUNY New Paltz and is working on his Master of Science and Doctor of Philosophy in social p sychology at the University of Florida. Working under Dr. Gregory Webster, most of his research focuses on the structure of social networks.