THE PURSUIT OF POSITIVE AFFECT IN TASK ADVICE NETWORKS:
EFFECTS ON INDIVIDUAL PERFORMANCE
Tiziana Casciaro
University of Toronto
March 2014
Working draft. Please do not circulate or cite.
THE PURSUIT OF POSITIVE AFFECT IN TASK ADVICE NETWORKS:
EFFECTS ON INDIVIDUAL PERFORMANCE
ABSTRACT
Research has demonstrated that people in organizations form advice networks not only for access
to task knowledge, but also for the positive affect they derive from social interaction, so much so
that they are willing to trade off task competence for the personal liking, pleasantness and energy
experienced in interactions with colleagues. The effects of such relational choices on individual
performance are poorly understood, however. I propose a theory of affect in advice networks
according to which an individual’s propensity to form task-related advice ties based on positive
affect experienced when interacting with a colleague enhances individual task performance. I
argue that this effect changes depending on the level of activation (a feeling of being energized)
generated by those interactions. Compared to forms of positive affect with neutral-to-moderate
levels of activation (feelings of enjoyment), the preferential selection of work partners that elicit
high-activation positive affect enhances task performance by stimulating an individual to engage
in task-oriented effort. Social-network and individual performance data on 430 technology
salespeople in a global IT corporation provide support for this theory. I discuss implications for
network research and rational action in organizations.
1
INTRODUCTION
As social entities whose raison d’être is enabling the collective performance of tasks that
individuals cannot perform in isolation, organizations are purposefully designed to induce
interactions between members with complementary expertise. Organizational theorists have
therefore been concerned with how to facilitate desired patterns of interaction to integrate the
differentiated competencies of organization members (Barnard, 1938; Blau & Schoenherr, 1971;
Lawrence & Lorsch, 1967; Mintzberg, 1979; Thompson, 1967). The social network of taskrelated advice that an individual forms with other organization members is a central mechanism
for such knowledge integration and the execution of interdependent tasks (Baldwin, Bedell, &
Johnson, 1997; Gargiulo, Ertug, & Galunic, 2009; Sparrowe, Liden, Wayne, & Kraimer, 2001).
How people go about seeking colleagues out for advice, therefore, is relevant to their ability to
carry out assigned tasks. Much evidence indicates that, in forming advice networks to perform
their job, people tend to seek coworkers out not only for task knowledge, but also for affective
inputs related to the hedonic rewards of social interaction (Bales, 1958; Hinds, Carley,
Krackhardt, & Wholey, 2000; Homans, 1961; Krackhardt, 1999; Krackhardt & Stern, 1998;
Roethlisberger & Dickson, 1939; Slater, 1955). The coexistence of affective and instrumental
motivations in advice networks is so prevalent that, across organizational contexts and tasks,
people have been shown to trade off task competence for the personal liking, pleasantness and
energy experienced in interactions with colleagues (Casciaro & Lobo, 2008).
How does the tendency to pursue positive affect in task-advice networks influence
individual performance? Existing theory offers ambivalent arguments in this regard. The
intertwining of instrumental and affective motivations in how people in organizations seek
colleagues out for task advice has been portrayed as a deviation from desired task-oriented
2
behaviour, whereby a logic of sentiments supplements and erodes the logic of cost and efficiency
represented by task requirements (Barnard, 1938; Merton, 1957; Roethlisberger & Dickson,
1939). This is because positive affect in social relationships tends to develop for reasons—such
as attribute similarity—that are tangential to task execution (Ibarra, 1992; Lincoln & Miller,
1979). Consistent with these arguments, performance benefits that individuals derive from their
position in organizational networks have been documented primarily for instrumental networks,
rather than expressive ones (for reviews, see Borgatti & Foster, 2003; Brass, Glaskiewicz, Greve,
& Tsai, 2004). For these reasons, people’s tendency to form task advice ties with colleagues
based on the positive emotions they feel for them may leave a reservoir of knowledge untapped
in organizations, potentially dampening performance (Casciaro & Lobo, 2008). Yet, network
research has also demonstrated that the affective content of social relationships can facilitate
organizational processes relevant to task execution, including the transfer of useful knowledge
(Levin & Cross, 2004) and the adaptive response to organizational crises (Krackhardt & Stern,
1998). Although these studies did not investigate how forming task-related advice ties based on
affective considerations influences individual performance, they pose a fundamental question
about relational choice in organizations. Does forging task advice ties for the increase in positive
affect experienced in interactions with colleagues hinder or benefit individual performance?
In this paper, I argue that an individual’s propensity to form task-related advice ties based
on positive affect experienced when interacting with a colleague enhances individual task
performance. This prediction reflects the central role of affect and motivation in performance:
affect serves a fundamental motivational function by stimulating engagement with and effort
toward assigned tasks (Buck, 1988), thus increasing the likelihood that they will be successfully
performed. Not all forms of positive affect that people draw from social interactions enhance
3
performance, however. I argue that the effect of positive affect experienced in interactions with
colleagues changes depending on the level of activation (i.e., a feeling of being energized)
generated by those interactions. It is specifically the choice of social interactions that induce a
state of high-activation positive affect—excitement (Izard, 1991), emotional energy (Collins,
1993) or energetic activation (Thayer, 1989)—that enhances task performance by stimulating
task-oriented engagement and effort. I test this argument with social network and individual
performance data on 430 technology salespeople operating in a multinational computer,
technology and IT consulting corporation headquartered in the United States.
POSITIVE AFFECT, ADVICE SEEKING, AND PERFORMANCE
The formal roles individuals perform in organizations require patterns of interaction with
colleagues that organization members may not willingly choose (Barnard, 1938; Blau &
Schoenherr, 1971; Lawrence & Lorsch, 1967; Mintzberg, 1979; Thompson, 1967). Within these
formal constraints, however, emergent task advice networks also reflect people’s discretionary
choice as to whom they wish to seek out as a source of task knowledge. Such choices are driven
by an individual’s subjective assessment of the resources embedded in interactions with different
coworkers (Merton, 1957).
Network scholars have identified several criteria people follow in forming and
maintaining task advice relationships. These criteria concern cognitive evaluations of the
expected instrumental value of task-advice interactions with a colleague (as indicated by a
colleague’s perceived task competence), and the expected accessibility of task advice (as
indicated by colleagues’ perceived willingness to share their knowledge and by their physical
proximity) (Borgatti & Cross, 2003; Hinds et al., 2000; Nebus, 2006). In addition to these
4
cognitive judgments of a colleague’s ability and willingness to provide task-relevant input,
affective judgments related to the affective rewards of a social relationships have also been
linked to the formation of task advice ties, such that people tend to seek out for task-related input
those with whom they experience positive affect (Bales, 1958; Hinds et al., 2000; Homans, 1961;
Krackhardt, 1999; Krackhardt & Stern, 1998; Roethlisberger & Dickson, 1939; Slater, 1955).
While we have clear theory and evidence regarding the criteria people tend to use as they
develop task advice relationships in an organization, we lack understanding of the consequences
for performance of building advice ties based on the positive emotions experienced in a social
relationship with a colleague. I propose that using positive affect as a criterion for task-advice
ties can be beneficial for individual performance. More precisely, I argue that seeking out for
task-related advice colleagues with whom an individual experiences an emotional state of
positive activation may increase the probability of successful task execution by stimulating the
advice seeker to engage in task-oriented effort.
Theoretical foundations for this argument rest in the microsociology of social
interaction advanced by Collins (1981). Collins argues that emotional energy is the main
motivating force in social life. It provides the impetus for working, consuming, investing, and
any other human activity. An individual experiences an increase or a decrease in emotional
energy based on how social encounters unfold through interaction rituals. This is how the
network of social relationships in an organization drives the energy that people feel at work
(Cross, Baker, & Parker, 2003). An individual who negotiates social interactions successfully
acquires an increment of positive emotional energy (Collins, 1981). “High levels of emotional
energy in an individual consist of feelings of enthusiasm and confidence; low levels are
manifested as apathy and depression” (Collins, 1993: 211). People preferentially pursue social
5
encounters that increase their emotional energy, and the confidence and enthusiasm an individual
derives from social interactions stimulates action based on the expectation that a desirable future
can be brought into the present (Barbalet, 1998; Collins, 1993).
Collins’ notion of emotional energy portrays relational choice and human activity as the
result of a subjective feeling of confidence and enthusiasm that an individual derives from social
interactions. Applied to advice seeking and individual performance, this argument implies that
people will seek colleagues out for task-related advice based on the feeling of excitement
experienced in interaction with them. Because emotional energy is synonymous with motivation
for action, in the context of relationships formed to seek out task-related advice, emotional
energy may serve as a stimulus for task-oriented action, hence boosting task performance, under
the assumption that engagement and effort increase the likelihood of executing tasks
successfully.
Research on the psychology of affect adds precision to Collins’ definition of emotional
energy. Affect refers to consciously accessible feelings (Fredrickson, 2001), and can be defined
in terms of two dimensions: valence and arousal (Diener & Emmons, 1984; Russell, 1979, 1980;
Watson & Tellegen, 1985). Valence (or hedonic tone) is a subjective feeling of pleasantness and
unpleasantness. Arousal is a subjective state of feeling activated or deactivated. The positive
affective resources an individual draws from interactions with another person, therefore, can be
characterized in terms of varying degrees of activation. Positive valence with high activation
represents an excited, enthusiastic mood. This is the affective state that Collins (1981, 1993)
refers to as emotional energy. A positive mood with high activation corresponds to a discrete
emotion of excitement (Cropanzano, Weiss, Hale, & Reb, 2003; Izard, 1991), which is a
subjective state of “feeling energized” based on the expectation of future rewards (Izard, 1991;
6
Lawler & Yoon, 1996). Positive valance with neutral-to-moderate activation represents instead a
pleasant, contented, happy mood. A positive mood with neutral-to-moderate activation
corresponds to a discrete emotion of enjoyment (Cropanzano et al., 2003; Izard, 1991), which is
a subjective state of “feeling gratified” due to rewards received through the interaction (Izard,
1991; Lawler & Yoon, 1996).
I propose that the level of activation in the positive affect that an individual draws from
social interactions is consequential for task performance due to its effects on task engagement
and effort. Collins emphasis on emotional energy—rather than any generic form of positive
emotion—as an engine of human activity is echoed by research in psychology and organizational
behavior. In general, high-activation emotions (both positive and negative) are associated with
higher levels of engagement and effort than low-activation emotions (Carver, 2004). More
specifically, positive emotions with neutral levels of activation, such as pleasantness, signal that
a goal has been reached and further effort is unnecessary; by contrast, positive high-activation
emotions, such as excitement, imply eagerness to achieve (Buck, 1988; Carver, 2003). In work
contexts, job engagement is thus defined as a motivational state of high levels of energy (Rich,
Lepine, & Crawford, 2010), whereby “goals are chosen and behavior is energized and directed in
relation to envisaged positive outcomes” (Warr & Inceoglu, 2012: 618), with energetic activation
serving as an antecedent of intrinsic motivation (Quinn, Spreitzer, & Lam, 2012). Consistent
with these insights, Seo, Bartunek and Feldman Barrett (2004; 2010) proposed theoretically and
documented empirically a direct effect of emotional activation on the amount of effort exercised
by an individual in pursuit of work goals.1 Adding further nuance to this result, Foo, Uy and
1
The evidsence Seo, Bartunek and Barrett (2010) provided for the effect of pleasantness on
motivation is problematic, due primarily to construct validity issues. In that study, persistence
(one of three components of motivation) was measured in terms of how long an investor held on
7
Baron (2009) found high-activation positive affect (measured as “enthusiastic,” “attentive,”
“proud,” “interested,” and “inspired”) to predict individual orientation toward the future and
effort toward tasks not immediately required. This result is consistent with the notion of
excitement as a motivational state based on the expectation of future rewards (Izard, 1991;
Lawler & Yoon, 1996), and the idea that confidence and enthusiasm an individual derives from
social interactions stimulates action based on the expectation that a desirable future can be
brought into the present (Barbalet, 1998).
Applied to task-advice relationships, these arguments paint a consistent picture in which
advice ties eliciting high-activation positive affect result in higher levels of task engagement and
effort than task-related relationships eliciting positive emotions with lower levels of activation.
Enjoyable task-advice relationships may boost psychological resilience and emotional well-being
(Fredrickson, 2001; Fredrickson & Branigan, 2005), but do not necessarily stimulate an
individual to take task-oriented action, because pleasantness, contentment and happiness may
signal that a desired end state has been reached and no further action is required. By contrast,
engagement and effort have clearer theoretical implications for job performance (Rich et al.,
2010).
Together, these insights suggest that the tendency to seek out for advice colleagues that
elicit feelings of excitement (a propensity I label excitement-based advice seeking) will stimulate
greater task engagement and effort than the tendency to seek out for advice colleagues that elicit
enjoyment (enjoyment-based advice seeking). To the extent that engagement and effort have
to a stock position. This variable could measure disengagement from the task, rather than
persistence. Indeed, persistence was negatively predictive of effort, signaling a potentially
negative indirect association between pleasantness and engagement with the task.
8
greater likelihood of producing performance gains than gratification and contentment, these
predictions follows:
Hypothesis 1: Ego’s propensity to seek advice from alters based on how exciting
social interactions with them are (excitement-based advice seeking) is positively
associated with ego’s task performance.
Hypothesis 2: Excitement-based advice seeking has a stronger association with ego’s
task performance more than ego’s propensity to seek advice from alters based on
how enjoyable social interactions with them are (enjoyment-based advice seeking).
METHOD
Site and sample
To test these hypotheses, I conducted a web-based survey of information technology
salespeople in a large, multinational computer and IT consulting corporation headquartered in
the United States. The study concerned the performance of salespeople operating in teams that
the company created with the explicit purpose of integrating hardware, software, and other
technology services to meet wide-ranging client needs in a coordinated and comprehensive
manner. Clients represented a wide range of industries, including public sector not-for-profit
organizations, as well as retail, service, and manufacturing companies. Salespeople serving these
clients were typically engineers, by training. They performed a variety of roles, ranging from
information technology specialists to relationship managers. Providing clients with integrated
products and services depended on effective knowledge sharing and coordination among team
members. Consequently, although salespeople’s performance was evaluated based on attaining
9
their individual sales quotas, doing so required them to draw from the specialized expertise and
client knowledge of their teammates. The need to integrate others’ knowledge to accomplish
individual goals provided the rationale for creating these particular team structures. Because of
the technical complexity of products and services and the size of clients requesting them, the
sales teams had standing supplier relationships with clients, rather than short-term project-based
engagements. Salespeople serving a client were not necessarily co-located, although most were
located within the same geographical area within the United States.
The study was sponsored and supported by senior management within the organization,
including the executive vice president of sales operations, who provided detailed information
regarding the structure, dynamics, and composition of sales teams. Their insights helped to select
and word the survey items so that respondents would accurately and consistently interpret them,
given their work context. The selection process for participants in the survey included several
steps. First, three executive vice presidents of regional sales in the United States generated a list
of sales teams that they deemed to be representative of either high-performing and lowperforming teams in their region. Out of the population of 800 sales teams, the executive vice
presidents selected 60 teams for the study, with equal proportions of high- and low-performing
teams. Rather than outliers in the distribution of team performance at the company, high- and
low-performing teams in the sample were representative of large swaths of the company’s team
population. In this organization, differences in team-level performance stemmed primarily not
from differences in the level of competence of team members, but rather from exogenous
characteristics of the client organization, such as industry competitive dynamics and the lifecycle of the client relationship, with new clients in fast-growing industries providing a better
platform for sales than long-standing clients in mature industries. Stratifying the sample of
10
salespeople by current team performance was intended therefore to yield a sample representative
of the variability of team environments in which salespeople operated in the company.
The leaders of the sampled teams were contacted and 53 agreed to participate, ranging
in size from 4 to 47 (mean = 17.01, S.D. = 10.93). Team leaders, in turn, identified a total of 898
salespeople in their teams. An invitation to participate in the web-based study was emailed to all
sampled salespeople, who received a personalized access code to the survey website with a
guarantee of confidentiality of individual survey responses. This invitation was followed up by
emails from the team leader as well as phone calls by senior researchers within the organization.
Of the sampled salespeople, 693 returned at least partially completed questionnaires. I excluded
from statistical analyses respondents who did not rate any team member. Salespeople who filled
out the questionnaire but left the organization before the end-of-year performance evaluation
were also excluded from the analyses. The final sample included 430 salespeople. There were no
significant differences between respondents and non-respondents along characteristics such as
tenure in the organization, formal role, or gender. Likewise, among those who did fill out the
survey, there were no significant differences, along either survey or non-survey measures,
between respondents who stayed at the company and those who left before the year-end
performance evaluation.
Network data
The web-based survey provided respondents (egos) with a roster of all members of their
sales team (alters). Respondents were asked to check off members of their team that they knew.
This question was particularly relevant to large teams, where not all salespeople knew each
other. Teammates that a respondent did not know were excluded from subsequent survey
questions. For each team member that salespeople had checked off as someone they knew,
11
respondents were asked to indicate their agreement or disagreement with each of several
statements on a five-point scale (1 = strongly disagree, 5 = strongly agree). Task-related advice
was measured with the item “I turn to this person for information or advice before making an
important decision about this account”. The items “Interactions with this person are enjoyable”
and “Interactions with this person are energizing” measured, respectively, the extent to which
ego experienced positive affect with neutral-to-moderate activation (enjoyment) and highactivation positive affect (excitement) when having social interactions with each alter. Company
executives and salespeople who pre-tested the survey preferred the term “energizing” over the
term “exciting” because the latter could be associated with the notion of arousal, which was
deemed as potentially awkward and misleading in a professional context.
To account for people’s demonstrated tendency to seek out for advice colleagues they
deem competent at the task (Borgatti & Cross, 2003; Casciaro & Lobo, 2008; Hinds et al., 2000),
I also included the question “This person is very competent at his/her job,” which measured the
respondent’s competence evaluations of each member of the team. As with the other network
questions, the wording was determined based on input from company executives, who requested
the addition of the qualifier “very” to prevent, or at least limit, the censoring of responses at the
high end of the response scale. I also wished to account for the known effect of perceived
accessibility on the likelihood of seeking a colleague out for task advice (Borgatti & Cross, 2003;
Nebus, 2006). To measure a salesperson willingness to share knowledge, company executives
suggested the item “This person volunteers information that helps the team better serve the
client.”
Survey respondents provided a total of 8327 ratings of team members.
Overview of modeling strategy
12
Testing this study’s hypotheses entailed two estimation steps. Simply put, the first
estimation step computes four regression lines (slopes) for each ego based on the association
between ego’s four ratings of his teammates (enjoyment, excitement, competence, and willing to
share knowledge) and ego’s report on turning to each teammate for task advice. Figure 1
illustrates this estimation step. This respondent, Bob, rated each of his 12 teammates for how
exciting and enjoyable he found interactions with them, and how competent and willing to share
knowledge he perceived them to be. Bob also reported on how much he turned to each of those
12 teammates for work advice. The first estimation step uses these ratings to estimate four
regression lines for Bob, where the y axis is the rating for turning to a teammate for advice, and
the x axis is, respectively, the rating for excitement, enjoyment, competence, and willingness to
share that Bob gave to each teammate. Figure 2 illustrates how variability in the slope
coefficients this estimation produces for different respondents in the sample.
The model estimating these regression lines is a three-level model. Level 1 is the dyadic
level. It includes all ratings provided by the 430 salespeople in the sample about their teammates
and estimates overall (not individual) slopes for excitement, enjoyment, competence and willingto-share ratings as predictors of advice seeking across all respondents. Level 2 is the individual
level. For example, Bob is the individual within which Bob’s 12 ratings of alters are nested.
Level 2 estimates both Bob’s intercept for advice seeking (which account for Bob’s general
tendency to go to colleagues for task advice) and Bob’s slope for each of the four ratings of alters
(for example, the slope for excitement represents the tendency for Person’s A advice ties to be
exciting relationships). Level 3 in the model is the team in which each individual was nested.
In step 2 of the estimation, the four individual-level slopes for each of the 430 sampled
individuals are used as independent variables predicting individual performance. The model
13
estimating individual performance is a two-level model. Level 1 is the individual. Level 2 is the
team within which individuals are nested. Below, I detail this analytical approach.
First estimation step: Modeling advice-seeking propensities
I used the 8327 dyadic ratings that respondents provided about their teammates to
construct individual-level measures of the propensity to form advice ties with alters that ego
perceived as, respectively, enjoyable, exciting, competent, and willing to share knowledge.
These four measures were estimated for each salesperson in the sample. To that end, I ran mixed
models with random intercepts and random slopes for each ego, as well as fixed effects for each
alter, an approach equivalent to Snijders and Kenny’s (1999). The random slopes for each ego
represent the competence-, excitement-, enjoyment- and sharing- based advice seeking displayed
by each individual, which are the relative tendency of an actor to seek others out for task-related
input (i.e., ego’s responses to the “turn to” survey item) based on ego’s subjective assessment of
their competence and their willingness to share knowledge, and of how enjoyable and exciting
interacting with them is, respectively. I estimated random-slope models because they yield the
best linear unbiased predictions for the slope coefficients for each ego. The random intercept for
ego and fixed effects for alter control for perceiver effects (an individual average tendency to
seek colleagues out for advice) and target effects (an alter’s average tendency to be sought out
for advice), consistent with Kenny’s (1994) social relations model. The mixed models have the
following functional form:
P(Yij= y)= a0 + at + ai + b1ix1ij + b2ix2ij + b3ix3ij + b4ix4ij + b5ix5ij +( ∑19
𝑙=6 𝑏𝑙 𝑥𝑙𝑖 + 𝑢𝑖 )+ eij
where:
yij is the “turn to” rating given by the ith ego to the jth alter,
14
(1)
a0 is the overall intercept
at is the random portion of the intercept for the tth team,
ai is the random portion of the intercept for the ith ego,
b1i is the slope on the competence rating for the ith ego (competence-based advice seeking),
x1ij is the competence rating given by the ith ego to the jth alter,
b2i is the slope on the excitement rating for the ith ego (excitement-based advice seeking),
x2ij is the excitement rating given by the ith ego to the jth alter,
b3i is the slope on the enjoyment rating for the ith ego (enjoyment-based advice seeking),
x3ij is the enjoyment rating given by the ith ego to the jth alter,
b4i is the slope on the willingness-to-share-knowledge rating for the ith ego (sharing-based
advice seeking),
x4ij
is the willingness-to-share-knowledge rating given by the ith ego to the jth alter,
b4i
is the slope on face-to-face communication for the ith ego,
x4ij
is the face-to-face communication rating given by the ith ego to the jth alter,
∑13
𝑙=1 𝑏𝑙 𝑥𝑙𝑖 + 𝑢𝑖 are the fixed effects for ego’s formal job role
eij is the random error, and
ai ~ N(0,ai2); at ~ N(0,aj2); b1i ~ N(b10, b12); b2i ~ N(b20, b22); b3i ~ N(b30, b32); b4i ~ N(b40,
b42); ul ~ N(0,u2).
The b1i, b2i, b3i, and b4i coefficients are the measures of competence-, excitement-,
enjoyment- and sharing- based advice seeking, respectively. The intercepts ai and at and the
slopes b1i- b4i are random because ego and team are random effects. b10 - b40 are the overall
slopes on the competent, excitement, enjoyment, and willingness-to-share-knowledge ratings.
The covariance of the random effects is unconstrained, which allows the random effects
parameters to be correlated.
The estimation of enjoyment-, excitement-, competence-, sharing-based advice seeking
included controls for the formal role performed by each salesperson in their team (
15
∑19
𝑙=6 𝑏𝑙 𝑥𝑙𝑖 + 𝑢𝑖 in equation 1), which might affect their patterns of interaction with other team
members due to varying degrees of task interdependence between formal roles. The company
defined fourteen formal roles in sales teams, including Client Executive/Manager, Client
Services Manager, Software Sales Representative, Business Development Executive, Sales
Representative, IT Architect, and Technical Specialist. The first estimation step includes thirteen
dummy variables for the formal role played by each ego in the sample, treating the “Other”
response option as the omitted category in the estimation. In supplementary analyses, I also
included dummy variables for the formal role of alters. The coefficients of dummy variables for
alters’ formal role were not significantly different from zero in any of the models, and their
inclusion substantially decreased sample size, because of missing data on the formal role of
alters who did not fill out the survey. To prevent inflating standard errors and increasing the
probability of Type II errors, I excluded dummies for alters’ formal role from the estimation.
I also wished to control for the impact of physical proximity on the probability of seeking
a colleague out for advice (Nebus, 2006). The data collection did not include measures of
physical or geographical proximity in the organization, partly because salespeople travelled
extensively, and had opportunities to meet each other at their clients’, making measures of
proximity in the confines of their organization less meaningful. But the network survey did ask
respondents to indicate which of the following methods of communication they used with each
alter: face to face (planned or unplanned); instant messaging; email; phone for 1-1 conversations;
multi-person conference call; and shared repository. I used a dummy variable for face-to-face
communication (planned or unplanned) as a proxy for physical proximity between ego and alter.
Overall, the first estimation step controls for four primary forms of non-independence of
observations: row effects (with ego random intercepts), column effects (with alter fixed effects),
16
dyad effects (with fixed effects for ego’s and alter’s formal role that account for unobserved task
interdependence), and cluster effects (with team random intercepts).
Second estimation step: Predicting individual performance
The dependent variable is this study was individual performance as measured by yearend performance ratings assigned to each salesperson by a line manager outside of the team
structure. These performance ratings were based mainly on the extent to which a salesperson
achieved individual sales-quota targets for the year. A minor component of performance
evaluations was a line manager’s assessment of how well a salesperson represented the products
and services of his division. As such, performance evaluations were based primarily on objective
financial results, and only marginally on subjective supervisor ratings, providing a relatively
unbiased measure of sales effectiveness at the individual level. Individual performance ratings
were expressed in integers, ranging from 1 to 4, with high values indicating better performance.
The enjoyment-, excitement-based advice seeking measures calculated with equation (1)
were independent variables in ordered logistic regressions predicting individual performance.
Competence- and sharing-based advice seeking measures from equation (1) served as two of the
controls in this second estimation step. I used an ordinal logistic model specification to test the
hypotheses because the dependent variable was categorical, ordered, and had four possible
outcomes (Long, 1997). With this specification, an underlying score is estimated as a linear
function of the independent variables and a set of cut points. The probability of observing
outcome Z corresponds to the probability that the estimated linear function, plus random error, is
within the range of the cut points estimated by the outcome:
𝛾𝑖𝑧 =P (κz−1 < β1b1i + β2b2i + β3b3i + β4b4i + …+ βkxki + ui ≤ κZ)
17
(3)
where b1i, b2i, b3i, and b4i are the estimates for competence-, excitement-, enjoyment- and
sharing-based advice seeking from equation (1), and ui is assumed to be logistically distributed.
Using maximum likelihood, the coefficients β1, β2,…, βk are estimated along with the cut points
κ1, κ2, …, κz−1, where Z is the number of possible outcomes.
To control for team-level non-independence of observations, I performed the ordered
logistic regression in equation (3) within a two-level generalized linear latent and mixed models
(GLLAMM) with the following functional form:
(𝑧)
𝑦𝑖 = log(
(𝑧)
𝛾𝑖
(𝑧)
1−𝛾𝑖
) = 𝛼 (𝑧) - (∑49
𝑡=1 𝛽𝑡 𝑥𝑡𝑖 + 𝑢𝑖 ),
𝑢𝑖 ~𝑁(0, 𝜎𝑢2 )
(4)
The GLLAMM model in equation (4) estimates individual performance with an ordered
logistical regression nested in team-level random effects. In alternative specifications, I ran
ordered logistic models predicting performance estimating either team-clustered robust standard
errors or including team-level fixed effects. The results of these analyses were entirely
comparable with those obtained with the models presented in the table below.
The GLLAMM model included a number of control variables. In addition to controlling
for the formal job role performed by salespeople on their team, I used the dyadic ratings to
construct measures to account for how each salesperson was perceived by his or her teammates,
on average, along the criteria of competence, enjoyment, excitement, and willingness to share
knowledge. For each of those network items, the measure
∑𝑛
𝑗=1 𝑟𝑗𝑖
(5)
𝑛
represents the average evaluation that each salesperson, i, received from his or her n team-mates
j. These measures account for the possibility that a salesperson’s performance may be a function
of her individual characteristics, as perceived by others, rather than her relational behavior within
18
the team. To account for the possibility that newer members of the organization or of the team
may be at a disadvantage in terms of both task-related knowledge and social capital, I controlled
for the length of time individuals had been members of the team (team tenure) with a four-level
variable: less than 1 year, 1–2 years, 2–3 years, and more than 3 years. Similarly, I controlled for
tenure in the company (company tenure) with a five-level variables: less than 1 year, 1–2 years,
2–3 years, 3–4 years, and more than 4 years.2 Because a number of salespeople supported
multiple teams simultaneously, I controlled for the percentage of time a salesperson devoted to a
team included in the sample (time percentage on team) with a four-level variable (less than 20%,
20% to 50%, more than 50%, and 100%), and with a count of the number of clients served by the
salesperson (number of clients). I also wished to consider the possibility that women may
achieve levels of performance systematically different from those of men in a technology
company such as the one in which the study was conducted, where 75 percent of the sales force
was male. For that reason, the models predicting individual performance also included a dummy
variable for gender (female).
RESULTS
First estimation step: Modeling advice-seeking propensities
Table 1 presents descriptive statistics and correlation coefficients for the variables
included in the mixed models estimating the propensity of respondents to seek advice from
colleagues based on their perceived competence, willingness to share knowledge, and the
experience of excitement and enjoyment when interacting with them. The correlation
2
Company management provided data on team and company tenure in the format described
above, rather than as continuous variables. Alternative specifications of the team tenure and
company tenure variables, including dummies for each level, did not alter the results.
19
coefficients among ratings of competence, willingness to share knowledge, enjoyment, and
excitement that an individual gave his teammates are relatively high, consistent with the wellknown halo effect, according to which interpersonal judgments along different dimensions tend
to co-vary, so that individual characteristics, such as intelligence, likeability, warmth, and
competence, tend to be highly correlated in people’s judgments of others (Ambady & Rosenthal,
1993; Dion, Walster, & Berscheid, 1972). The random intercepts for ego and fixed effects for
alter, as well the simultaneous estimation of the effects of all four ratings (Equation 1) partial out
sources of this common variation.
Table 2 shows that ratings of competence, excitement, enjoyment and willingness to
share knowledge all have positive and statistically significant effects on the probability of
seeking a teammate out for task advice. These results indicate that the tendency to seek advice
from a colleague is positively associated with how ego perceives a colleague’s competence and
willingness to share knowledge, as well as the excitement and enjoyment experiences by ego in
interactions with alters. Face-to-face communication—a proxy for physical proximity to a
colleague—was also positively related to advice seeking. This reliance on competence, positive
affect, willingness to share knowledge, and proximity as criteria for selecting sources of task
advice is consistent with existing theory and evidence (Casciaro & Lobo, 2008; Collins, 1993;
Hinds et al., 2000; Nebus, 2006), providing evidence for the internal and external validity of this
study.
In supplemental mixed models, I also tested the possibility that competence, excitement,
enjoyment and willingness-to-share ratings might have multiplicative, as well as additive, effects
on advice seeking. To that end, I estimated the effect of six two-way interaction terms between,
respectively, ratings of competence, excitement, enjoyment and willingness to share knowledge.
20
These two-way interactions were not statistically significant in any of the mixed models
predicting advice seeking.
Second estimation step: Predicting Individual Performance
In the second estimation step, I used ego’s random slope coefficients associated with
ratings of excitement, enjoyment, competence, and willingness to share knowledge obtained with
the mixed model in equation (1) to measure ego’s tendency to seek task advice from colleagues
ego based on feelings of excitement (excitement-based advice seeking) and enjoyment
(enjoyment-based advice seeking) experienced when interacting with colleagues.
Table 3 provides descriptive statistics and correlation coefficients for these predictors,
as well as all control variables, including competence-based advice seeking and sharing-based
advice seeking. Among the four measures of advice-seeking propensities, excitement-seeking
has the highest correlation with individual performance (r=.11, p<.05). The four propensity
measures also vary with regard to their mean, with sharing-based advice seeking more common
(μ=.45) than competence-based advice seeking (μ=.12). The standard deviations for the four
measures of advice-seeking propensities also indicate variability across respondents. For
instance, the standard deviation for the individual slopes measuring excitement-based advice
seeking (coefficient b2i in equations 1 and 3) was .12, ranging from a minimum of -.18 to a
maximum of .51. This variability in individual slopes indicates that, while on average
salespeople in the sample treated the excitement experienced in interactions with a colleague as a
reason to seek them out for advice, some salespeople valued that excitement more than others.
The correlation coefficients among the average evaluations of competence, willingness
to share knowledge, enjoyment, and excitement that an ego received from teammates (i.e., alters’
rating of ego) are high, consistent with the halo effect discussed above (Ambady & Rosenthal,
21
1993; Dion et al., 1972). These correlation patterns notwithstanding, the four team-level
evaluations of an individual have different patterns of association with performance: only an
individual’s competence, as perceived by team mates, correlates significantly with individual
performance (r=.17, p<.01). I also ran t-tests comparing the average performance and
competence evaluations of salespeople in high- and low-performing teams. There were minimal
differences in the individual performance of salespeople operating in high-performing teams
(μ=2.88) and in low-performing teams (μ=2.71) (t = 2.07, p < .05), and no difference in average
ratings of competence received by salespeople in the two groups, indicating that the selection of
teams for the study did not unduly bias the distribution of individual performance and
competence in the sample.
Table 4 presents the results of the two-level GLLAMM model estimating individual
performance with an ordered logistical regression nested in team-level random effects. The
coefficient for excitement-based advice seeking was positive and statistically significant.
Consistent with hypothesis 1, the greater a person’s tendency to seek out for task advice
colleagues with whom ego experiences feelings of excitement was positively associated with
individual performance. By contrast, the results provide no evidence for enjoyment-based advice
seeking having a significant impact on task performance. A test of two linear hypotheses rejected
the null that the coefficients for excitement-seeking advice seeking and enjoyment –seeking
advice seeking are equal (p < .05), supporting Hypothesis 2, which predicted that excitementbased advice seeking would have a stronger positive association with task performance than
enjoyment-based advice seeking.
These results indicate that having advice ties with colleagues who elicit feelings of
excitement—emotional energy—is distinctly predictive of individual performance, even after
22
accounting for an individual’s tendency to seek advice from colleagues based on their
competence, willingness to share knowledge and enjoyment of interactions with them. In
supplementary models not included in Table 2, I tested for the possibility that the four adviceseeking variables might interact with the other three advice-seeking propensities to influence task
performance. None of these interaction terms was statistically significant. The lack of a
significant interaction effect between excitement-based and competence-based advice seeking, in
particular, indicates that the performance improvement induced by positive high-activation
positive affect was not contingent on the amount of task knowledge accessed through the
interaction. Yet, it is difficult to draw firm conclusions from null results concerning second-order
effects that are difficult to detect in multilevel models that include a high number of predictors.
Only one control variable had a significant effect on task performance: the average
competence ratings received by a salesperson from her team mates. This effect was to be
expected, as there should be consistency between teammates’ ratings of how knowledgeable a
colleague is about her job and that colleague’s job performance. By contrast, the average
enjoyment, excitement and generosity ratings received by a salesperson from his team mates did
not predict individual performance. The lack of a significant effect of average ratings of
excitement received by an individual rules out of possibility that the effect of excitement-based
advice seeking on individual performance may be an artefact of that individual’s own ability to
generate excitement in social interactions. Creating excitement does not predict individual
performance; seeking excitement does. This pattern of results held when the measures for alters’
ratings of ego along the competence, enjoyment, excitement and generosity dimensions were
tested one at a time, indicating that, in spite of the high correlation among these four group-level
assessments of an individual, it is specifically the team’s average evaluation of a team member’s
23
competence that predicts individual performance. The Variance Inflation Index (VIF) of these
four control variables ranged between 4.05 and 4.25, well below conventionally used rules of
thumb that indicate excessive or serious multicollinearity if any of the VIF are greater than 10
(Aiken & West, 1991; O'Brien, 2007).
In supplemental analyses, I checked the robustness of these findings by performing the
estimation using a conservative sample that included respondents who had provided a minimum
of five ratings of alters and whose ratings displayed a standard deviation greater than 0.5.
Restricting the sample to those who had provided at least five ratings accounted for potential bias
introduced by variability in the amount of relational data available for different individuals. By
restricting the sample to those whose ratings had a standard deviation greater than 0.5, I
addressed the possibility that invariant ratings might signal low response quality. The results
obtained with the full and the restricted sample were entirely comparable. In additional models, I
tested for team-level effects not with a two-level GLLAMM model with random effects for team,
but rather with a dummy variable denoting whether the team a salesperson operated in was
designated as high-performing by sales executives in the company. The results of these model
specifications were consistent with those in Table 4.
Finally, I wished to address the possibility that a salesperson’s extraversion may
underlie both excitement-based advice seeking and task performance. To that end, in
supplemental models, I included ego’s average rating of alters as enjoyable and exciting. These
variables serve as proxies for extraversion, since extraverts tend to enjoy interacting with people
and draw energy from social interactions. The inclusion of these measures did not alter the
findings. The lack of an effect of ego’s average ratings of alters as enjoyable and exciting also
24
addresses the possibility that strong performers may simply find colleagues more exciting, and
thus it rules out a potential form of reverse causality.
DISCUSSION
The purpose of this study was to investigate the performance implications of people’s
tendency to choose colleagues as sources of task advice based on the positive affect experienced
in interactions with them. I started from the premise that advice networks in organizations reflect
not only the instrumental resources someone can contribute to a task (e.g., knowledge), but also
the affective resources that people derive from social interaction (Bales, 1958; Hinds et al., 2000;
Homans, 1961; Krackhardt, 1999; Krackhardt & Stern, 1998; Roethlisberger & Dickson, 1939;
Slater, 1955). Although it is evident that task-oriented social interaction involves the pursuit of
both instrumental and affective resources, the effects of building advice networks based on the
affective content of social relationships with colleagues on people’s task performance have been
poorly understood.
The sole contributor to individual performance emerging from this study of adviceseeking behavior is an individual’s tendency to have task-advice ties that elicit high-activation
positive affect. This result persisted after accounting for the potential performance boost an
individual might gain by seeking out for advice especially competent colleagues or gaining
privileged access to their knowledge. Caution should be used, however, in interpreting the lack
of evidence for positive effect of competence- and sharing-based advice seeking (Cortina &
Dunlap, 1997; Cortina & Folger, 1998). Finding no support for the role of competence-based
advice seeking in individual performance, for instance, may be due to the specific distribution of
competence in the organization studied. In this complex and interdependent task environment,
25
the baseline of individual competence was high, and seeking advice from teammates with greater
competence relative to this selected set of colleagues may not have significantly increased the
task knowledge an actor got access to. Moreover, although we partly accounted for unobserved
task interdependence with fixed effects for ego’s and alter’s formal role, company management
did not provide us with a direct measure of the degree of task interdependence between a team’s
formal roles. As a result, our measure of competence-based advice seeking may not accurately
reflect the complementarity of competencies, or lack thereof, in a given dyad.
This interpretative caution noted, the documented effect of excitement-based advice
seeking on an individual’s job performance has three primary implications. First, it suggests that
the affective content of social relationships can serve a fundamental motivational function for
task-related action in organizations. Research in sociology, psychology and organizational
behavior consistently documents how a positive affective state of energized activation motivates
engagement and effort, and goal-oriented action toward achieving envisaged positive outcomes
(Buck, 1988; Carver, 2003, 2004; Collins, 1993; Foo et al., 2009; Quinn et al., 2012; Seo et al.,
2004; Seo et al., 2010; Warr & Inceoglu, 2012). In the context of task advice relationships, highactivation positive affect can thus stimulate engagement and effort toward assigned tasks (Rich
et al., 2010), and therefore increase the likelihood that they will be successfully performed.
Network research on the strength of ties has noted the motivational role of affect in task
interaction, evoking affective constructs, such as emotional closeness and trust, to explain what
induces social actors to invest time and effort to transfer useful and complex knowledge to their
colleagues (Hansen, 1999; Levin & Cross, 2004). These accounts portray affect as motivating
social actors to provide valued resources to others, but they largely leave to the background the
question of how affect experienced in social interaction may, in turn, stimulate actors to take
26
action toward work tasks. The present study underscores this very aspect of advice seeking in
organizations: affect as a motivational force that can enhance individual performance by
encouraging an actor to engage with and exercise effort toward work tasks, independent of the
task knowledge gained through the advice network.
Second, different forms of positive affect experienced in social interaction influence
task performance differently. Network research has typically treated relational affect as generally
positive or negative (Heider, 1958; Homans, 1961; Labianca & Brass, 2006; Labianca, Brass, &
Gray, 1998; Sampson, 1968), and has characterized the affective experience with constructs—
such as friendship, closeness, and liking—that lend themselves to ambiguous interpretation.
Friendship, for instance, takes many forms, ranging from enjoyable socializing to deep emotional
bonds of trust, which are mechanisms with potentially different outcomes. To address this
ambiguity, I measured directly two emotional responses with theoretical relevance to my
argument and clear definitions in psychological models of the affective experience. This
approach adds precision to traditional network measures of expressive ties, and it alleviates
concerns with the single-item measures to which network studies are often bound. Moreover, the
present findings indicate that a finer characterization of the emotional content of social
relationships is necessary to understand how positive affect may enhance task performance. In
this study, the pursuit of high-activation positive affect yielded unique performance advantages
over forms of positive affect with lower levels of activation. The documented link between
emotional activation and engagement and effort (Carver, 2004; Collins, 1993; Quinn et al., 2012;
Seo et al., 2010; Warr & Inceoglu, 2012) helps to explain the effects of excitement-based advice
seeking on performance. Positive emotions with moderate or neutral levels of activation, such as
enjoyment, signal that a goal has been reached and, therefore, further effort is unnecessary; by
27
contrast, positive high-activation emotions signal that a desirable end-state is possible, triggering
eagerness to achieve (Barbalet, 1998; Carver, 2003). Theories of affect in organizational
networks should therefore be sensitive to the motivational foundations of social interaction
between organization members. A failure to do so may yield misleading conclusions about the
relevance, or irrelevance, of affect for task-related action in organizations.
Third, these findings demonstrate that eschewing the pursuit of task competence for the
sake of positive affect may not be as irrational as it may appear at first glance. Prior research has
documented people’s tendency to form task-related ties based on their personal feelings for
coworkers, raising the possibility that such behaviour might prevent a significant reservoir of
task knowledge from being tapped in organizations (Casciaro & Lobo, 2008). I find that, at the
individual level, preferentially seeking for advice exciting colleagues may be rational, because
increasing one’s motivation to carry out work tasks can enhance an actor’s performance
independent of seeking out and accessing superior task competence. This may be the case for at
least two reasons. First, to the extent that organizations hire and retain people based on their
possessing basic abilities to perform assigned roles, choosing task partners with greater
competence relative to this selected set of coworkers may not increase significantly how much
task knowledge an actor gets access to. Second, seeking advice from colleagues with
significantly greater competence relative to other organization members may not stimulate an
actor toward applying their superior knowledge to task-oriented effort. For instance, particularly
knowledgeable colleagues may trigger upward comparisons that lower the advice seeker’s
confidence and energy (Tesser, Millar, & Moore, 1988).
The relevant question for organizations, then, is: what makes a social interaction
energizing? A number of theories in sociology, psychology and organizational behavior have
28
tackled this question (for a review, see Quinn et al., 2012). Extant theorizing and empirical
research indicate that it is not simply the existence and identification of social resources, but
rather the perception that desirable rewards are achievable that generates a motivational state of
energetic activation (Collins, 1993; Cross et al., 2003; Higgins, 1997; Izard, 1991). This
characterization of excitement as sparked by the prospect of future rewards begs further
questions concerning which social relationships are more likely to be associated with such
potential rewards. Classic social psychology suggests that task interdependence among
organizational actors may be predictive of high-activation positive affect, because the need to
work closely together to achieve superordinate goals enhances the intensity of emotional
responses within social interactions (Sherif, 1966; Sherif, Harvey, White, Hood, & Sherif, 1961).
Research on the emergence of affective commitment and micro orders has confirmed the link
between higher levels of interdependence and emotional attachment (Lawler, Thye, & Yoon,
2008). The recurrence and exclusivity of social interactions may increase the emotional energy
they generate (Collins, 1981, 1993). Individual differences may also explain variability in
relational excitement (Cross et al., 2003). The hallmark of extraversion, for instance, is lively
sociability—the enjoyment of others' company (McCrae & Costa, 1987). In the present study,
manifestations of extraversion were controlled for with measures of ego’s average rating of alters
as enjoyable and exciting, since extraverts tend to enjoy interacting with people and draw energy
from such social interactions. But while the results of this study cannot straightforwardly be
attributed to traits like extraversion, it is possible that individual differences may underlie how
social interactions operate as sources of emotional energy. And, certainly, “goals have an
energizing function” (Locke & Latham, 2002: 706). To the extent that goal-setting occurs
through social interaction with coworkers, the choice of interaction partners can influence self29
regulation in service of task performance (Latham & Locke, 1991). Despite these insights, the
distinct importance of excitement-based advice seeking demonstrated by this study demands a
deeper exploration of the relational and structural dynamics that determine the emergence of
high-activation positive affect in organizations.
Future research can also investigate the negative sphere of the affective experience in
intra-organizational networks. I focused on positive affective responses because my theory
concerned the social resources—and not the liabilities—that people seek out from interactions
with coworkers. Nevertheless, negative relational content can hinder individual and group
performance (Labianca & Brass, 2006; Sparrowe et al., 2001), and the tendency to either avoid
or seek out relationships that trigger negative high-activation emotion, such as fear and anger,
may play an important role in sustaining motivation.
The results of this study also reflect the specific distribution of task competence in the
organization I studied. It is conceivable that, in organizations populated with employees with
limited task competence, competence-seeking behaviour would emerge as essential to ensuring
adequate task performance. To the extent that organizations tend to select employees that have
basic task-relevant skills, however, widespread incompetence is both improbable and
unsustainable in organizations. The results of this study are therefore likely to emerge in other
common organizational contexts. The external validity of these findings is further corroborated
by research showing that interpersonal affective evaluations—such as personal liking,
pleasantness or energy—contribute to the formation of task-advice networks independent of
either organizational context or task characteristics, suggesting that these basic interpersonal
responses may not vary significantly by organizational and task environment (Casciaro & Lobo,
2008). The results of this study also replicate patterns of advice network formation documented
30
in a number of organizational contexts (for a review, see Nebus, 2006), suggesting that the
context I studied shared basic characteristics with other organizations (Johns, 2006).
These avenues for future research notwithstanding, this study sheds new light on the
motivational function that relational affect plays in task interaction and performance. Managers
often design organizational structures and systems to encourage interactions among coworkers
with the greatest amount of task competence. Without accounting for the affective content of a
social interaction, however, recognizing that a colleague has desirable task resources may not be
sufficient to enhance performance. In this sense, the affective motives that people apply to task
interactions do not simply “contaminate” their rational pursuit of work tasks. Rather, they
contribute to an individual’s ability to achieve task goals, and should therefore be considered an
indispensable component of any theory of task-oriented social action in organizations.
31
REFERENCES
Aiken, L. S., & West, S. G. 1991. Multiple Regression: Testing and Interpreting Interactions.
Newbury Park, CA: Sage Publications.
Ambady, N., & Rosenthal, R. 1993. Half a minute: Predicting teacher evaluations from thin
slices of nonverbal behavior and physical attractiveness. Journal of Personality and
Social Psychology, 64(3): 431-441.
Baldwin, T. T., Bedell, M. D., & Johnson, J. L. 1997. The social fabric of a team-based M.B.A.
program: Network effects on student satisfaction and performance. Academy of
Management Journal, 40(6): 1369-1397.
Bales, R. F. 1958. Task roles and social roles in problem-solving groups. Readings in social
psychology, 3.
Barbalet, J. M. 1998. Emotion, Social Theory, and Social Structure. Cambridge, UK:
Cambridge University Press.
Barnard, C. I. 1938. The Functions of the Executive. Cambridge: Harvard University Press.
Blau, P. M., & Schoenherr, R. A. 1971. The Structure of Organizations. New York: Basic
Books.
Borgatti, S. P., & Cross, R. 2003. A relational view of information seeking and learning in social
networks. Management Science, 49(4): 432-445.
Borgatti, S. P., & Foster, P. C. 2003. The network paradigm in organizational research: A review
and typology. Journal of Management, 29(6): 991-1013.
Brass, D. J., Glaskiewicz, J., Greve, H. R., & Tsai, W. P. 2004. Taking stock of networks and
organizations: A multilevel perspective. Academy of Management Journal, 47(6): 795817.
Buck, R. 1988. Human motivation and emotion (2nd. ed.). New York: John Wiley & Sons.
Carver, C. S. 2003. Pleasure as a sign you can attend to something else: Placing positive feelings
within a general model of affect. Cognition & Emotion, 17(2): 241-261.
Carver, C. S. 2004. Negative affects deriving from the behavioral approach system. Emotion,
4(1): 3-22.
Casciaro, T., & Lobo, M. S. 2008. When competence is irrelevant: The role of interpersonal
affect in task-related ties. Administrative Science Quarterly, 53: 655-684.
Collins, R. 1981. On the microfoundations of macrosociology. American Journal of Sociology,
86(5): 984-1014.
32
Collins, R. 1993. Emotional energy as the common denominator of rational action. Rationality
and Society, 5(2): 203-230.
Cortina, J. M., & Dunlap, W. P. 1997. On the logic and purpose of significance testing.
Psychological Methods, 2(2): 161-172.
Cortina, J. M., & Folger, R. G. 1998. When Is It Acceptable to Accept a Null Hypothesis: No
Way, Jose? Organizational Research Methods, 1(3): 334-350.
Cropanzano, R., Weiss, H. M., Hale, J. M. S., & Reb, J. 2003. The structure of affect:
Reconsidering the relationship between negative and positive affectivity. Journal of
Management, 29(6): 831-857.
Cross, R., Baker, W., & Parker, A. 2003. Whats creates energy in organizations? Mit Sloan
Management Review, 44(4): 51-+.
Diener, E., & Emmons, R. A. 1984. The independence of positive and negative affect. Journal
of Personality and Social Psychology, 47: 1105-1117.
Dion, K., Walster, E., & Berscheid, E. 1972. What is beautiful is good. Journal of Personality
and Social Psychology, 24(3): 285-290.
Foo, M. D., Uy, M. A., & Baron, R. A. 2009. How Do Feelings Influence Effort? An Empirical
Study of Entrepreneurs' Affect and Venture Effort. Journal of Applied Psychology,
94(4): 1086-1094.
Fredrickson, B. L. 2001. The role of positive emotions in positive psychology - The broadenand-build theory of positive emotions. American Psychologist, 56(3): 218-226.
Fredrickson, B. L., & Branigan, C. 2005. Positive emotions broaden the scope of attention and
thought-action repertoires. Cognition & Emotion, 19(3): 313-332.
Gargiulo, M., Ertug, G., & Galunic, C. 2009. The Two Faces of Control: Network Closure and
Individual Performance among Knowledge Workers. Administrative Science Quarterly,
54(2): 299-333.
Hansen, M. T. 1999. The search-transfer problem: The role of weak ties in sharing knowledge
across organization subunits. Administrative Science Quarterly, 44(1): 82-111.
Heider, F. 1958. The psychology of interpersonal relations. New York: Wiley.
Higgins, E. T. 1997. Beyond pleasure and pain. American Psychologist, 52(12): 1280-1300.
Hinds, P. J., Carley, K. M., Krackhardt, D., & Wholey, D. 2000. Choosing work group members:
Balancing similarity, competence, and familiarity. Organizational Behavior and Human
Decision Processes, 81(2): 226-251.
33
Homans, G. C. 1961. Social Behavior: Its Elementary Forms. New York: Harcourt, Brace and
World.
Ibarra, H. 1992. Homophily and differential returns: Sex differences in network structure and
access in an advertising firm. Administrative Science Quarterly, 37: 422-447.
Izard, C. E. 1991. The Psychology of Emotions. New York: Plenum Press.
Johns, G. 2006. The essential impact of context on organizational behavior. Academy of
Management Review, 31(2): 386-408.
Kenny, D. A. 1994. Interpersonal Perception: A Social Relations Analysis: The Guilford Press.
Krackhardt, D. 1999. The Ties that Torture: Simmelian Tie Analysis in Organizations. Research
in the Sociology of Organizations, 16: 183-210.
Krackhardt, D., & Stern, R. 1998. Informal Networks and Organizational Crises: An
Experimental Simulation. Social Psychology Quarterly, 51: 123-140.
Labianca, G., & Brass, D. J. 2006. Exploring the social ledger: Negative relationships and
negative asymmetry in social networks in organizations. Academy of Management
Review, 31(3): 596-614.
Labianca, G., Brass, D. J., & Gray, B. 1998. Social networks and perceptions of intergroup
conflict: The role of negative relationships and third parties. Academy of Management
Journal, 41(1): 55-67.
Latham, G. P., & Locke, E. A. 1991. Self-regulation through goal-setting. Organizational
Behavior and Human Decision Processes, 50(2): 212-247.
Lawler, E. J., Thye, S. R., & Yoon, J. 2008. Social exchange and micro social order. American
Sociological Review, 73(4): 519-542.
Lawler, E. J., & Yoon, J. 1996. Commitment in exchange relations: Test of a theory of relational
cohesion. American Sociological Review, 61: 89-108.
Lawrence, P. R., & Lorsch, J. W. 1967. Differentiation and integration in complex organizations.
Administrative Science Quarterly, 12(1): 1-47.
Levin, D. Z., & Cross, R. 2004. The strength of weak ties you can trust: The mediating role of
trust in effective knowledge transfer. Management Science, 50(11): 1477-1490.
Lincoln, J. R., & Miller, J. 1979. Work and friendship ties in organizations - comparative
analysis of relational networks. Administrative Science Quarterly, 24(2): 181-199.
Locke, E. A., & Latham, G. P. 2002. Building a practically useful theory of goal setting and task
motivation - A 35-year odyssey. American Psychologist, 57(9): 705-717.
34
Long, J. S. 1997. Regression models for categorical and limited dependent variables. Beverly
HIlls, CA: Sage.
Merton, R. K. 1957. The role-set: problems in sociological theory. British Journal of Sociology,
8: 106-120.
Mintzberg, H. 1979. The structure of organizations. Englewood Cliffs, N.J.: Prentice-Hall.
Nebus, J. 2006. Building collegial information networks: A theory of advice network generation.
Academy of Management Review, 31(3): 615-637.
O'Brien, R. M. 2007. A caution regarding rules of thumb for variance inflation factors. Quality
& Quantity, 41(5): 673-690.
Quinn, R. W., Spreitzer, G. M., & Lam, C. F. 2012. Building a Sustainable Model of Human
Energy in Organizations: Exploring the Critical Role of Resources. Academy of
Management Annals, 6: 337-396.
Rich, B. L., Lepine, J. A., & Crawford, E. R. 2010. Job engagement: Antecedents and effects on
job performance. Academy of Management Journal, 53(3): 617-635.
Roethlisberger, F. J., & Dickson, W. J. 1939. Management and the Worker. Cambridge, MA:
Harvard University Press.
Russell, J. A. 1979. Affective space is bipolar. Journal of Personality and Social Psychology,
37(3): 345-356.
Russell, J. A. 1980. A circumplex model of affect. Journal of Personality and Social
Psychology, 39: 1161-1178.
Sampson, F. 1968. A Novitiate in a Period of Change: An Experimental and Case Study of
Social Relationships. Cornell University: Doctoral dissertation.
Seo, M. G., Barrett, L. F., & Bartunek, J. M. 2004. The role of affective experience in work
motivation. Academy of Management Review, 29(3): 423-439.
Seo, M. G., Bartunek, J. M., & Barrett, L. F. 2010. The role of affective experience in work
motivation: Test of a conceptual model. Journal of Organizational Behavior, 31(7):
951-968.
Sherif, M. 1966. In common predicament: Social psychology of intergroup conflict and
cooperation. Boston: Houghton Mifflin.
Sherif, M., Harvey, O. J., White, J., Hood, W., & Sherif, C. 1961. Intergroup Conflict and
Cooperation: The Robber's Cave Experiment. Institute of Intergroup Relations:
Norman: University of Oklahoma.
35
Slater, P. E. 1955. Role differentiation in small groups. American Sociological Review, 20(3):
300-310.
Snijders, T. A. B., & Kenny, D. A. 1999. The social relations model for family data: A multilevel
approach. Personal Relationships, 6(4): 471-486.
Sparrowe, R. T., Liden, R. C., Wayne, S. J., & Kraimer, M. L. 2001. Social networks and the
performance of individuals and groups. Academy of Management Journal, 44(2): 316325.
Tesser, A., Millar, M., & Moore, J. 1988. Some affective consequences of social-comparison and
reflection processes - The pain and pleasure of being close. Journal of Personality and
Social Psychology, 54(1): 49-61.
Thayer, R. E. 1989. The Biopsychology of Mood and Arousal. New York: Oxford University
Press.
Thompson, J. D. 1967. Organizations in action. New York: McGraw-Hill.
Warr, P., & Inceoglu, I. 2012. Job Engagement, Job Satisfaction, and Contrasting Associations
With Person-Job Fit. Journal of Occupational Health Psychology, 17(2): 129-138.
Watson, D., & Tellegen, A. 1985. Toward a consensual structure of mood. Psychological
Bulletin, 98(2): 219-235.
36
FIGURE 1. Modeling advice-seeking propensity based on ego’s evaluations of alters
Advice
seeking
Bob
Criterion
Legend: Each dot represents how Bob rated each of her 12 teammates along one of
four criteria (enjoyment or excitement experienced in interaction with that
teammate, or the teammate’s competence or willingness to share knowledge)
FIGURE 2. Variability of advice-seeking behavior
Advice
seeking
Michael
Bob
Anne
Joel
Criterion
Legend: Each regression line represents a respondent’s tendency to seek out for
advice colleagues with a given rating for a given criterion. For example, if the
criterion is the excitement experienced in interactions with teammates, the line
represents a person’s excitement-based advice seeking.
37
TABLE 1
Means, standard deviations, and correlations of variables in first estimation step
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Variable
Mean s.d.
1
2
3
4
5
Advice seeking
3.05 1.14
Competence rating
3.91 .96 .52
Excitement rating
3.55 1.00 .55 .70
Enjoyment rating
3.89 1.00 .54 .72 .73
Willingness to share knowledge rating
3.72 1.05 .60 .69 .73 .73
Face-to-face communication
.15 .36 .11 .12 .13 .14 .13
Formal role 1
.11 .31 .16 .10 .06 .09 .08
Formal role 2
.09 .29 -.08 -.01 -.04 -.02 -.04
Formal role 3
.04 .19 .03 -.02 .01 -.02 -.02
Formal role 4
.03 .18 -.04 -.04 -.05 -.04 -.04
Formal role 5
.03 .18 .00 -.04 -.03 -.02 -.02
Formal role 6
.08 .26 -.05 .05 .02 .05 .04
Formal role 7
.05 .21 .04 .04 .03 .05 .04
Formal role 8
.03 .16 -.01 -.03 -.01 -.01 .00
Formal role 9
.01 .11 -.01 -.01 -.01 -.02 .01
Formal role 10
.05 .21 -.04 -.02 .01 -.01 -.01
Formal role 11
.02 .12 .03 .00 .03 .01 .04
Formal role 12
.06 .24 -.05 -.04 -.05 -.05 -.06
Formal role 13
.01 .08 .09 .06 .08 .06 .09
9
-.04
-.04
-.06
-.04
-.03
-.02
-.04
-.02
-.05
-.01
10
10 Formal role 4
11 Formal role 5
-.03
12 Formal role 6
-.05
13 Formal role 7
-.04
14 Formal role 8
-.03
15 Formal role 9
-.02
16 Formal role 10
-.04
17 Formal role 11
-.02
18 Formal role 12
-.05
19 Formal role 13
-.01
n = 8327.
Coefficients greater than .021 are significant at p<.05;
Coefficients greater than .031 are signficant at p<.01
38
11
12
13
14
-.05
-.04
-.03
-.02
-.04
-.02
-.05
-.01
-.06
-.05
-.03
-.06
-.04
-.07
-.02
-.04
-.02
-.05
-.03
-.06
-.02
-.02
-.04
-.02
-.04
-.01
15
6
7
8
.08
-.03
-.02
-.02
.00
.03
.00
.08
-.01
-.06
.06
.02
.00
-.11
-.07
-.06
-.06
-.10
-.08
-.06
-.04
-.08
-.04
-.09
-.03
-.06
-.06
-.06
-.09
-.07
-.05
-.04
-.07
-.04
-.08
-.02
16
17
18
-.02
-.01 -.03
-.03 -.06 -.03
-.01 -.02 -.01 -.02
TABLE 2
Three-level mixed model with random intercepts and random slopes for egos, random effects for
teams, and fixed effects for alters, predicting advice-seeking propensities
Model 1
Predictors
Competence rating
Excitement rating
Enjoyment rating
Willingness-to-share-knowledge rating
Face-to-face communication
Formal role 1
Formal role 2
Formal role 3
Formal role 4
Formal role 5
Formal role 6
Formal role 7
Formal role 8
Formal role 9
Formal role 10
Formal role 11
Formal role 12
Formal role 13
.12
.17
.21
.44
.11
.26
-.14
.14
-.08
.18
-.25
.04
-.05
-.12
-.10
-.07
-.07
.61
Log restricted-likelihood
-9532.38 ***
n = 8372
Standard errors are in parentheses
*p < .05; **p < .01; ***p < .001
39
(.02)
(.02)
(.02)
(.02)
(.03)
(.08)
(.09)
(.12)
(.11)
(.12)
(.08)
(.10)
(.13)
(.17)
(.11)
(.18)
(.09)
(.29)
***
***
***
***
***
**
*
TABLE 3
Means, standard deviations, and correlations of variables in second estimation step
Mean
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Individual performance
Competence-based advice seeking
Excitement-based advice seeking
Enjoyment-based advice seeking
Sharing-based advice seeking
Alters' rating of ego's competence
Alters' rating of ego as exciting
Alters' rating of ego enjoyable
Alters' rating of ego as wiling to share knowledge
Female
Team tenure
Time percentage on the team
Number of clients
Organizational tenure
27 Alters' rating of ego enjoyable
28 Alters' rating of ego as wiling to share knowledge
29 Female
30 Team tenure
31 Time percentage on the team
32 Number of clients
33 Organizational tenure
n = 430
Coefficients greater than .10 are significant at p<.05;
Coefficients greater than .12 are signficant at p<.01
40
s.d.
20
21
22
23
24
25
.78
.81
.79
-.12
.27
.02
-.23
.13
2.79
.12
.17
.21
.45
3.95
3.61
3.95
3.81
.27
2.74
2.47
3.54
4.73
.83
.16
.12
.15
.19
.49
.48
.46
.51
.45
1.16
1.41
1.27
.66
-.08
.11
.00
.02
.17
.09
.08
.07
.04
.06
-.06
-.05
.02
.00
-.43
-.19
-.02
.02
.04
.04
-.01
.03
.03
-.04
.06
-.32
-.02
.05
.07
.07
.06
-.11
-.07
-.04
-.14
.03
-.33
.04
.00
.00
.00
.05
.04
.01
.09
-.11
-.10
-.03
-.08
-.08
.01
-.01
.00
.08
.01
26
27
28
29
30
31
32
.80
.79
-.10
.16
.11
-.21
.04
.83
-.09
.20
.08
-.22
.10
-.06
.20
.16
-.27
.11
-.07
-.04
.09
-.01
.11
-.05
.31
-.18
-.07
-.10
TABLE 4
Two-level generalized linear latent and mixed model (GLLAMM) estimating individual
performance with an ordered logistical regression nested in team-level random effects
Predictors
Model 2
Excitement-based advice seeking
Enjoyment-based advice seeking
Competence-based advice seeking
Sharing-based advice seeking
Alters' rating of ego's competence
Alters' rating of ego as exciting
Alters' rating of ego enjoyable
Alters' rating of ego as wiling to share knowledge
Female
Team tenure
Time percentage on the team
Number of clients
Organizational tenure
Formal role 1
Formal role 2
Formal role 3
Formal role 4
Formal role 5
Formal role 6
Formal role 7
Formal role 8
Formal role 9
Formal role 10
Formal role 11
Formal role 12
Formal role 13
Log likelihood
n = 430
Standard errors are in parentheses
*p < .05; **p < .01; ***p<.001
41
.73
-.13
-.41
.09
.53
.10
-.14
-.26
.15
.03
-.02
.00
-.01
-.21
-.04
-.05
-.05
-.09
-.09
-.23
.18
-.02
-.18
.48
.11
.36
-503.244 ***
(.37) *
(.35)
(.30)
(.24)
(.16) **
(.16)
(.18)
(.16)
(.09)
(.04)
(.04)
(.04)
(.06)
(.17)
(.19)
(.19)
(.27)
(.18)
(.14)
(.17)
(.28)
(.27)
(.20)
(.34)
(.16)
(.56)