Accuracy in social network perception is broadly defined as the degree of congruence between the perceived and actual network of relationships formed via social interactions among people^{[1, 2]}. Accurate perception of organizational social networks, such as who talks with whom or who trusts whom, is a crucial determinant of organizational power and influence, and underlies effective leadership and organizational decision-making^{[1, 3]}. Research has linked organizational leaders' accurate perception of networks among employees and important external entities to accurate organizational problem diagnosis, critical ally selection, and judicious team member selection^[4–6]. Despite emerging consensus concerning its importance, only a handful of studies have investigated factors influencing accuracy in network perception, and only a very limited number of potential determinants have been explored^{[1, 2, 6–9]}. In the present study, we seek to extend current understanding about network perception by investigating the impact of holistic-analytic thinking styles on how accurately people perceive social network relationships and how effectively they decide on coalition selection.

Current psychological research suggests that individual thinking style varies along the holistic-analytic continuum. Holistic style involves a tacit recognition of the essential relatedness among people, objects, and events, and consequently, a habitual orientation to the contexts and web of relationships within which people, objects and events subsist^[10]. Analytic style, in contrast, involves the assumption that the world is a union of independent entities free of constraints, and consequently a tendency to focus on the internal properties of a target, in abstraction from the contextual and relational matrix^[10].

We posit that the different orientations that holistic thinkers and analytic thinkers have toward relationship and object have implications for social network perception. Holistic thinkers, when navigating the social landscape, are likely to automatically pick on relationships among people, objects, and events, and therefore, in theory should be better positioned to detect and encode social relationships. Analytic thinkers, however, cognitively attuned to independence and internal attributes, will be less likely to register the relational contexts and interdependence of experiential phenomena^[10]. As a result, analytic thinkers should be less accurate than holistic thinkers in perceiving social relationships, as relationships are not important in their conception of the world and would not be given much attention. Though intuitively reasonable, weather and how holistic-analytic thinking style affects social network perception has yet to be studied.

Our objective, therefore, in the present study, is to make two central contributions toward the existing literature on network perception and holistic-analytic thinking styles. First, we test the hypothesis that individual variations along holistic-analytic continuum affects how accurately people perceive social network relationships. Second, we further link holistic-analytic thinking to organizational decision-making by hypothesizing and testing its impact on coalition selection, via the mediation (at least in part) of network perception accuracy. In doing so, we advance scholarly understanding of cognitive accuracy in the domain of social networks and probe potential organizational implications of holistic-analytic thinking style. In the following sections, we first review the separate literatures on network perception and holistic-analytic thinking style. We then develop hypotheses on specific connections between these concepts. Finally, we test our hypotheses on a sample of 281 undergraduate students at a public university.

Existing research suggest that individuals with accurate knowledge of organizational social networks can gain greater influence and get things done more efficiently^[1]. Organizational leaders with more accurate perception of informal networks among employees are in a better position to form effective teams, diagnose organizational problems and lead organizational changes^{[4, 5, 11, 12]}. Despite the importance of accurate knowledge about informal organizational networks, research suggests that individuals are often not good at perceiving social network structure, and there are often substantial differences between the actual networks formed through social interactions among people and people's perception of these network relationships^{[4, 13–15]}. Though a critical piece of information for effective organizational leadership and decision-making, it has been investigated in only a handful studies why some people are more accurate than others in social network perception. These studies suggest a few key mechanisms underlying network perception accuracy, namely, reliance on cognitive schemas, attention to relationships, cognitive ability, and information availability. We briefly discuss each in turn.

First, reliance on network-related schemas reduces accuracy in network perception. Research suggests individuals often rely on default cognitive schemas (e.g., the transitivity principle or the small world principle) when encoding, representing, and inferring social relationships^{[6, 16–19]}. For example, the transitivity principle states: "If a transitive relation exists among three members of a set, such as $X$, $Y$, and $Z$, then the existence of a relation between $X$ and $Y$ $(XrY)$ and between $Y$ and $Z$ $(YrZ)$ implies that it will also exist between $X$ and $Z$ $(XrZ)$" in [6](p.348). Schemas, such as the transitivity principle, are intuitively appealing and permit efficient cognition of social network structure. Consequently, schemas offer the advantage of decreasing cognitive resources needed for social network perception. Cognitive benefits of their use notwithstanding, these default schemas are often inconsistent with actual social network patterns. In the $X$-$Y$-$Z$ network described above, there are many situations where $X$ and $Z$ are not related. However, relying on default schemas, individuals tend to "fill in the blanks" by perceiving transitivity in intransitive relations^[20–22]. In summary, automatic network-related schema use may result in less network perception accuracy.

Second, greater attention to social relationships enhances accuracy in network perception. Devoting more attention to social relationships leads to more effortful cognition and reduces the chance of automatic reliance on default schema, resulting in more accurate perception of social relationships^[9]. Indeed, research suggests that attention to social relationships is an important mechanism linking a few individual and situational factors to accuracy in network perception. Casciaro^[2] suggests that individuals with stronger needs for affiliation and achievement were more motivated to learn organizational social relationships, and therefor reported more accurate perception of the friendship and advice networks in their organization. Research also found that, in an effort to improve their disadvantaged situations, low-power individuals tended to allocate more attention to organizational social relationships and resultantly developed more accurate perception of those relationships^{[9, 23–25]}. In contrast, with less need to attend to social relationships for career advantages, individuals occupying places of organizational power were less accurate in perceiving organizational social networks^[2].

Third, cognitive ability to process network-related information impacts accuracy in network perception. Learning social relationships is a complex task. When the patterns of social relationships are inconsistent with the default schemas, accurate perception of social networks is even more cognitively demanding, because individuals have no short-cut but have to memorize and retrieve each of the dyadic relations (present or missing) as discrete pieces of information through brute force^{[6, 26]}. Positive affectivity has been found to enhance accuracy in perceiving friendship networks because positive feelings increase cognitive organization and flexibility, expanding individual capacity for discerning, sorting and integrating more network-related information^{[8, 27–29]}. Increased cognitive flexibility, in particularly, should enable individuals to more effectively process information on variations in social network patterns. Research also suggests that prior experience with intransitive social relationships enhances accurate perception of incomplete social networks^[6]. Prior experience with deviate network patterns not only sensitizes individuals to the inapplicability of default schema, but also may trigger the formation of unitized cognitive representations of alternative network patterns that people can readily draw upon to facilitate the organization of discrete dyadic information, leading to more efficient encoding and recall, and more accurate perception^[6].

Fourth, information availability influences network perception. Individuals have direct information on their own friendship, but only indirect information such as hearsay or assumptions pertinent to other's relationships. As a result, individuals are generally less aware of others' friendship, relative to their own^[30]. Occupying central positions in a network gives people access to information about social relationships in different parts of the network, leading to more accurate network perception^{[7, 8]}. Finally, demographic and social similarities improve perception of other's network relationships, as they increase information availability^[7].

In summary, a number of variables influence how accurately people perceive social networks. Using network-related schema while reducing cognitive demand may decrease accuracy in network perception. Greater attention to social relationships, network information processing capacity, and greater access to relationship-relevant information, on the other hand, may improve accuracy in network perception. The present study extends existing literature by investigating holistic-analytic thinking style as another determinants of accuracy in network perception. We theorize on the relationship that holistic-analytic thinking style may have with accuracy of network perception by explicating how holistic-analytic thinking style may be related to network-related schema, attention to social relationships and network-related cognitive capacity.

Different social historical conditions generate different thinking styles. East Asians and North Americans systematically vary across a number of perceptual and cognitive variables summarized in the model of holistic-analytic thinking style^[10]. Key variables on which holistic-analytic thinking styles are defined include locus of attention (field vs. parts), causal theory (interactionism vs. dispositionism), perception of change (cyclic vs. linear), and attitude toward contradiction (naive dialecticism vs. formal logic)^[31]. Locus of attention refers to the orientation of individual attention to parts or whole. Causal theory refers to the extent to which individuals perceive events as caused by internal attributes or as the result of complex interactions among external forces. Perception of change refers to the extent to which individuals see the phenomenal world as constituted of relatively stable essences or as a continuous temporal flux and flow. Finally, attitude toward contradiction refers to the extent to which individuals resolve contradictory propositions by rejecting one or by finding a dialectical common ground. Locus of attention and causal theory are the most relevant for network perception, on which we elaborate below.

Locus of attention refers to the orientation of individual attention to different aspects of the phenomenal world. For holistic thinkers, the world is characterized by interconnectedness among people, objects and events. For analytic thinkers, however, the world is in nature an aggregate of atomistic elements^{[10, 32, 33]}. To make sense, holistic thinkers habitually focus on relationships among elements and on the context to which the elements belong, while analytic thinkers attend to particular objects and their constituent properties, to the neglect of the context or its relations to other objects^{[10, 34]}. As a result, holistic thinkers tend to grasp the "whole picture" with ease, while analytic thinkers tend to dwell on the "parts". Additionally, holistic thinkers are more accurate at detecting relationships among objects but are less accurate than analytic thinkers when they have to isolate and analyze an object while ignoring the field in which it is embedded^[35].

Empirically, Abel and Hsu^[36] investigated locus of attention by asking participants to describe what they saw when presented with Rorschach cards. Holistic thinkers, i.e., the Chinese American participants, were more likely to give so-called "whole-card" responses, responses that consider all aspects of the card as a gestalt. Analytic thinkers, i.e., the European American counterparts, however, were more likely to give "part" responses, focusing on a single aspect of the card. Masuda and Nisbett^[37] provided additional evidence for different locus of attention between holistic thinkers and analytic thinkers. Participants were presented with realistic, animated scenes and asked to describe what they perceived. American participants focused on details of the focal object in the scene, while Japanese participants focused more on the background of the focal object, and noted relations involving different aspects of the scene.

Differences in causal theory constitute another important dimension of the holistic-analytic distinction. Casual theory refers to the habitual attributions people make in an effort to explain why people behave in certain way or why certain events occur. Holistic thinking style is interactionist in causal attribution, explaining events in terms of interconnections between actors within complex situational contexts^{[10, 31]}. Analytic thinking style, in contrast, is dispositionist, explaining behaviors as generated by actors' internal attributes^[38–41]. Further, being able to consider interconnections among a wide range of objects in the field, holistic thinkers tend to perceive a larger number of factors as potentially relevant to explaining a given event^{[31, 42]}.

Though holistic-analytic thinking styles have been investigated primarily as between-culture cognitive differences, evidence for within-culture variations also exists. Americans more interested in social interactions are more field dependent (even when adjusting for general intelligence) than are people with less social interest^{[35, 43]}. Korean oriental medicine students are more holistic than Korean students of other majors^[44]. A few recent studies have in fact conceptualized holistic-analytic thinking styles as within culture individual differences. Choi, et al.^[31] and Lechuga, Santos, and Garza-Caballero^[45] developed and validated the Analysis-Holism Scale in the U.S., Korea and Mexico, respectively. Monga and John^[46] explored how individual differences in holistic-analytic thinking styles moderate the effect of negative publicity information on consumer brand perception. Still, very little is known about the antecedents and consequences of individual differences in holistic-analytic thinking styles. Furthermore, its consequences for organizational behavior and decision-making are yet to be investigated. The present research adopts the individual difference approach and aims to determine if and how holistic-analytic thinking styles affect network perception and organizational decision-making.

Integrating network perception and thinking style research, we posit that holistic thinking style, relative to analytic style, is both a boon and bane to network perception. More specifically, we argue that while holistic thinkers' attention to field may enhance network perception accuracy, their interactionist causal attribution style may increase misperception of relationships where they are absent, resulting in less accurate perception.

As suggested earlier, attention to field captures the extent to which individuals attend to the whole and interconnections among the elements. While holistic thinkers habitually attend to relationships among elements in the field, analytic thinkers tend to focus on object properties, to the neglect of interconnections among objects^[10]. Our earlier review also suggests that paying attention to relationships is an important requirement for individuals to accurately recognize, encode, and recall information about social relationships^{[2, 9]}. It follows that holistic thinkers, being cognitively attentive to relationships, may have an advantage in network perception, relative to analytic thinkers.

Furthermore, holistic thinkers may have developed greater cognitive capacity in processing network-related information so that they are more accurate than analytic thinkers even when people are equally attentive to relationships. Research on experiential learning suggests that prior task experience increases efficiency in performing similar tasks. Through repeated experience with a task, individuals not only acquire knowledge content (i.e., knowledge related to the task), but also gain insights into the learning processes, that is, knowledge about how to assimilate or process particular kinds of information and apply it to new situations^[47]. Habitual attention to relationships should expose holistic thinkers to a greater variety of social network patterns and provide them with more practice in organizing and encoding network information. Experience of exceptions to the default rules should help people the limited applicability of those rules, and consequently reduce people's reliance on default schema. Such exceptions should also trigger and foster the development of alternative strategies to organizing information on dyadic social relationships into integrated structures that enable more efficient and effective encoding and recall^{[6, 48]}. Analytic thinkers, in contrast, may be at a disadvantage due to their lack of experience in processing social relationship information. When required to attend to social relationships, they may be able to restrain from automatic reliance on default schema but may have to encode and recall each dyadic relationship through brute mental force.

In short, habitual orientation to relationship should enhance holistic thinker's cognitive ability to accurately encode and recall social relationships. Supporting the above reasoning, a few studies of covariation detection found holistic thinkers to be more accurate than analytic thinkers in covariation judgments^{[34, 49]}. Therefore, we expect attention to field to be positively associated with accuracy in network perception.

H1: Greater attention to field will be associated with more accurate network perception.

Recognizing relationships as ontologically basic may increase attention to relationship, it may also increase the tendency to presume the existence of relationships. Indeed, holistic thinkers are interactionist, that is, they tend to see a lot of factors as potentially related and assume a complex matrix of interconnections as causally related to a given effect^{[31, 42, 50]}. Consistently, research suggests that holistic thinkers are more likely than analytic thinkers to note relationships or report co-variation between elements in the environment^{[34, 37, 49]}. As noted in our earlier review, individuals often misperceive the nature of social networks by assuming that a relationship exists when it does not^{[22, 51–53]}. Consequently, the inclination to assume complex interconnections in understanding and explaining worldly phenomena may make holistic thinkers more vulnerable to this error. We therefore have the following hypothesis:

H2: Greater interactionism in casual theory will be associated with less accurate network perception.

Social network research suggests that accurate network perception guide behaviors and decision-making in organizational settings^{[4, 5]}. We therefore suggest that holistic-analytic thinking style should also have consequence for organizational decision-making via the mediation of network perception. A critical problem confronted by many organizational actors is to develop effective coalitions so as to sway others' attitudes and opinions, or to promote new ideas^[1]. Selecting a maximally influential coalition, however, requires accurate perception of organizational social relationships such as who are friends with whom and who influences whom^[54]. Indeed, Janicik and Larrick^[6] have demonstrated a positive association between accurate perception of influence relationships and the effectiveness of subsequent coalition choice. We expect, therefore, holistic thinking to indirectly influence the effectiveness of coalition choice via the mediation of network perception. More specifically, attention to field may positively relate to the effectiveness of coalition selection through its positive association with accurate network perception. Interactionism in causal theory may negatively relate to effectiveness of coalition selection via its negative association with accuracy in network perception.

H3: Attention to field will positively influence the effectiveness of coalition choice via the mediation of network perception.

H4: Interactionism in causal theory will negatively influence the effectiveness of coalition choice via the mediation of network perception.

A related question is whether holistic-analytic thinking is directly associated with the effectiveness of coalition choice. Effective coalition selection requires people to reconstruct the structure of network relationships among all organizational members so as to select one or more individuals that can collectively influence the most people. Existing studies on network perception have focused on recognition and recall of dyadic ties as indicators of network perception. In the preset study, accuracy in network perception is also measured in terms of accuracy in recalling dyadic ties. Reconstructing the whole network, however, requires not just accurate encoding and recalling of dyadic relationships in the network (i.e., a part of the network), but also integrating the information in a way that reflects the whole network and the relative position of its members. The latter process is particularly important because individuals with the same number of ties to others are not necessarily equally positioned to exert influence or access resources^[55]. While accurate recall of dyadic relationships is a critical step in reconstructing the whole network, effective integration of dyadic information seems to involve a different cognitive process, particularly when the network deviates from the default cognitive schema^{[6, 48]}.

We suggest that holistic thinker's habitual attention to field may not only enhance the encoding and recall of dyadic relationships, but also provide an edge in integrating dyadic information. Attention to the field as opposed to a particular part entails instantaneously integrate and combine a broad set of contextual and relational features. Research shows that when presented with Rorschach cards, holistic thinkers were more likely to give "whole-card" responses, that is, responses based on all aspects of the card, or its Gestalt as a whole^{[10, 36]}. Norenzayan, et al.^[38] also reported that holistic thinkers categorized a focal object according to the number of common features it shares with a group of objects, rather than any single shared feature. This focus on the overall similarity based on combination of multiple shared features suggests an integration procedure that enables cognition of the whole simultaneously. Given attention to field's association with the capacity for combination and integration of a wide range of objects, relative to a more analytic focus on a single focal object and its attributes, we suggest that attention to field will also directly affect the effectiveness of coalition choice above and beyond the mediated effect via network perception.

H5: Attention to field will have a direct positive association with coalition choice.

In summary, we have advanced a model in which variations in two dimensions of holistic-analytic thinking, namely, attention to field and interactionist casual theory, relate to accuracy in network perception in opposite directions, and further influence coalition selection via the mediation of network perception. Figure 1 is a graphic presentation of the model.

Data for the present study were collected with an online survey and a lab task administered to undergraduate students in a public university in the United States. Participants were recruited via the online research participation system in the university. Undergraduate students are allowed to take part in faculty research to earn extra credits. 281 undergraduate students completed both the online survey and the network perception task. As presented in Table 1, 55% of the respondents are male; 77% are U.S citizens, while 23% are international students from East Asian (Chinese mainland; Hong Kong, China; Taiwan, China; South Korea and Japan). 49% are freshmen, 15% are sophomores, 19% are juniors, 15% are seniors and 3% are super seniors.

Upon arrival at the designated computer lab, participants first filled out an informed consent form, which explained the nature and procedure of the study, and assured them of information confidentiality and voluntary nature of the study. Participants then filled in an online questionnaire that consists of three major sections. The first section assessed participants' thinking style. The second section asked about participants' ego-centric social networks. The third section of the questionnaire focused on demographic information about the participants. The questionnaire took about 30 minutes to complete.

After completing the questionnaire, participants worked on a network perception task developed by Janicik and Larrick^[6]. Building upon the classical studies of network accuracy, the task presented participants with a novel organizational network, then measured their network perception accuracy^{[18, 20, 22, 30]}. Specifically, participants were instructed to learn the influence relations in a group of five scientists in a fictitious biotech company called "MedPro". Participants first received the following general instructions: "MedPro is a relatively new yet growing biotech company specializing in gene therapy. The management of MedPro has put together a five-member advisory board to oversee its new product development. The board consists of five of MedPro's top scientists: Steve, Bob, Doug, Mike, and Ken. Because they have worked together in the past, relationships have formed, and now certain scientists influence other scientists on the board. Your task in this experiment is to learn who influences whom among the members of the advisory board. You will have 1 min to study the information on the overhead projector. You will then be asked several questions. Do not take notes or write anything down on paper."

Participants were then given one minute to study the influence relations listed in Figure 2. The text in Figure 2(A) was displayed on a large screen in the front of the computer lab for the participants to study. Participants were not allowed to take notes during this time. Figure 2(B) diagrams the actual pattern of the influence relations listed in Figure 2(A) but was not shown to the participants. After one minute had passed, the projector was turned off and participants were asked to complete a questionnaire that tested their ability to recall the influence relations among the board members. Ten yes/no questions were asked regarding who influences whom (see Appendix for the list of questions).

Following Janicik and Larrick^[6], participants were given a second questionnaire that contained only one question upon completion of the network questions. The question described a scenario involving the board members in the influence relations and instructed participants to select two board members with whom they would like to form a coalition. The question was phrased as follows: "Suppose you are hired by MedPro as a marketing consultant. After extensive market analysis, you realize that drastic changes are needed in the development plan of one particular new drug. You have been informed that it is difficult to get the advisory board to endorse new ideas because of all the tense working relationships among its members. If possible, you would like to form a coalition with two of the board members in order to sell your ideas to the rest of the board. The board must agree to any changes in the development plan. You are planning on meeting with these two individuals to discuss your ideas. Remember, the five individuals are Steve, Bob, Mike, Ken, and Doug. Which two members would you select?"

The network-learning task took about 5 minutes to complete. Following completion of the task, a research assistant collected the materials and explained the purpose of the study. As hypothesized earlier, we expected locus of attention and causal theory to influence how accurately participants recall information on network relationships and how they select coalition partners. The appropriate coalition choices will be described below in the measure section.

Dependent variables. We had two dependent variables, namely, accuracy in network perception and effectiveness of coalition choice. As pointed out earlier, participants answered ten questions that tested their ability to recall the influence relations among the board members. The number of correct recalls was used as an indicator of accuracy in network perception. The score ranged from 0 to 10, with an average of 7.48 in this study.

Given that the goal for participants in forming a coalition was to "sell your ideas to the rest of the board, " the more board members participants could influence via the two selected board members, the more effective their choices would be; and the most effective choice could reach all members of the board via existing influence relations. We therefore calculated the number of unselected board members who could be reached indirectly through the selected board members to measure effectiveness of coalition choice. For example, if Steve and Ken were selected, all the other three board members would be influenced via these individuals (score = 3). In contrast, if Mike and Bob were selected, none of the remaining three would be influenced (score = 0). Thus, scores for effectiveness of coalition choice could vary from 0 (if Mike and Bob, or Steve and Doug were selected) to 3 (if Steve and Ken, or Doug and Ken were selected). Participants in the present study had an average score of 1.62.

Independent variables. We had two independent variables, namely, locus of attention and causal theory. Both were assessed using the analysis-holism scale developed and validated by Choi and his colleague^[31]. The scale has altogether 24 items capturing four dimensions of differences between holistic thinkers and analytic thinkers, with six items for each dimension. On a 5-point scale ($1=$ Strongly Disagree, $2=$ Disagree, $3=$ Neutral, $4=$ Agree, $5=$ Strongly Agree), participants rated their degree of agreement or disagreement with each of those statements.

Sample items for locus of attention include "The whole, rather than its parts, should be considered in order to understand a phenomenon." "It is more important to pay attention to the whole than its parts." and "The whole is greater than the sum of its parts." Higher scores indicate higher levels of attention to field. The scale has acceptable reliability, with Cronbach's alpha being 0.69^[56]. Sample statements for causal theory include "Everything in the universe is somehow related to each other." "Nothing is unrelated.", and "Everything in the world is intertwined in a causal relationship." Higher scores indicate higher levels of interactionism in causal attribution. The scale has acceptable reliability, with Cronbach's alpha being 0.80^[56].

Control variables. We collected data on a few control variables. Female is a dummy variable with 1 being female and 0 indicating male. We created the variable East Asian based on participants? country of origin, with 1 indicating an East Asian country or region (including Chinese mainland; Hong Kong, China; Taiwan, China; South Korea and Japan in the present study), and 0 indicating other countries. Years in college refer to participants' academic year at the time of the survey ($1=$ Freshman; $2=$ Sophomore; $3=$ Junior; $4=$ Senior; $5=$ Super senior). Participants with higher levels of general cognitive ability may be more efficient in encoding and recalling information and be more accurate in their perception. We therefore asked participants to report their GPA as a proxy for their general cognitive ability.

Finally, we controlled for two social network variables: Network size and experience with missing ties. As Janicik and Larrick^[6] suggest, prior experience with social relationships may influence network perception because they foster the development of network-related schematic knowledge that facilitate the encoding and recalling of network relationship. Participants identified a list of network contacts based on multiple relationships in the study and social domains, and then reported the strength of the relations among all possible pairs of contacts on a four-point scale ($1=$ Close; $2=$ Neither close nor distant; $3=$ Distant; $4=$ Negative). The number of network contacts was computed as an indicator of network size. Network size may indicate the pool of opportunities for seeing different types of relationships. Following Burt^[55], and Janicik and Larrick^[6], experience with missing ties was defined as experience with relationships where the two parties "are just acquaintances, rarely spend time together, do not know each other, don't like each other or don't enjoy one another's company". Such situation indicates instances in which the participant has ties to two contacts not directly related to each other^[6]. We counted the total number of missing relationships participants have among their network contacts, and then divided it by the total possible ties among participants' ego network contacts. Experience with missing ties captures the extent to which people has prior knowledge of missing ties, a situation inconsistent with the transitivity principle, and has been linked to more accurate perception of social networks inconsistent with transitivity schema^[6].

Table 1 provides the means, standard deviations, scale reliabilities (in parentheses), and intercorrelations of all variables in the analysis. As expected, accuracy in network perception was positively associated with attention to field ($r=0.18$, $p<0.01)$, and negatively associated with interactionism in causal theory ($r=-0.17$, $p<0.01$). Effectiveness of coalition was positively associated with accuracy in network perception ($r=0.34$, $p<0.01)$, and attention to field ($r=0.16$, $p<0.05)$, but was, unexpectedly, not significantly associated with interactionism. A closer look at the intercorrelations points to the possibility of correlative suppression effect between interactionism and attention to field in their correlations with accuracy in perception and effectiveness of coalition choice^[57]. Given that attention to field was positively correlated with both accuracy in network perception ($r=0.18$, $p<0.01$) and effectiveness of coalition choice ($r=0.16$, $p<0.05$), and that interactionism was positively correlated with attention to field ($r=0.32$, $p<0.01$), the hypothesized negative correlations that interactionism may have with accuracy in network perception and effectiveness of coalition choice could have been suppressed in bivariate situations when attention to field was not controlled for. We can expect the negative association between interactionist causal theory and accuracy of network perception to become stronger, and the insignificant negative association between interactionism and effectiveness of coalition choice to become significant, once attention to field is controlled for. Existing research on holistic-analytic thinking suggests that East Asians have a more holistic thinking style, while Europeans and North Americans have a more analytic thinking style. In the present study, being East Asian had a marginally significant positive association with interactionism but was not significantly associated with attention to field. Additional analysis suggests that East Asian participants ($N=64$) in our sample scored an average of 3.82 in interactionism and 3.42 in attention to field, whereas other participants $(N=217)$ scored an average of 3.69 in interactionlism and 3.32 in attention to field. The differences between East Asians and other participants were in the expected direction but were not statistically significant. Finally, consistent with existing findings (6), prior experience with missing ties was positively associated with accuracy in network perception ($r=0.23$, $p<0.01$) and with effectiveness of coalition choice ($r=0.18$, $p<0.01$).

We conducted two sets of hierarchical multiple regressions to test our hypotheses. The first set of multiple regressions predicted accuracy in network perception with attention to field and interactionism (See Table 2). The second set of regressions assessed if attention to field and interactionism influence the effectiveness of coalition choice via the mediation of network perception (See Table 3).

Predicting accuracy in network perception. Model 1 in Table 2 is the baseline model with all control variables, including female, East Asian, year in college, network size and experience with missing ties. The regression coefficient for network size was negative and significant, while the one for experience with missing ties was positive and significant. The other three control variables were not significant. Finally, as we pointed out earlier, GPA was considered as a proxy for general cognitive ability, and we had planned to include it in the models as an important control variable. However, with only 166 out of 281 participants reporting their GPA, 41% of the sample would have been dropped from the analysis if we were to include GPA. We decided not to include GPA after conducting two sets of analyses as described below.

First, we conducted $t$-tests of group means. Compared with participants with a GPA score, those without one were more likely to be female (difference $=0.29$, $p=0.00$), non-East Asian (difference $=-0.18$, $p=0.00$), and freshman (difference $=-1.76$, $p=0.00$). Participants without a GPA value also scored higher in attention to field (difference $=0.12$, $p=0.03$) and reported marginally more effective coalition choice (difference $=0.19$, $p=0.09$). However, the two groups did not differ significantly in accuracy of network perception (difference $=-0.03$, $p=0.45$), locus of attention (difference $=-0.07$, $p=0.16$), network size (difference $=0.02$, $p=0.49$), and experience with missing ties (difference $=0.03$, $p=0.19$). Second, we repeated the analyses reported here with GPA controlled for and with the sample size dropped to 166. Analyses based on 166 participants yielded results that were, across the board, identical to those reported in this paper1. Taken together, these results suggest that leaving GPA out would not compromise the validity of our analysis and results. We then examined the effects of attention to field and interactionism on accuracy in network perception by adding them to the baseline model one at a time. Supporting hypotheses H1 and H2 respectively, attention to field had a significant positive association (Model 3 in Table 2: $b=0.57$, $p<0.01$), and interactionism had a significant negative association (Model 2 in Table 3: $b=-0.72$, $p<0.01$) with accuracy in network perception. Model 4 included both attention to field and interactionism, yielding results consistent with those from Models 2 and 3. It must be noted, however, that the effects of attention to field and interactionism on accuracy in network perception were both stronger in Model 4 than in Models 2 and 3, as the coefficients for attention to field increased from 0.57 ($p<0.01$) to 0.86 ($p<0.01$), and the coefficients for interactionism changed from $-0.72$ $(p<0.01)$ to $-0.1.02$ ($p<0.01)$. These patterns were consistent with our earlier suggestion of suppression between attention to field and interactionism in their association with accuracy in network perception. Predicting effectiveness of coalition choice. Table 3 presents results on the effectiveness of coalition choice. We first estimated the baseline model with all control variables included (Model 1). Female was positively associated with the effectiveness of coalition choice. Network size had a negative association with the dependent variable, while experience with missing ties had a positive association with it. We then added attention to field and interactionism to the baseline model one by one. Attention to field had a significant positive association (Model 2: $b=0.28$, $p<0.05$) with effectiveness of coalition choice. Interactionism was only significant at 0.1 level (Model 3: $b=-0.20$, $p>0.1$). However, when entered into Model 4 together with attention to field, interactionism became significant (Model 4: $b=-0.33$, $p<0.01$). Similarly, the coefficient for attention to field also increased from 0.28 in Model 2 to 0.37 ($p<0.01$) in Model 4. These changes corroborated our earlier observation of suppression effects between the two variables in their association with effectiveness of coalition choice. In Model 5, accuracy in network perception was added to the baseline model, yielding a significant positive association with effectiveness of coalition choice $(b=0.17$, $p<0.01)$. Finally in Model 6, attention to field and interactionism were entered, together with the hypothesized mediator, accuracy in network perception. Internationalism became insignificant. Attention to field remained significant, but its coefficient decreased from 0.37 ($p<0.01$) to 0.25 ($p<0.05)$. Taken together, results from Table 2 and Table 3 supported hypotheses H3, H4 and H5. Attention to field and interactionism were related to effectiveness of coalition via the mediation of accuracy in network perception. While fully mediating between interactionism and effectiveness of coalition, accuracy in network perception was a partial mediator for attention to field. The results met all four conditions discussed by Baron and Kenny^[58] to establish mediation: 1) Both attention to field and interactionism were significantly associated with the hypothesized mediating variable, accuracy in network perception (Table 2, Models 2 to 4); 2) In the absence of accuracy in network perception, attention to field and interactionism were all significantly associated with the hypothesized outcome variable, effectiveness of coalition choice (Table 3, Models 2 to 4); 3) Accuracy in network perception was significantly related to the outcome variable (Table 3, Model 5); and 4) the associations that attention to field and interactionism had with effectiveness of coalition choice were significantly diminished or became insignificant once accuracy in network perception was taken into consideration (Table 3, Model 6). Further, Sobel tests showed that the indirect effects of attention to field and interactionism on effectiveness of coalition choice via accuracy in network perception were significantly different from zero^{[59, 60]}. Therefore, accuracy in network perception fully mediated between interactionism and effectiveness of coalition choice, and partially mediated between attention to field and effectiveness of coalition choice.

In summary, our analyses yielded three major noteworthy findings. First, two dimensions of holistic-analytic thinking styles had opposite effects on accuracy of network perception: While attention to field enhanced, interactionist causal theory decreased accuracy in network perception. Second, holistic-analytic thinking styles had consequences for organizational decision-making. While attention to field positively influenced effectiveness of coalition choice, interactionism had a negative influence. Third, accuracy in network perception fully mediated between interactionism and effectiveness of coalition choice but was a partial mediator between attention to field and coalition choice.

Our objective in this study is to integrate insights from research on holistic-analytic thinking styles with research on network perception to advance our knowledge about the roots of organizational power and effective decision-making. Two of the dimensions, namely, attention to field (as opposed to parts) and interactionism (as opposed to dispositionism) were the focus. Findings in this study support the proposal that individual differences in holistic-analytic thinking styles significantly influence perception of social relationships, and through perception, effectiveness of decision-making. First, we showed significant multifaceted relationships between holistic-analytic thinking and accuracy in network perception: While people with greater attention to field had more accurate network perception, those with interactionist inclination reported more errors in their perception. Further, we linked holistic-analytic thinking styles to organizational decision-making such as the selection of effective coalition, through the mediation of network perception. People with interactionist causal theory, compared with those with a dispositionist inclination, were less effective in identifying the most influential coalition, which was fully accounted for by their less accurate perception of influence relationship. In contrast, people with greater attention to field were able to select more influential coalition, and this relationship was only partially channeled via network perception, probably because attention to field not only enhances the encoding and recalling of dyadic relationships, but also promotes the integration of discrete information to obtain an understanding of the whole.

Our study makes three important theoretical contributions. First, our study broadens existing literature on thinking styles by linking individual differences in holistic-analytic thinking styles to network perception and illustrating its potential implications for organizational decision-making. With participants largely coming from the same culture, the study also demonstrates the usefulness of an individual differences approach to holistic-analytic thinking in understanding perception and decision making within the same culture. Further, the finding that attention to field and interactionist casual theory have opposite effects on network perception and coalition selection suggests the need to consider individual dimensions of the holistic-analytic thinking style separately in theory construction and empirical investigation. In our study, the correlation between attention and casual theory is moderate ($r=0.32$, $p<0.01$). In Choi, et al.^[31], the inter-correlations among the sub-dimensions of holistic-analytic thinking are rather small (range from $r=0.07$, insignificant, to $r=0.25$, $p<0.01)$. As such, treating the sub-dimensions of holistic-analytic thinking styles separately may not only contribute to theoretical precision, but also makes for more valid methodological practice.

There is a growing literature on cultural as well as individual differences in holistic-analytic thinking styles, and their implications for judgment and perception in various domains^[10]. However, research on the organizational implications of holistic-analytic thinking styles is very scarce. Linking holistic-analytic thinking styles to network perception and coalition selection, our study suggests the great potential for theoretical advancement and practical insights to investigate holistic-analytic thinking styles in relation to important organizational processes and outcomes, such as power and influence, leadership, and decision-making. Although the present study focused on undergraduate participants working on fictitious network scenarios, severely limiting the degree to which we can extrapolate the findings to real organizational settings, it does point to new directions that future organizational research can pursue.

Second, our study contributes to the literature on network perception. While the importance of accurate network perception in navigating social and organizational life has long been acknowledged^{[1, 54]}, existing literature offers limited insights into the individual, relational and contextual factors that mold individual perception. Our study constitutes a valuable addition to this literature by being the first to identify and test holistic-analytic thinking styles as important determinants of accuracy in network perception.

Third, our study responds to the call by social network researchers for more effort toward incorporating individual psychological differences such as motivation and cognition into network research^[61–63]. Researchers have used the pipe and prism metaphors to describe the function of social networks^[64]. As pipes, social networks channel the flow of a variety of information and resources. Research stressing the role of social networks in job search, knowledge sharing and creativity, and wellbeing are all exemplary of the pipe perspective. From this perspective, research has integrated individual psychology in two ways. One approach investigates how individual psychology influences the network positions individual occupy^{[62, 65, 66]}. The other focuses on how personality and social network position combine to influence important outcomes such as creativity and performance^{[67, 68]}.

As prisms, networks are viewed not just pipes conveying resources, but also informational cues on which people rely on to form perceptions of themselves and others^{[64, 69]} or to diagnose problems in work groups and organizations^{[4, 5]}. From this perspective, the integration of individual psychology with network research is natural and deep-rooted. As Sociologist Thomas stated in the early part of last century, "If men define situations as real they are real in their consequences.'' (in Colkart 1951: 81). Network cognition plays important roles in shaping individual interests and motivation^[55], and individual and organization actions that have consequence across the individual, group and organizational level of analysis^{[1, 4, 5]}.

The present study contributes to integrating individual psychology with the prism perspective of social network research. Research in this tradition is very limited, with many important questions yet to be addressed. For example, what kind of inference or perception people form about individuals or organizations that occupy different network positions or possess ego networks with different structure and composition? What factors influence the correspondence or discrepancies between objective network structure and perceived network structure? What are the consequences of these inferences and perceptions? The present study provides initial evidence for the role of holist-analytic thinking styles in shaping individual network perception and organizational decision-making based on such perception.

There is a broad agreement that organizational leadership and effective decision-making requires accurate perception of social networks and the ability to choose and marshal team members based on the perception of individual location in the organizational network^[4–6]. Organizational scholars have suggested that accurate network perception is a key cognitive player in the sphere of organizational power and influence^{[1, 54]}. Though the empirical focus on network perception is important for establishing its role in organizational leadership and effective decision-making, our research suggests that future exploration of the impact of thinking styles on accurate network perception and effective decision-making would further deepen our understanding in these domains. More broadly, the present study calls for greater research attention to the implications of holistic-analytic thinking styles in organizational settings.

Questions Used to Assess Participants? Recall of Influence Relationships

1. Does Bob influence Doug? __Yes__No

2. Does Ken influence Bob? __Yes__No

3. Does Steve influence Bob? __Yes__No

4. Does Doug influence Mike? __Yes__No

5. Does Ken influence Steve? __Yes__No

6. Does Doug influence Ken? __Yes__No

7. Does Steve influence Doug? __Yes__No

8. Does Mike influence Bob? __Yes__No

9. Does Bob influence Steve? __Yes__No

10. Does Mike influence Ken? __Yes__No

Note: Adapted from Janick & Larrick (2005), p.359.