You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/315298440

Connected to Create: A Social Network Analysis of Friendship Ties and


Creativity

Article  in  Psychology of Aesthetics Creativity and the Arts · August 2017


DOI: 10.1037/aca0000117

CITATIONS READS
21 3,021

3 authors:

Alexander S McKay Pawel Grygiel


Virginia Commonwealth University Jagiellonian University
23 PUBLICATIONS   239 CITATIONS    87 PUBLICATIONS   713 CITATIONS   

SEE PROFILE SEE PROFILE

Maciej Karwowski
University of Wroclaw
215 PUBLICATIONS   5,844 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

critical thinking and creativity View project

Forma kształcenia a trajektorie rozwoju teorii umysłu View project

All content following this page was uploaded by Alexander S McKay on 20 March 2017.

The user has requested enhancement of the downloaded file.


Psychology of Aesthetics, Creativity, and the Arts © 2017 American Psychological Association
2017, Vol. 0, No. 999, 000 1931-3896/17/$12.00 http://dx.doi.org/10.1037/aca0000117

Connected to Create: A Social Network Analysis of


Friendship Ties and Creativity

Alexander S. McKay Paweł Grygiel


The Pennsylvania State University Educational Research Institute

Maciej Karwowski
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

The Maria Grzegorzewska University


This document is copyrighted by the American Psychological Association or one of its allied publishers.

The purpose of this article was to examine the relationship between friendship ties and creativity. Based
on the homophily hypothesis, we predicted friendship ties would more likely occur between people
similar to each other on creativity-related attributes. We also predicted students would be more likely to
report friendship ties with peers who have higher creativity scores in general. Across a pilot and primary
study, we examined the relationship between friendship strength among high school students in a pilot
study (Study 1) and friendship nominations among elementary school students in a primary study (Study
2) with creativity. In Study 2, but not Study 1, we found that friendship nominations were more likely
to occur when scores on a creativity task were similar. In both studies, we found that popularity was
positively related to originality (Study 1) and creativity (Study 2). The results indicate that elementary
school students nominated peers as friends who are similar to them when it comes to creativity and that
there is a positive relationship between popularity and creativity.

Keywords: creativity, social networks, friendship, homophily, similarity-attraction

Supplemental materials: http://dx.doi.org/10.1037/aca0000117.supp

Although layperson stereotypes align with the idea that “cre- Creativity research has a long and fruitful history of studying the
ative people” are “mad geniuses” with mental disorders, working relationship between creativity with individual differences (e.g.,
in isolation on radical new discoveries (Kaufman, Bromley, & personality; Feist, 2010) and contextual factors (e.g., characteris-
Cole, 2006; Simonton, 2014), this is often not the case (Sawyer, tics of the workplace; Amabile, Conti, Coon, Lazenby, & Herron,
2012). Rather, many inventors and radical creators like Bill Gates, 1996). There have, however, been fewer studies examining the role
Jane Austen, Ada Lovelace, Steve Jobs, Frida Kahlo, and Mark of social connections with creativity in everyday settings (for
Zuckerberg were or are well connected to others to develop and exceptions see Kéri, 2011; Lau & Li, 1996; Simonton, 1984). This
implement their ideas within their disciplines (Sawyer, 2012). lack of research is unfortunate because studies examining the role
These social connections allow them to gain new information, of social connections, primarily in the organizational literature,
identify others who they can work, and gain “buy in” to implement have found that they are important drivers of individual creativity
their ideas. (Baer, Evans, Oldham, & Boasso, 2015). Indeed, as Simonton
(1984) argued, “A successful ‘social psychology of creativity’
demands that the creative individual be placed within a network of
interpersonal relationships and group influences” (p. 1273). Thus,
examining social relations is important to identify the full range of
Alexander S. McKay, Department of Psychology, The Pennsylvania antecedents for creativity.
State University; Paweł Grygiel, Department of Educational Measurement, Over the past decade, researchers in the organizational sci-
Educational Research Institute; Maciej Karwowski, Creative Education ences have begun to use social network methods to examine the
Lab, Department of Educational Sciences, The Maria Grzegorzewska Uni- relationship between social connections and creativity (e.g.,
versity. Cattani & Ferriani, 2008; Perry-Smith, 2006; Perry-Smith &
Study 1 was supported by a grant funded to Maciej Karwowski by Polish Shalley, 2003). These studies have primarily examined the role
Ministry of Science and Higher Education (Iuventus Plus Program -
of information sharing ties and have rarely examined how other
0193/IP3/2015/73). We thank Diane Felmlee and Anne DeLessio-Parson
for their helpful feedback on an earlier version of this article.
informal ties like friendship are related to creativity. Baer et al.
Correspondence concerning this article should be addressed to Alexan- (2015) “encourage future research to more systematically con-
der S. McKay, Department of Psychology, 142 Moore Building, Pennsyl- sider the role that tie type and the content conveyed through ties
vania State University, University Park, PA 16802. E-mail: asm273@ may play in determining the effects of certain network features
psu.edu on innovation” (p. 215).

1
2 MCKAY, GRYGIEL, AND KARWOWSKI

In the current study, we expand on previous network research Jouyner, 2004; McPherson et al., 2001). These relationships form
and examine a different type of tie with regard to creativity in through various mechanisms like feelings of trust, ease of com-
everyday settings. Specifically, we examine how friendship ties munication, beliefs that friendship will be reciprocated, and that
relate to creativity across two studies in two school settings (ele- someone is not alone in their values and preferences.
mentary and high schools). This natural setting is where students Although homophily has been demonstrated across a variety of
have the opportunity to develop friendships and engage in creative factors, we expand on this previous research and examine the
activities, which likely occur in tandem. Often times this creativity relationship between creativity and friendship ties. Creativity will
is everyday creativity like “little-c” and “mini-c” creativity rather influence friendship formation because two people will have sim-
than “pro-c” or “Big-C” creativity (Kaufman & Beghetto, 2009). ilar values and preferences, which will lead people to have positive
We investigate the relationship between friendship ties and interpersonal experiences (Izard, 1960). For example, the person-
creativity using two perspectives. First, we examine the relation- ality factor openness to experience, which is closely related to
ship between friendship and creativity based on the homophily or creativity, influences the hobbies that people engage in (Wolfradt
similarity-attraction hypothesis. That is, people are attracted to and & Pretz, 2001). Thus, creative people might be drawn to others
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

select friends that are similar to themselves (Byrne, 1971; McPhe- who are similarly creative because both share a preference for
This document is copyrighted by the American Psychological Association or one of its allied publishers.

rson, Smith-Lovin, & Cook, 2001). Second, we argue that creative engaging in similar hobbies and obtaining new experiences. If
students are more popular within their friendship networks based
people did not share the same values and preferences, then either
on prior research. Because social network analyses are less com-
person has the option to change their values and preferences or end
mon in the creativity literature, we begin with an overview of its
the friendship (McPherson et al., 2001). Additionally, people sim-
basic assumptions. We then discuss the two perspectives on the
ilar with regard to creativity might feel more comfortable sharing
relationship between friendship ties and creativity. Last, we test
their ideas without fear that people will be biased toward those
our hypotheses in two studies.
ideas (Mueller, Melwani, & Goncalo, 2012). Dissimilar people
might lead one person (or both) to feel uncomfortable sharing new
Social Network Analysis ideas with each other for fear of criticism. Thus, we predicted:
Social network analysis is a methodological and analytical ap-
Hypothesis 1: Similarity in creativity-related factors will pre-
proach to studying how nodes are connected or unconnected to
other nodes through various ties (Borgatti, Everett, & Johnson, dict stronger friendship ties.
2013; Wasserman & Faust, 1994). Nodes are people, teams, ani- Although people might be drawn to similar others, people are
mals, countries, and so forth, and are distinguished from each other likely drawn to creative others in general. That is, people might
based on categorical (e.g., race, ethnicity) or continuous (e.g., age, report friendship ties with creative others, whether or not this
personality, or a country’s GDP) attributes. Ties or relations be- friendship is reciprocated. In line with this notion, Lau and Li
tween two nodes can be anything from friendship, information (1996) examined the relationship between peer and teacher per-
sharing, marriage, or kinship. These ties can be dichotomous (e.g.,
ceived creativity and popularity among students in primary school
married or not married) or valued/weighted (e.g., friendship
(Grade 5) in China. They found that “popular” students were
strength). The information that is acquired from the network is
significantly more likely to be perceived as creative by their peers
often used to describe various everyday social systems (e.g., or-
and teacher compared with students who were in an “average”
ganizational dynamics, friendship in school settings).
group. Additionally, Kéri (2011) found in a sample of adults that
Researchers also use social network measures (e.g., centrality,
the size of one’s primary network (i.e., close friends and family)
subgroups) as predictors or outcomes when testing research ques-
positively predicted a person’s creative achievements, but the size
tions and hypotheses. These hypotheses might be at the node level
(e.g., does age predict one’s location in the network?), dyad level of one’s secondary network (i.e., who they send Christmas cards to
(e.g., are two people similar in personality traits more likely to by mail or electronically) was unrelated to creative achievement.
become friends?), or network level (e.g., how does turnover in an Despite the above studies providing support for the relationship
organization influence the structure of the network?). In summary, between popularity and creativity, they are limited in certain ways.
social network analysis allows researchers to determine how var- First, Lau and Li (1996) focused on perceived creativity rather than
ious social relationships predict behavior. This information can be using more objective measures. Thus, people might infer that
useful for understanding how social relations predict workplace popular students are more creative when they are not. Second, Kéri
performance, unhealthy behaviors (e.g., smoking), and more. In (2011) used the interview measure of social relationships (Brugha
the current study, we apply social network analysis to examine the et al., 1987) to assess one’s primary network. This measure utilizes
relationship between friendship ties and creativity. an open-format where people must recall their close friends and
family. Thus, memory biases might influence one’s recall of their
friendship network (Borgatti et al., 2013). To overcome these
The Role of Friendship Ties With Creativity
potential limitations, we measured creativity objectively with a
Based on the homophily and similarity-attraction hypotheses, creative performance test and measure popularity using social
we argue that people will report friendships with similar creative network methods in smaller group of students where recall errors
others. People often form relationships (e.g., friendship, marriage, are less likely to occur. Based on the above research, we predicted:
and exchange) with people who are similar to them on various
attributes like gender, race, social status, education, religion, and Hypothesis 2: Popularity will be positively related to creative
personality (Byrne, Griffitt, & Stefaniak, 1967; Izard, 1960; Kao & performance.
CONNECTED TO CREATE 3

The Present Studies number of elements and sets used for the creation of the individual
drawing. Transformativeness scores range from 2 (one element
The goal of the present study was to test two hypotheses on the from one set) to 20 (all 16 elements from the 4 sets). Originality
role of friendship ties with creativity. For Hypothesis 1, we exam- was scored by an external judge trained in the scoring of the TCI
ined the relationship between friendship ties and similarity on using quarter point increments on a 5-point scale.
various creativity-related attributes. For Hypothesis 2, we exam- Network data. Students were asked to evaluate their class-
ined whether popularity predicted creative performance on a cre-
room peers on five criteria using a 1–5 scale. The five criteria
ativity test. We tested these hypotheses across two studies. In
were: (a) who they liked to spend their holidays with, (b) who they
Study 1, which was a pilot study, we examined the relationship
would go to the cinema with, (c) who they would talk with about
between friendship strength and creativity among high school
personal troubles, (d) who can help if they have problems with
students. In our primary study (Study 2), we examined the rela-
school subjects, and (e) and who they think is gifted. This resulted
tionship between friendship nominations and creativity among
in five matrices with rows indicating each student’s evaluation of
elementary school students (Grade 5).
all other students and columns indicating the other students’ eval-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

uation of a given student.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Study 1: Pilot Study To identify a parsimonious number of matrices, we conducted


principal components analysis (PCA) on the five matrices using
Method UCINET (Borgatti, Everett, & Johnson, 2002).1 To conduct PCA
in UCINET, a correlation matrix is calculated across the five
Participants and procedure. Students (N ! 100) from six matrices for each classroom, which is then used as the input for the
high school classrooms in Poland participated in the study. There PCA analysis. Across the six classes, one component emerged
were nine students in Class A, 22 in Class B, 14 in Class C, 15 in with an eigenvalue greater than one and extracted between 65 and
Class D, 25 in Class E, and 15 in Class F. Students first rated all 69% of the total variance (see Table 1). All five criteria loaded on
other students within the same classroom on five friendship criteria the one “friendship” component. Based on the results of the PCA,
and then completed the Creative Behavior Questionnaire and the scores across the five criteria were averaged into a single friend-
test of creative imagination. ship matrix, which was used for all analyses.
Measures. Two characteristics of the data must be considered. First, the
Creative Behavior Questionnaire. We used the 60-item KANH network data is valued. Valued data allows us to examine friend-
Creative Behavior Questionnaire (Gralewski & Karwowski, 2013, ship strength, but does not allow us to examine certain measures
2016; Karwowski, 2008a) to assess creative and imitative atti- like betweenness and closeness centrality, which are moderately
tudes. Each score is composed of two subscales. The Creative correlated with creativity (Baer et al., 2015). However, these other
Attitudes Scale includes the nonconformism personality subscale measures are also beyond the scope of the current study. Although
(" ! .82) and the heuristics behavior subscale (" ! .79) assessing valued data can be dichotomized, doing so results in a loss of
cognitive approaches to problem solving. The Imitative Attitudes useful information regarding friendship strength and choosing a
scale includes the conforming personality subscale (" ! .79) and point to dichotomize the data is often a difficult decision (e.g., if
the algorithmic behavior subscale (" ! .69) assessing cognitive a value of 3 were chosen as a cutpoint to dichotomize the data, 2.9
approaches to problem solving. Each subscale has 15 items, which and 3.0 would be coded as 0 and 1, respectively). Indeed, Was-
are rated on a 3-point scale. Participants indicate whether an item serman and Faust (1994) note, “The graph for a valued relation
was reflective of them (score of 2), they do not know if it reflects must convey more information by representing the strength of an
them (score of 1), or if it does not reflect them (score of 0). Scores arc or a line” (p. 136). In this study, we are able to test our
ranged from 0 to 60 for the creative and imitative attitude scales. hypotheses with valued data. Thus, we retained the valued data
Test of creative imagination. The test of creative imagination rather than dichotomize it.
(TCI; Karwowski, 2008a, 2008b) is a nonverbal, culture-fair test Second, the six classrooms were sociocentric (i.e., closed net-
and was used to assess creative performance. Participants were works), which is distinguished from egocentric networks. Egocen-
given a sheet of paper with 16 elements on it: 4 straight lines, 4 tric networks might involve one person listing who is in their
dots, 4 curvy lines, and 4 semicircles. They were instructed to use primary friend group and secondary friend group (e.g., Kéri,
some or all 16 elements (but not more) to draw figures that 2011). This results in many stand-alone personal networks and
represent something that does not exist, but should exist. These data analysis is centered on the person providing the data. Socio-
could be anything from new appliances, inventions, or expressed centric networks, on the other hand, have a boundary of who is
ideas. Participants were instructed to come up with ideas that were included. In the current study, this was the total number of students
as original as possible and provide a short description of their in each classroom. Thus, network size was fixed to classroom size.
drawing and what the figure could be used for. Participants worked Network measures. We used two social network measures to
on the task for 30 min. test our hypotheses. First, to test Hypothesis 1, we created five
The test was scored for (a) fluency, (b) transformativeness, and difference score matrices using participants’ scores on the Creative
(c) originality. Fluency was defined as the number of responses Behavior Questionnaire and TCI. There were matrices for each of
provided. Responses were not counted if: (a) more than 16 ele- the six classes: (a) creative attitudes, (b) imitative attitudes, (c)
ments were used, (b) a description of the drawing was not pro-
fluency, (d) transformativeness, and (e) originality. The values in
vided, or (c) a drawing was of a common object (e.g., a table).
Transformativeness was defined as the elaboration and extent of
the drawing visualization. Scores were calculated as the average 1
Principal Axis Factoring is not available in UCINET.
4 MCKAY, GRYGIEL, AND KARWOWSKI

Table 1
Principal Component Analysis Results and Network-Level Descriptive Results Across the Six Classes (Study 1)

Measures Class A (N ! 9) Class B (N ! 22) Class C (N ! 14) Class D (N ! 15) Class E (N ! 25) Class F (N ! 15)

Principal component analysis


Holiday .87 .89 .90 .87 .89 .89
Cinema .88 .84 .86 .88 .88 .85
Personal troubles .82 .72 .82 .80 .83 .88
Academic troubles .68 .83 .80 .76 .72 .73
Gifted .87 .73 .80 .82 .68 .78
Eigenvalue 3.42 3.24 3.48 3.43 3.25 3.44
% Total variance extracted 68.4 64.7 69.7 68.6 65.0 68.7
Cronbach’s " .88 .86 .89 .88 .86 .88
Network-level measures
Density M (SD) 3.88 (1.03) 2.56 (1.00) 3.08 (1.03) 2.81 (1.07) 2.33 (1.02) 2.97 (1.03)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Indegree centrality M (SD) 3.88 (.78) 2.56 (.47) 3.08 (.79) 2.81 (.56) 2.33 (.40) 2.97 (.28)
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Indegree centrality min./max. 2.28/4.68 1.56/3.31 1.52/4.40 1.70/3.54 1.47/2.86 2.67/3.71


Outdegree centrality M (SD) 3.88 (.37) 2.56 (.49) 3.08 (.35) 2.81 (.49) 2.33 (.45) 2.97 (.40)
Outdegree centrality min./max. 3.20/4.20 1.71/3.45 2.63/3.59 1.77/3.70 1.40/3.06 2.33/3.53

the five matrices were the absolute difference between each par- equal to or greater than 1 SD above density. Thus, as one moves
ticipant’s individual score and each person within each network. In right across the graphs, greater restrictions are placed on the
other words, each network allows us to determine if there is a friendship ties displayed, which helps determine which friendship
relationship between similarity/difference scores on the creative ties in the network are stronger. In the second and third graphs,
attributes and friendship ties. Low scores in each of the five thick lines indicate mutual ties (i.e., both people reported a friend-
difference networks indicate two people are similar to each other ship tie above the threshold) and thin lines indicate one person
and high scores indicate two people are different from each other. reported a friendship tie above the threshold and the other person
These networks are sometimes referred to as reversed homophily did not. A notable feature of metric MDS is that two nodes are
matrices. placed closer together if friendship tie strength is stronger and two
Second, to test Hypothesis 2, we operationalized popularity nodes are more distant if friendship tie strength is weaker. For
using degree centrality within each of the six classrooms. Degree example, on the left side of the graph nodes E15 and E22 are
centrality for valued data involves indegree and outdegree ties. placed close together because they provided a score of 4.6 for each
Because there were six networks, we obtained standardized degree other. To visualize where creative people are within each network,
centrality scores, which allow for comparisons across networks of we weighted the size of the nodes by students’ originality scores.
different sizes. Standardized indegree ties for each person is the Larger nodes indicate higher scores and smaller nodes indicate
average friendship tie strength going into the person from all lower scores.2
people within the network. Thus, it is the entire network’s average Examining the graph for Class E, it appears that both hypotheses
perception of that person as a friend, which indicates a person’s could be supported. Specifically, throughout each of the network
overall popularity within the network. Standardized outdegree ties graphs, people with higher originality scores were closer to other
are each person’s average friendship tie strength going out of that people with higher original scores, and people with lower origi-
person to all other people in the network. Thus, it is the person’s nality scores were closer to other people with lower originality
average perception of other people as their friend. scores. Additionally, there are also cases where more creative
people have more incoming ties even after placing greater restric-
Results and Discussion tions on the strength of friendship ties displayed. This was also the
case with the graphs for the other classrooms (not displayed).
Table 1 provides network-level measures for each of the six
To test Hypothesis 1, which stated that similarity in creativity-
networks. Density, which is the average tie strength across all ties
related factors would predict stronger friendship ties, we con-
within the classroom, varied among the six classes. It is important
ducted Quadratic Assignment Procedure (QAP) correlation and
to note that indegree and outdegree centrality means are the same
regression analyses in UCINET. QAP analyses are a method to
as density (i.e., network level). This is because all ties serve as
correlate or regress a dependent network on one or more indepen-
both incoming and outgoing ties, but centrality scores differ for
dent networks. In the current study, friendship ties (dependent
each person (node level), which is why SDs for indegree and
network) were regressed on the five creative attribute matrices
outdegree centrality vary across the classes.
(independent networks). With QAP analyses, observed coeffi-
Before testing our hypotheses, we created graphs for each of the
cients are obtained and p values are based on comparing the
six networks. To create the graphs, we used metric multidimen-
observed coefficients with thousands of pairs of matrices that are
sional scaling (MDS) in the program NetDraw (Borgatti, 2002),
which is a visualization program accompanying UCINET. Figure similar to the input matrices, but are independent of one another
1 contains three graphs for the largest classroom (i.e., Class E).
The first graph includes all ties, the second graph includes ties 2
Graphs for the other classes are available in an online supplemental
equal to or greater than density, and the third graph includes ties file.
CONNECTED TO CREATE 5
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 1. Class E network in Study 1 with (a) all ties present, (b) ties equal to or greater than density present,
and (c) ties equal to or greater than 1 SD above density present. Thick lines indicate mutual ties (i.e., both people
reported a friendship tie above the threshold), whereas thin lines indicate nonmutual ties (i.e., one person
reported a friendship tie at or above the threshold and the other person did not). Node size was weighted by
originality scores (i.e., larger nodes indicate higher originality scores).

(Borgatti et al., 2013). The p value obtained for each coefficient is sion, we conducted QAP correlation analyses for each class using
based on the proportion of the coefficients among the independent friendship ties and similarity/difference matrices. This information
matrices that were similar in size to the observed coefficients. is available for each of the six classes in an online supplemental
Thus, QAP analyses are a permutation technique and stable p file. However, the results were highly consistent with those of
values are based on using a large number of permutations. We QAP regression analysis, which is reported in Table 2. As Table 2
conducted QAP correlations analyses and Double Dekker Semi- shows, there was only one coefficient supporting the hypothesis. In
Partialing Multiple Regression QAP analyses in UCINET (Bor- Class C, transformativeness was significantly negatively related to
gatti et al., 2002) using 2,000 permutations in each of the six friendship ties. Notably, there were other marginally significant
classes. Dyadic sample sizes were generally large enough in each correlations, but there was no consistent pattern among the coef-
of the six classes to conduct the analyses for each network sepa- ficients. Thus, Hypothesis 1 was unsupported.
rately (N ! 72– 600). To test Hypothesis 2, which stated that popularity would be
As mentioned, low scores in each of the five difference net- significantly positively related to creative performance, we com-
works indicate two people are similar to each other and high scores bined the data across the six classes and conducted three ordinary
indicate two people are different. Therefore, a negative correlation least squares (OLS) regression analyses, one for each creativity
or # value would indicate similarity on the creativity attributes measures (fluency, transformativeness, and originality). Popularity
predicts stronger friendship ties, whereas a positive correlation or was conceptualized using indegree centrality. We controlled for
# value would indicate similarity on the creativity attributes pre- outdegree centrality and each student’s score on the creative and
dicts weaker friendship ties. Before conducting the QAP regres- imitative attitude scales. Because students were nested within one

Table 2
QAP Regression Results Across the Six Classes (Study 1)

Class A Class B Class C Class D Class E Class F


(N ! 72) (N ! 462) (N ! 182) (N ! 210) (N ! 600) (N ! 210)
Variables b (#) SE b (#) SE b (#) SE b (#) SE b (#) SE b (#) SE

Intercept 4.01 !!!


2.44 !!!
3.35 !!!
2.72 2.46
!!! !!!
2.95!!!

Creative attitudes $.08 ($.24) 3.46 .02 (.10)† .61 .02 (.08) 1.11 .03 (.15)! .96 % .01 (.03) .54 % .01 (.01) 1.06
Imitative attitudes $.05 ($.21) 2.65 .01 (.02) 1.03 $.03 ($.10) 1.39 & $.01 ($.03) .86 & $.01 ($.02) .53 % .01 (.04) .60
Fluency .09 (.14) 12.12 $.04 ($.09) 5.27 .06 (.04) 14.31 $.24 ($.17) 14.29 $.10 ($.15) 6.41 .06 (.13)† 2.18
Transformativeness .09 (.17) 4.77 $.01 ($.02) .51 $.07 ($.38)!! 1.55 $.03 ($.16)† .75 $.02 ($.10)† .46 $.02 ($.10) 1.10
Originality .03 (.05) 11.78 .02 (.05) 4.71 .10 (.09) 11.25 .22 (.25)† 8.63 .09 (.13) 6.40 $.02 ($.06) 1.34
R2 .10! .01† .12!!! .08!! .01! .02†
Adjusted R2 .03 .001 .09 .05 .004 $.005
Note. QAP ! Quadratic Assignment Procedure. Ns for each class are based on the number of dyads in the network.

p % .10. ! p % .05. !! p % .01. !!! p % .001.
6 MCKAY, GRYGIEL, AND KARWOWSKI

Table 3
Correlations Among the Variables in the Analyses (Study 1)

Measures M SD 1 2 3 4 5 6

1. Fluency 2.36 2.18 —


2. Transformativeness 12.55 6.49 .10 —
3. Originality 2.61 2.52 .79!!! .16 —
4. Imitative attitudes 27.18 6.45 .27!! $.15 .22! —
5. Creative attitudes 37.99 6.36 .39!!! $.07 .36!!! .10 —
6. Indegree centrality .98 .18 .27!! .12 .30!! .02 .05 —
7. Outdegree centrality .99 .15 .21! .03 .11 $.01 .05 .45!!!
Note. N ! 100.
!
p % .05. !! p % .01. !!!
p % .001.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

of six classes, we created five dummy-coded variables and con- Study 2: Primary Study
trolled for classroom membership with Class A serving as the
referent group. Before conducting the OLS regression analysis, we
examined the bivariate correlations among the centrality scores Method
and the creativity factors collapsed across the six classes (see Participants and procedure. We used data from the larger
Table 3). Indegree centrality had significant positive bivariate longitudinal study “School Effectiveness Research” (Dolata,
correlation with fluency and originality. Thus, Hypothesis 2 had 2014), which included a representative sample of pupils from
initial support. We then conducted the OLS regressions to formally Polish elementary schools. The study used a stratified two-stage
test Hypothesis 2. As Table 4 shows, creative attitudes signifi- cluster sampling procedure. The strata were determined by type of
cantly positively predicted fluency and originality scores. Imitative urbanization and the number of classrooms within a school. Within
attitudes also significantly positively predicted fluency scores. strata, schools were sampled with a probability proportional to size
Specific to our hypothesis, indegree centrality significantly posi- (number of students). For the current study, only students from
tively predicted originality scores and accounted for 3.4% of the fifth grade (12-year-olds) were included. The final sample for the
variance after controlling for classroom membership, creative and current study was 5,659 students from 295 classrooms in 173
imitative attitudes, and outdegree centrality.3 Thus, Hypothesis 2 elementary schools (49.7% female). The average classroom size
was partially supported. was 19.18 (SD ! 4.45, range ! 7–30).
Overall, the results of Study 1 failed to support Hypothesis 1, Measures.
but partially supported Hypothesis 2. Specifically, students’ sim- Creativity. The Test of Creative Imagery Abilities (TCIA) was
ilarity on various creativity-related attributes failed to predict used to measure creative imagination (Jankowska & Karwowski,
friendship strength, but popularity was positively related to orig- 2015). The TCIA test booklet consists of seven tasks. The first
inality. stage of solving each task is to generate, in an oral or written form,
Although this pilot study provides an initial test of our hypoth- as many images as possible based on a simple graphic sign, called
eses, it is important to consider three limitations. First, the overall the initial figure. Next, participants select what they think is the
and classroom sample sizes were small. This resulted in our most original of the images given and create a drawing accompa-
statistical tests being underpowered, which might have influenced nied by a brief description. The instructions stress the possibility of
our not finding support for Hypothesis 1. Second, we did not have elaborating and changing the selected image and adding any ele-
demographic information for participants. Previous research has ments to it in such a way as to create something even more
found that men perceive themselves to be more creative than original. The drawings and descriptions of the images made are
women (Furnham, Hosoe, & Tang, 2001; Karwowski, 2011; Kar- assessed on three scales based on the conjunctional model of
wowski, Gralewski, & Szumski, 2015) and men are also perceived creative imaging ability (Dziedziewicz & Karwowski, 2015): (a)
by others as being more creative than women (Lau & Li, 1996; vividness scale, (b) originality scale, and (c) transformativeness
Proudfoot, Kay, & Koval, 2015). Last, we only used one concep- scale. Each scale is scored according to the criteria discussed in
tualization of friendship: friendship strength. An alternative con- detail and illustrated with examples elsewhere (see Jankowska &
ceptualization of friendship would be friendship nomination. Al- Karwowski, 2015). According to these criteria, it is possible to
though we argued friendship strength is an advantage of the score 0, 1, or 2 points on each scale for each drawing. The scores
current study, it can also be a limitation. These measures are on each scale are computed by adding up the points scored for all
conceptually different and could result in a different pattern of the drawings. Rather than using the three separate scale scores, an
results. Friendship strength involves making judgments of how overall score may be used. In the current study, we averaged the
close another person is to them, whereas friendship nominations
are simply whether someone is considered a friend or not. Utilizing
3
these alternate conceptualizations might result in a different pat- There was one univariate outlier on originality (z ! 4.51). We con-
tern of results. The purpose of the second and primary study was ducted the analyses after removing this participant and the results remain
unchanged. Indegree centrality remained a significant predictor of origi-
to expand on the pilot study with a larger number of students and nality, b ! .83 (# ! .25; SE ! .37), p ! .025, and accounted for an
classes, control for gender, and use friendship nominations as the additional 3.7% of the variance in originality scores after controlling for the
dependent variable rather than friendship strength. other predictors.
CONNECTED TO CREATE 7

Table 4
Regression Results With the Three Creativity Measures Regressed on Creative and Imitative
Attitudes and Indegree- and Outdegree-Centrality (Study 1)

Fluency Transformativeness Originality


Variables b (#) SE b (#) SE b (#) SE

Intercept $3.89 2.60 6.28 (.21) 9.68 $2.96 3.21


Class B ! 1 $.74 ($.14) .98 3.24 (.29) 3.67 $1.06 ($.18) 1.22
Class C ! 1 $2.04 ($.33)! .86 5.34 (.17) 3.20 $2.12 ($.29)! 1.06
Class D ! 1 $1.59 ($.26) .96 3.15 (.32) 3.57 $1.57 ($.22) 1.19
Class E ! 1 $1.04 ($.21) 1.06 4.69 ($.01) 3.96 $1.45 ($.25) 1.31
Class F ! 1 .90 (.15) .88 $.21 ($.11) 3.26 .78 (.11) 1.08
Imitative attitudes .06 (.19)! .03 $.11 ($.03) .10 .05 (.13) .04
Creative attitudes .09 (.26)!! .03 $.03 ($.03) .11 .10 (.26)!! .04
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Outdegree centrality .18 (.05) .41 .64 (.06) 1.51 $.44 ($.11) .50
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Indegree centrality .52 (.16) .33 1.96 (.21) 1.24 .89 (.24)! .41
F 7.52!!! 1.24 5.41!!!
R2 .43 .11 .35
Adjusted R2 .37 .02 .29
Note. N ! 100. Degrees of freedom for the three regression analyses was 9, 99. Classes were dummy coded
and controlled for in each analysis. Class A served as the referent group.
!
p % .05. !! p % .01. !!! p % .001.

scores on the three scales for an overall score (" ! .91). For each • Nodeocov ! The tendency for people with higher lev-
student, their overall creativity score was standardized within els of the covariate (creativity, in the current study)
classrooms (M ! 0, SD ! 1). For students without creativity sending more friendship nominations to other people
scores missing data were replaced with classroom means. (i.e., outgoing ties).
Network data. Children peer networks were identified by us- • Absdiff ! Reversed homophily. A positive value means
ing a sociometric procedure developed by Coie, Coppotelli, and there is a tendency to reciprocate friendship nomina-
Dodge (1982). Students were asked to nominate schoolmates from tions among people who are dissimilar on creativity,
the same classroom with whom they most liked to play. Thus, ties whereas a negative value means there is a tendency to
were dichotomous and network size was fixed to the classroom reciprocate friendship nominations among people who
size (i.e., sociocentric network). are similar on creativity. This parameter was used to
Network Measures and statistical procedure. To assess the test Hypothesis 1.
relevance of the different structural processes, we fitted an expo- • Nodematch ! Gender homophily. A positive value
nential random graph model (ERGM) to each observed classroom indicates boys are more likely to nominate other boys
network. ERGM allows us to test both of our hypotheses simul- and girls nominate other girls, whereas a negative value
taneously and control for the effects of the other variables. We are indicates boys are more likely to nominate girls and
also able to control for other network characteristics (e.g., reci- girls are more likely to nominate boys.
procity of friendship nominations) in the same analysis. The out-
come variable in ERGMs must be dichotomous (friendship nom-
inations in this study). For more information on ERGMs, see Results and Discussion
Borgatti et al. (2013, pp. 139 –145) and Wimmer and Lewis First, we fit the models for each of the 295 classes separately.
(2010). Specifically, in our main ERGM, we include the following Rather than report the results for all 295 classes separately, we
structural parameters, which are described below: (a) reciprocity combined the results through a meta-analysis. Conducting a meta-
(mutual), (b) three effects based on creativity (nodeicov, nodeo- analysis allows us to present ERGM results from all classes
cov, and absdiff), and (c) controlling for effects based on gender succinctly, and to evaluate overarching tendencies across all the
homophily (nodematch). All models were fitted using the “ergm” networks rather than within each network separately. To test
package (Hunter, Handcock, Butts, Goodreau, & Morris, 2008) in whether between classrooms variance was zero, we used the Q
R (R Development Core Team, 2013). We also provide an expla- statistic. Meta-analysis were conducted in R with the “metafor”
nation and definition of the effects used in the analyses: package (Viechtbauer, 2010).
• Edges ! A parameter that is similar to the intercept in The ERGM results are presented in Table 5. In support of
regression models and should not be directly interpreted. Hypothesis 1, we found a significant negative relationship
• Mutual ! The tendency to reciprocate friendship nomina- between the reversed homophily for creative performance and
tions (positive value means I like X and X likes me). friendship nominations, after controlling for the other variables
• Nodeicov ! The tendency for people with higher levels in the model, b ! $0.20, p % .001. In other words, students
of the covariate (creativity, in the current study) to were more likely to nominate another person as a friend when
receive more friendship nominations from other people they performed similarly on the creativity test. In support of
(i.e., incoming ties). This parameter was used to test Hypothesis 2, we found a positive relationship between inde-
Hypothesis 2. gree ties and creativity scores, b ! 0.33, p % .001. Students
8 MCKAY, GRYGIEL, AND KARWOWSKI

Table 5
Meta-Analysis ERGM Results for the Formation of Friendship Ties (Study 2)

ERGM parameter Heterogeneity


Parameter Estimate (SE) 95% CI Tau2 (SE) I2 H2 Q

Edges $2.26 (.05) !!!


[$2.36, $2.16] .55 (.06) 84.73% 6.55 1513.44!!
Mutual 1.33 (.04)!!! [1.25, 1.41] .24 (.03) 59.07% 2.44 696.77!!
Nodeicov (CRE) .33 (.08)!!! [.18, .48] .95 (.12) 69.69% 3.30 861.14!!
Nodeocov (CRE) .17 (.07)! [.03, .31] .76 (.10) 64.85% 2.84 769.98!!
Absdiff (CRE) $.20 (.06)!!! [$.32, $.08] .41 (.07) 49.34% 1.97 571.06!!
Nodematch (GENDER) 1.51 (.04)!!! [1.43, 1.59] .32 (.03) 80.70% 5.18 1273.71!!
Note. N ! 295 classrooms. A restricted maximum likelihood estimation was used. ERGM ! exponential
random graph model; CI ! confidence interval; CRE ! creativity; GENDER: 0 ! females, 1 ! males; Tau2 !
estimate of total amount of heterogeneity; I2 ! statistic estimates (in percent) how much of the total variability
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

in the effect size estimates (that is composed of heterogeneity and sampling variability) can be attributed to
This document is copyrighted by the American Psychological Association or one of its allied publishers.

heterogeneity among the true effects; H2 ! the ratio of the total amount of variability in the observed outcomes
to the amount of sampling variability; Q ! omnibus test for residual heterogeneity (Cochran Q test).
!
p % .05. !! p % .01. !!! p % .001.

with higher creative performance scores were more likely to be underpowered. Thus, the results of Study 2 are more reliable
nominated as a friend by their peers. when determining support for Hypothesis 1. Last, we used two
different conceptualizations and methodological approaches for
General Discussion friendship across the two studies. Specifically, we used friend-
ship strength in Study 1 and friendship nominations in Study 2.
Creativity research has focused heavily on studying individ- The differing pattern of results might indicate that how re-
ual differences and contextual factors as predictors of creative searchers measure friendship ties might influence the pattern of
performance. Less attention has been given to the relationship
the results observed. Thus, the current study provides both
between social relationships like friendship with creativity. The
theoretical and methodological contributions. Theoretically, we
purpose of the present article was to test two hypotheses ex-
provide support for the notion that people nominate similar
amining the relationship between friendship ties and creativity
creative others as friends, which aligns with the homophily and
in a natural setting (i.e., schools and classrooms). Across two
similarity-attraction hypotheses. Methodologically, we identify
studies (a pilot and primary study), we found mixed support for
that different conceptualizations of friendship might lead to a
Hypothesis 1. Specifically, friendship strength was unrelated to
different pattern of results. Thus, future research examining
creative similarity among high school students (Study 1), but
friendship ties and creativity should consider how they measure
elementary school students were more likely to nominate their
friendship.
peers as friends when they had similar scores on a creativity test
The results of both studies also demonstrated a positive
(Study 2). However, we found support for Hypothesis 2 in both
studies. That is, there was a positive relationship between relationship between popularity and creativity, which supports
friendship strength and originality scores (but not fluency or Hypothesis 2 and aligns with previous research (Kéri, 2011;
transformativeness in Study 1) and students were more likely to Lau & Li, 1996). Specifically, we found a positive relationship
nominate peers who had higher creative performance scores as between friendship strength and originality scores (Study 1) and
friends (Study 2). The results indicate that friendship ties, students with higher creativity scores were more likely to be
which have received less attention in the mainstream creativity nominated by their peers (Study 2). Although creative people
literature, and the social networks and creativity literature, are can be disagreeable (Hunter & Cushenbery, 2015) and “rule
related to creativity. breakers” (Gino & Ariely, 2012; Gino & Wiltermuth, 2014),
As mentioned, the results across both studies presented they also possess other positive social traits like ambition and
mixed support for Hypothesis 1. Across the two studies, the openness to experience (Feist, 2010) and are often perceived as
results indicate that people are more likely to nominate as more sexually attractive than less creative people (Kaufman et
friends those who are similar on creativity-related attributes al., 2014). To sell creative ideas, people must clash with norms
(Study 2), but not necessarily report stronger friendship ties and overcome biases associated with creative ideas (Mueller et
with them (Study 1). There are three potential reasons for the al., 2012). Such people might be difficult to handle interper-
different results. First, age differences (i.e., high school vs. sonally at times, but they also inspire and attract others. This is
elementary school students) might have influenced the results. analogous to the role of leadership in organizational settings
That is, the similarity-attraction effect might differ across age (Mumford, Hunter, Eubanks, Bedell, & Murphy, 2007). Leaders
groups when it comes to creativity. Although this is a possible must have the capacity to go against norms at times and be
interpretation, that requires future research to rule out, we creative themselves to inspire their subordinates to be creative
believe the next two interpretations are more likely the reason and develop their own capabilities. Thus, examining a person’s
there was a difference. Second, and previously mentioned, the friendship network early in life might be useful to identify who
first study had a smaller sample size, leading to the study being emerges as future leaders.
CONNECTED TO CREATE 9

The current study examines how one’s friendship network re- Third, we focused on domain-general creativity in the current
lates to individual creativity. We examine how friendship ties study, which does not consider the specific creative domains
among people in everyday environments influences or is influ- (e.g., art, theater, and science). Focusing on domain-specific
enced by creativity. Based on the results, we argue that social creativity factors might provide a different pattern of results.
network approaches can help provide a more nuanced view for That is, people interested in artistic creativity might be attracted
how individual differences, contextual sources, and social rela- to others who are interested in artistic creativity and not those
tionships influence creativity. interested in scientific creativity. Such studies might utilize
subgrouping analyses to determine if people cluster together
based on similar interests. Additionally, some “outsider” move-
Limitations and Directions for Future Research
ments might include creative people who might not be per-
Although the two studies expand on the previous social net- ceived as popular. Greater specificity in future research might
works and creativity literature, it is not without some limitations. lead to different conclusions.
First, both studies were cross-sectional and we cannot establish Last, network tools can be used to examine connections
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

causality. We cannot say with any certainty that popularity causes across multiple levels of analysis. For example, Kenett, Beaty,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

people to come up with more creative ideas or that generating Silvia, Anaki, and Faust (2016) used network analyses to un-
more creative ideas causes one to become more popular. In fact, derstand how cognitive processes influence creativity. This
we believe this relationship is reciprocal. People who are open to work aligns with the current study as both utilize network
experiences and engage in creative activities might have more measures to better understand creativity. The current studies
opportunities to meet and attract others. People who increase their examining social relationships and their study developing
friendship network might lead them to have more information sematic network at the cognitive level. Developing a greater
available to them, which can be integrated and used to form understanding of creativity requires understanding how connec-
creative ideas. tions at multiple levels influence creativity.
Second, we only assessed friendship ties with other students
within the same classroom. In other words, we did not obtain
information on each student’s larger friendship network that ex- Conclusion
tends beyond the classroom. The method used differs from previ- Creativity occurs at the intersection of individual differences,
ous creativity and social network (i.e., Kéri, 2011). It is important contextual factors, and interpersonal relationships. The current
to note that each method has its advantages and disadvantages. If study examined the relationship between friendship ties and cre-
we had chosen to have people list their friends, memory biases ativity, extending the previous network-creativity literature by
might influence the results and it might have been difficult to considering a different type of tie. The current study serves as a
obtain creativity attributes from the people listed by the primary call for future research to investigate the relationship between
participant. Thus, we would be unable to examine the similarity- interpersonal factors and creativity.
attraction hypothesis that we tested in the current study. The design
used in the current study allowed us to obtain complete data on the
creativity measures and friendship ties among people who spend a References
large portion of their day together. It is important to note that the Amabile, T. M., Conti, R., Coon, H., Lazenby, J., & Herron, M. (1996).
results were similar across Kéri’s study and the current study, Assessing the work environment for creativity. Academy of Management
which helps alleviate the limitations raised from using either Journal, 39, 1154 –1184. http://dx.doi.org/10.2307/256995
method. However, it would be beneficial to utilize both methods in Baer, M., Evans, K., Oldham, G. R., & Boasso, A. (2015). The social
future research. network side of individual innovation: A meta-analysis and path-analytic
Aside from the limitations, social networks have received integration. Organizational Psychology Review, 5, 191–223. http://dx
little attention in the creativity literature. Thus, the topic is .doi.org/10.1177/2041386614564105
wide-open for future research. We offer four broad ideas. First, Barbot, B., Lubart, T. I., & Besançon, M. (2016). “Peaks, slumps, and
bumps”: Individual differences in the development of creativity in
it can be used to examine social ties among various eminent
children and adolescents. In B. Barbot (Ed.), New directions for child
creators. For example, future research could examine how em- and adolescent development (Vol. 151, pp. 33– 45). New York, NY:
inent artists influence other eminent artists from similar and Wiley. http://dx.doi.org/10.1002/cad.20152
other artistic movements. This could be useful for determining Borgatti, S. P. (2002). NetDraw software for network visualization. Lex-
how interpersonal relationships shape one’s creative outputs ington, KY: Analytic Technologies.
over time. The same forces could be considered for other Borgatti, S. P., Everett, M. G., & Johnson, J. C. (2002). Ucinet for
domains like science, music, and literature. Various social Windows: Software for social network analysis. Harvard, MA: Analytic
interactions might have influenced various artistic movements Technologies.
and breakthrough scientific discoveries. Borgatti, S. P., Everett, M. G., & Johnson, J. C. (2013). Analyzing social
Second, it would be interesting to examine how social relation- networks. London, England: Sage.
Brugha, T. S., Sturt, E., MacCarthy, B., Potter, J., Wykes, T., & Bebbing-
ships relate to creativity across various developmental periods.
ton, P. E. (1987). The interview measure of social relationships: The
Although previous scholars have considered individual differences description and evaluation of a survey instrument for assessing personal
in various developmental periods (Barbot, Lubart, & Besançon, social resources. Social Psychiatry, 22, 123–128. http://dx.doi.org/10
2016), the results of the current study indicate social relationships .1007/BF00584017
might play an important role with creativity over different devel- Byrne, D. (1971). The attraction paradign. New York, NY: Academic
opmental phases. Press.
10 MCKAY, GRYGIEL, AND KARWOWSKI

Byrne, D., Griffitt, W., & Stefaniak, D. (1967). Attraction and similarity of Karwowski, M. (2008b). Measuring creativity using the test of creative
personality characteristics. Journal of Personality and Social Psychol- imagination (TCI). Pt. 2. On validity of the TCI. The New Educational
ogy, 5, 82–90. http://dx.doi.org/10.1037/h0021198 Review, 15, 216 –231.
Cattani, G., & Ferriani, S. (2008). A core/periphery perspective on indi- Karwowski, M. (2011). It doesn’t hurt to ask . . . But sometimes it hurts to
vidual creative performance: Social networks and cinematic achieve- believe: Polish students’ creative self-efficacy and its predictors. Psy-
ments in the Hollywood film industry. Organization Science, 19, 824 – chology of Aesthetics, Creativity, and the Arts, 5, 154 –164. http://dx.doi
844. http://dx.doi.org/10.1287/orsc.1070.0350 .org/10.1037/a0021427
Coie, J. D., Dodge, K. A., & Coppotelli, H. (1982). Dimensions and types Karwowski, M., Gralewski, J., & Szumski, G. (2015). Teachers’ effect on
of social status: A cross-age perspective. Developmental Psychology, 18, students’ creative self-beliefs is moderated by students’ gender. Learn-
557–570. http://dx.doi.org/10.1037/0012-1649.18.4.557 ing and Individual Differences, 44, 1– 8. http://dx.doi.org/10.1016/j
Dolata, R. (Ed.). (2014). Czy szkoła ma znaczenie? Analiza zróżnicowania .lindif.2015.10.001
efektywności nauczania na pierwszym etapie edukacyjnym [Does school Kaufman, J. C., & Beghetto, R. A. (2009). Beyond big and little: The four
c model of creativity. Review of General Psychology, 13, 1–12. http://
matter: An analysis of the variability in teaching effectiveness at the first
dx.doi.org/10.1037/a0013688
educational stage]. Warszawa, Poland: Instytut Badań Edukacyjnych.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Kaufman, J. C., Bromley, M. L., & Cole, J. C. (2006). Insane, poetic,


Dziedziewicz, D., & Karwowski, M. (2015). Development of children’s
This document is copyrighted by the American Psychological Association or one of its allied publishers.

lovable: Creativity and endorsement of the “mad genius” stereotype.


creative visual imagination: A theoretical model and enhancement pro-
Imagination, Cognition and Personality, 26, 149 –161. http://dx.doi.org/
grammes. Education 3–13, 43, 382–392. http://dx.doi.org/10.1080/
10.2190/J207-3U30-R401-446J
03004279.2015.1020646
Kaufman, S. B., Kozbelt, A., Silvia, P., Kaufman, J. C., Ramesh, S., &
Feist, G. J. (2010). The function of personality in creativity: The nature
Feist, G. J. (2014). Who finds Bill Gates sexy? Creative mate prefer-
and nurture of the creative personality. In J. C. Kaufman & R. J. ences as a function of cognitive ability, personality, and creative
Sternberg (Eds.), The Cambridge handbook of creativity (pp. 113–130). achievement. The Journal of Creative Behavior. Advance online publi-
New York, NY: Cambridge University Press. http://dx.doi.org/10.1017/ cation.
CBO9780511763205.009 Kenett, Y. N., Beaty, R. E., Silvia, P. J., Anaki, D., & Faust, M. (2016).
Furnham, A., Hosoe, T., & Tang, T. L. P. (2001). Male hubris and female Structure and flexibility: Investigating the relation between the structure
humility? A cross-cultural study of ratings of self, parental, and sibling of the mental lexicon, fluid intelligence, and creative achievement.
multiple intelligence in America, Britain, and Japan. Intelligence, 30, Psychology of Aesthetics, Creativity, and the Arts, 10, 377–388.
101–115. http://dx.doi.org/10.1016/S0160-2896(01)00080-0 Kéri, S. (2011). Solitary minds and social capital: Latent inhibition, general
Gino, F., & Ariely, D. (2012). The dark side of creativity: Original thinkers intellectual functions and social network size predict creative achieve-
can be more dishonest. Journal of Personality and Social Psychology, ments. Psychology of Aesthetics, Creativity, and the Arts, 5, 215–221.
102, 445– 459. http://dx.doi.org/10.1037/a0026406 http://dx.doi.org/10.1037/a0022000
Gino, F., & Wiltermuth, S. S. (2014). Evil genius? How dishonesty can Lau, S., & Li, W.-L. (1996). Peer status and perceived creativity: Are popular
lead to greater creativity. Psychological Science, 25, 973–981. http://dx children viewed by peer and teachers as creative? Creativity Research
.doi.org/10.1177/0956797614520714 Journal, 9, 347–352. http://dx.doi.org/10.1207/s15326934crj0904_6
Gralewski, J., & Karwowski, M. (2013). Polite girls and creative boys? McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather:
Students’ gender moderates accuracy of teachers’ ratings of creativity. Homophily in social networks. Annual Review of Sociology, 27, 415–
The Journal of Creative Behavior, 47, 290 –304. http://dx.doi.org/10 444. http://dx.doi.org/10.1146/annurev.soc.27.1.415
.1002/jocb.36 Mueller, J. S., Melwani, S., & Goncalo, J. A. (2012). The bias against
Gralewski, J., & Karwowski, M. (2016). Are teachers’ implicit theories of creativity: Why people desire but reject creative ideas. Psychological
creativity related to the recognition of their students’ creativity? The Science, 23, 13–17. http://dx.doi.org/10.1177/0956797611421018
Journal of Creative Behavior, n/a. Advance online publication. http:// Mumford, M. D., Hunter, S. T., Eubanks, D. L., Bedell, K. E., & Murphy,
dx.doi.org/10.1002/jocb.140 S. T. (2007). Developing leaders for creative efforts: A domain-based
Hunter, D. R., Handcock, M. S., Butts, C. T., Goodreau, S. M., & Morris, approach to leadership development. Human Resource Management
M. (2008). ergm: A package to fit, simulate and diagnose exponential- Review, 17, 402– 417. http://dx.doi.org/10.1016/j.hrmr.2007.08.002
Perry-Smith, J. E. (2006). Social yet creative: The role of social relation-
family models for networks. Journal of Statistical Software, 24, a54860.
ships in facilitating individual creativity. Academy of Management Jour-
http://dx.doi.org/10.18637/jss.v024.i03
nal, 49, 85–101. http://dx.doi.org/10.5465/AMJ.2006.20785503
Hunter, S. T., & Cushenbery, L. (2015). Is being a jerk necessary for
Perry-Smith, J. E., & Shalley, C. E. (2003). The social side of creativity: A
originality? Examining the role of disagreeableness in the sharing and
static and dynamic social network perspective. The Academy of Man-
utilization of original ideas. Journal of Business and Psychology, 30,
agement Review, 28, 89 –106.
621– 639. http://dx.doi.org/10.1007/s10869-014-9386-1
Proudfoot, D., Kay, A. C., & Koval, C. Z. (2015). A gender bias in the
Izard, C. E. (1960). Personality similarity and friendship. The Journal of
attribution of creativity: Archival and experimental evidence for the
Abnormal and Social Psychology, 61, 47–51. http://dx.doi.org/10.1037/ perceived association between masculinity and creative thinking. Psy-
h0040056 chological Science, 26, 1751–1761. http://dx.doi.org/10.1177/
Jankowska, D. M., & Karwowski, M. (2015). Measuring creative imagery 0956797615598739
abilities. Frontiers in Psychology, 6, 1591. http://dx.doi.org/10.3389/ Sawyer, R. K. (2012). Explaining creativity: The science of human inno-
fpsyg.2015.01591 vation. New York, NY: Oxford University Press.
Kao, G., & Jouyner, K. (2004). Do race and ethnicity matter among Simonton, D. K. (1984). Artistic creativity and interpersonal relationships
friends? Activities among interracial, interethnic, and intraethnic ado- across and within generations. Journal of Personality and Social Psy-
lescent friends. The Sociological Quarterly, 45, 557–573. http://dx.doi chology, 46, 1273–1286. http://dx.doi.org/10.1037/0022-3514.46.6.1273
.org/10.1111/j.1533-8525.2004.tb02303.x Simonton, D. K. (2014). The mad-genius paradox: Can creative people be
Karwowski, M. (2008a). Measuring creativity using the test of creative more mentally healthy but highly creative people more mentally ill?
imagination (TCI). Pt. 1. Presentation of a new instrument to measure Perspectives on Psychological Science, 9, 470 – 480. http://dx.doi.org/
creative potential. The New Educational Review, 14, 44 –53. 10.1177/1745691614543973
CONNECTED TO CREATE 11

R. Development Core Team. (2013). R: A language and environment for ican Journal of Sociology, 116, 583– 642. http://dx.doi.org/10.1086/
statistical computing. Vienna, Austria: Foundation for Statistical Com- 653658
puting. Wolfradt, U., & Pretz, J. E. (2001). Individual differences in creativity:
Viechtbauer, W. (2010). Conducting meta-analyses in R with the metafor Personality, story writing, and hobbies. European Journal of Personal-
package. Journal of Statistical Software, 36, 1– 48. ity, 15, 297–310. http://dx.doi.org/10.1002/per.409
Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and
applications. Cambridge, United Kingdom: Cambridge University
Press. http://dx.doi.org/10.1017/CBO9780511815478 Received July 9, 2016
Wimmer, A., & Lewis, K. (2010). Beyond and below racial homophily: Revision received October 11, 2016
ERG models of a friendship network documented on Facebook. Amer- Accepted December 20, 2016 !
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

View publication stats

You might also like