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

net/publication/228472617

Communication Networks in Routine and Non-routine Tasks

Article · January 2009

CITATIONS READS

0 2,116

1 author:

Anssi Smedlund
Finnish Institute of Occupational Health
45 PUBLICATIONS 1,154 CITATIONS

SEE PROFILE

All content following this page was uploaded by Anssi Smedlund on 24 March 2014.

The user has requested enhancement of the downloaded file.

Helsinki University of Technology
Department of Industrial Engineering and Management

Communication Networks in Routine and Non-routine

Tasks

Anssi Smedlund, Helsinki University of Technology

Emily Choi, University of California, Berkeley

Working Paper No 2009/2

Espoo, Finland 2009
Helsinki University of Technology
Department of Industrial Engineering and Management

Helsinki University of Technology

Department of Industrial Engineering and Management
Laboratory of Work Psychology and Leadership
PO Box 5500
FIN-02015 HUT, Espoo, Finland
 COMMUNICATION NETWORKS IN ROUTINE AND NON-ROUTINE TASKS

ANSSI SMEDLUND1
Helsinki University of Technology
BIT Research Centre
P.O. BOX 5500
02015 HUT
FINLAND
Tel: +358 40 533 7452
[email protected]

EMILY CHOI
UC Berkeley
Haas School of Business
545 Student Services #1900
Berkeley, CA 94704
Tel: (510) 643-1408
[email protected]

Working paper

September 16, 2009

1
Corresponding author
 Communication Networks in Routine and Non-routine Tasks

ABSTRACT

This paper explores the communication network structures in routine and non-routine tasks in a

professional service firm. It also investigates the relationship between network centrality and

employee performance depending on the degree that an individual’s formal role is non-routine.

The communication structure to accomplish routine, day-to-day tasks differs significantly from

the communication structure to accomplish non-routine, ambiguous tasks. Employees who are in

a non-routine role benefit from centrality more than employees in a routine role. For employees

in non-routine roles, increased centrality predicts improved measured performance in

innovativeness and productivity, but counter-intuitively decreases performance measured by

demand from clients. The findings suggest that the central, most productive and innovative

employees in a professional service firm do not necessarily generate the most revenue.

Keywords: networks, work tasks, professional service firm

1
 Communication Networks in Routine and Non-routine Tasks

This paper builds on previous social network research that has shown a relationship between

network structure and performance (i.e. Uzzi, 1996; Hansen, 1999; Sparrowe et al., 2001; Burt,

2004; Cross and Cummings, 2004). By distinguishing between routine and non-routine tasks of

employees, this empirical paper aims to explore how centrality predicts performance, depending

on the employee’s role. We conducted this research in a professional service firm by using a

sociometric survey-based dataset of task-related communication. We also gathered subjective

performance indicators of the employees of the company.

In a professional service firm, work is highly knowledge intensive, because knowledge

and employee skills are both a resource and an outcome of the work. This makes day-to-day

routine work tasks highly dependent on active communication among the employees, not to

mention work tasks that are aimed to create totally new ideas. We expect that by exploring the

communication network structure in the professional service firm, and by identifying the best

performing employees embedded in them, it is possible to understand and improve the

management of knowledge-intensive organizations.

The communication network in an organization reveals structures that are either sparse or

dense. Sparse networks are filled with structural holes (Granovetter, 1973), where brokering

employees function as gatekeepers of flows of knowledge (Burt, 1992). Dense networks form

closures among the employees (Coleman, 1988), and every one is embedded in the network with

redundant ties (Uzzi, 1996). There has been a debate about sparse and dense network structures

and their effects on performance (Burt, 1992; Podolny and Baron, 1997; Hansen, 1999; Reagans

and Zuckerman, 2001; Reagans and McEvily, 2003). Densely connected networks limit the

2
inflow of diverse and fresh insights to the firm (Hansen, 1999; Ahuja, 2000; Reagans and

McEvily, 2003), but make it possible to facilitate the creation of value with mechanisms of

coordination, trust-building, and the production of sanctioning mechanisms (Coleman, 1988;

Coleman, 1990). Sparse structures with a large number of indirect ties create value for actors by

offering control and information benefits, and allowing the flow of complex and non-redundant

information (Burt, 1992; Hansen, 1999; Reagans and Zuckerman, 2001).

Experimental laboratory research conducted in the 1950’s and 1960’s shows evidence

that there is no optimal network structure, but many (Bavelas, 1951; Shaw, 1964). The optimal

communication network structure is related to the type of tasks. For example, sparse and

decentralized communication network structures are better in solving complex tasks, whereas

dense and centralized network structures are suitable for routine-like, simple tasks (ibid.).

Centralized networks channel information to a focal employee. The closer the others are to the

focal person, the faster the problem is solved. When the tasks become more complex, the

problems related to the task become unmanageable for the focal employees and their immediate

contacts to handle. Then the answer to the problem is sought from more distant sources, which

will make the communication network of complex tasks decentralized.

Although strong evidence has been presented that the relationship between individual

formal role and individual performance is mediated by centrality of the person in the network

(Ahuja, Galletta and Carley, 2003), it seems that previous social network research has tested the

employee’s position in communication networks without considering that the network structure

depends on the type of task.

In this paper, we assume that the nature of the tasks affects the communication network

structure. Routine tasks create a dense and interconnected communication network, whereas non-

3
routine tasks result in a sparse and decentralized structure. An employee’s position in the

network yields different performance results, depending on the network structure. Centrality in

general is important for performance, but depends on the match between the task and the

employee’s formal role. In the analysis, we interacted roles that require non-routine work with

centrality measures in routine and non-routine communication networks and found different

effects.

Next hypotheses for research are presented based on theory. Then in the methods part,

data collection and the variables for OLS regression are described. The results parts of the paper

include the description of the structural characteristics of the routine and non-routine networks,

and the results of the OLS regression analysis. Finally discussion of the results and limitations as

well as future research directions is provided.

4
 Theory

The flows of knowledge have been a popular topic in social network research (i.e. Hansen, 1999;

Reagans and Zuckerman, 2001; Hansen, 2002; Reagans and McEvily, 2003). Knowledge-related

studies have examined mainly how network structures transfer different types of knowledge

(Hansen, Nohria and Tierney, 1999; Reagans and Zuckerman, 2001). This paper does not

distinguish between different types of knowledge, such as explicit or tacit, but concentrates on

the communication network structures that are related either to routine or non-routine tasks of

employees. Communication relationships are some of the fundamental relationships between

employees besides such relationships as advice, friendship, support and influence (Ibarra, 1993).

Routine tasks have behavioral and cognitive definitions (Becker, 2005). The cognitive

view of routines is supported by Simon, March, Nelson and Winter (March, Simon and

Guetzkow, 1958 [1993]; Cyert and March, 1963 [1992]; Nelson and Winter, 1982). In the

cognitive view, routines are defined as knowledge that is embedded in standardized procedures

or rules (Nelson and Winter, 1982), organizational memory, truce in controlling intra-

organizational conflicts, or even normative targets of conduct (Cohendet and Llerena, 2003).

The behavioral approach to routines, considers routines as a behavioral regularity, a

recurrent interaction pattern (Becker, 2005). In this view, routines have a certain recurring

frequency and follow a sequence within a certain amount of time. Routine work also has a pre-

defined outcome. Established routines, in the behavioral sense, enable better coordination in a

company, provide stability to behavior, are sub-conscious and require limited cognitive resources

from the employee (Becker, 2004). Behavioral routines can also be defined as inclinations

towards certain type of behavior, when triggered by an external force (Becker, 2005).

5
 Most of the empirical research on routines has been done from the behavioral

perspective, because the cognitive dimension of routines is hard to operationalize and observe

(Lillrank, 2003). The definition of routine work differs from company to company, but it is still

rather easy to define them from the behavioral perspective by distinguishing them from non-

routine tasks. Non-routine tasks involve managing semi-structured or unstructured problems

(Pava, 1983), and are directed to something where the process is complex and the result of the

work is uncertain and unspecified. Salter and Gann (2003) provide a good analogy. In

engineering consultancy, routine work is to build a strip of freeway as non-routine work is to

build the Millennium Bridge in London.

Based on the previous literature, it can be stated that in routine tasks, employees are more

likely to communicate with those close to them who work in the same project to get the job done

as efficiently as possible, which results in a cohesive and centralized network structure. In non-

routine tasks, solutions for problems are more likely to be sought from more distant colleagues

and contacts accumulated from earlier projects, which results in a sparse and decentralized

network structure.

Centrality is the most commonly used structural measure in social network analysis,

because the central employees are the ones that are the most connected to others, and they are

most likely to possess large amounts of information, and are able to influence others more

effectively. Central employees affect the whole structure of the network by their communication

with a large number of other employees (Carley, 1991). According to the basic idea of social

capital (eg. Nahapiet and Ghoshal, 1998), more relationships is always better than less, and the

central employees in networks are more likely to perform better.

6
 The communication network structures in routine and non-routine tasks are

fundamentally different from each other. Employees embedded in these networks have different

kinds of benefits from centrality, which affects their performance. Central employees in routine

tasks are likely to act as focal individuals in projects who manage and integrate the work of

others, where as central individuals in non-routine tasks are likely to act as brokers and hubs of

new ideas.

H1a: Centrality in the communication network in non-routine tasks positively affects

performance

H1b: Centrality in the communication network in routine tasks positively affects performance

In a communication network for routine tasks that involves performing pre-determined,

sequential tasks that require precision, stability and reliability, and contains a low level of

uncertainty, central employees are in a better position to integrate and organize the work. They

are in control of the flow of information with dyadic ties to a large number of other employees,

which improves the solving of well-defined problems, but if the task presents complex or

ambiguous problems, then the centralized structure impedes solving the problem (cf. Bavelas,

1951; Blau and Scott, 1962; Shaw, 1964; Scott, 2003).

In the communication network for non-routine tasks, central employees are in an

advantageous position compared to their less central colleagues (cf. Burt, 1992). Non-routine

tasks involve high uncertainty and are ambiguous and unspecified by nature. When employees in

the organization seek out possible ideas and clues to solve their non-routine tasks, they reach out

to more distant sources of information, which results in a sparse and decentralized

7
communication network structure. Central employees in this communication network are the

ones that have an abundance of different kinds of information and have better access to new

ideas and new developments compared to their less central colleagues. Therefore, the central

employees in the communication network for non-routine tasks are brokers and gatekeepers of

new ideas, which will result in improved performance in their work (cf. Hansen, 1999; Burt,

2004).

Individual work roles also explain the employee performance in the organization (Ahuja,

Galletta and Carley, 2003). In this paper, the role characteristics are divided into routine and non-

routine tasks, and the employee roles in the case company tend to be either routine or non-

routine. In an established professional service firm, employees with routine roles engage mainly

in work in well-specified consulting projects by utilizing their own competence, company

procedures and databanks, and their experiences from earlier similar projects. The employees in

non-routine roles are likely to orchestrate the work of others, manage many projects

simultaneously, sell new consulting projects to clients, and develop and redefine firm-internal

processes and service offerings. Therefore, the employees in non-routine roles are involved in

solving more ambiguous problems compared to the employees in routine roles. The employees in

non-routine roles are more likely to benefit from a central network position in a non-routine

network, but less likely to benefit from centrality in routine tasks.

H2: Centrality in the non-routine communication network increases the performance of

employees with non-routine roles more than the centrality in the routine communication network

increases the performance of employees with routine roles.

8
 Methods

The research was conducted among the employees of an architects’ office in Northern Europe.

Architecture is highly knowledge-intensive work, but the competencies and processes in

architecture are well established and standardized. According to the CEO of the company, the

basic work of architects in routine roles in the business has remained quite unchanging for the

past 20 years, despite of the IT process innovations along the way, which have increased the

overall productivity of the design work.

The routine work of an architect in the company concerns drawings of buildings, parts of

buildings or public spaces and constructions for the clients. The work requires a Master’s degree

in architecture and application of certain IT-tools, processes and conventions in the field. The

work is project-based, which means that the work in the organization is managed in projects, and

that the employees keep a close track of their working hours per project. Billable design projects

for the client are separated from other types of projects. In 2007, the average percentage of

billable client work for the architects was 73%, and around one half of the architects billed over

80% of their total working hours from the clients. In addition to billable client work, the

employees in the company participate, depending on their formal role, in internal development

projects, marketing projects and more innovative building development projects, which require

sharing and developing new ideas.

The network data was gathered in 2006 and the performance data one year after that. The

purpose of the network questionnaire was to find out the routine and non-routine communication

network structures among the employees. The network survey questionnaire was detailed in

terms of direction and frequency of communication relationships, and took around 30-45 minutes

for the respondents to answer. A total of 84 out of the 93 employees answered the survey, but 5

9
of them left the non-routine question unanswered. Between 2006 and 2007 the company was

growing aggressively because of the construction boom in its market, and 26 new professional

architects were hired during this period. After the survey and before the end of 2007, 10

architects had left the company.

There were five formal roles of employees in the company. The majority of the

employees held a professional role (N=39), and their tasks were to concentrate on actual

architecture project work, managed by project managers (N=20) and senior project managers

(N=9). The top managers (N=6) were the original founders and owners of the company, while

the middle managers (N=7) were responsible for managing certain types of design projects (ie.

retail, sport arenas, interior etc.). The middle managers together with the top management

formed the management team of the company, where the strategic focus areas and organizational

development issues were discussed. In addition, there were also employers in administrative

roles (N=11), whose tasks were to provide accounting, payroll, IT support and front desk

services in the company.

In this paper, the employees working in professional roles (N=39) were considered to

perform routine-type work, and the employees in all manager roles (N=33) to perform non-

routine work. In the testing of the hypotheses, we indicate non-routine roles with a dummy

variable, 1 indicating non-routine and 0 otherwise.

Data collection

In the network survey questionnaire, sociocentric and egocentric data gathering methods were

combined. This was done by letting the respondents define their own networks from a roster of

names that included everyone in the organization, before answering detailed questions about

communication in routine and non-routine tasks. The survey was designed as a free-choice

10
survey with two-way directed questions (for the methodology, see Wasserman and Faust, 1994).

Demographic data, as well as the timesheets of the respondents and non-respondents were

acquired from the personnel records of the company.

Behavioral definitions of routines and non-routines were used, and to highlight the

differences in these tasks, exact wordings of the survey questions were reviewed and modified

several times by the authors and their colleagues before putting the survey online. Brief phone

discussions with a highly tenured professional made sure that the questions would be understood

correctly by the respondents. According to the open feedback gathered at the end of the

questionnaire, the wordings of the questionnaire were generally well understood among the

respondents.

Examples of routine tasks in the questionnaire included tasks that are delivered to the

client, are well specified in advance, recurring, and belong to the respondent’s line of expertise.

The non-routine tasks were defined through communication of ideas: the respondents were asked

to name those who they go to and those who come to them in “light bulb moments” at work. In

the non-routine task questions, there were no examples of certain types of non-routine tasks

because of their ambiguous nature.

In the questionnaire it was highlighted that the answers included every means of

communication (face-to-face, phone, email etc.), and that all the answers of the respondents were

subjective estimates of the actual communication. The frequency scale in communication was set

to options of 4) daily, 3) weekly, 2) once per month, 1) less than once per month, or 0=not at all.

One year after the network survey, the same web-based survey instrument was used to

measure the employee status in the organization in innovativeness. In this fixed-choice one-way

questionnaire (Wasserman and Faust, 1994) the respondent was not given a roster of names of

11
others working in the company, but the respondent was asked to name five others in their

organization in terms of 1) promoting new ideas and 2) being as a source of new ideas. Outlines

of the network questionnaire and the innovativeness colleague ratings are presented in the

Appendix.

Dependent variables

We measured three types of performance. The first was the number of billable hours from the

client. These were the working hours that the employees mark down in their timesheets and that

are later billed from the client based on the terms of the project agreement. The second measure

was project productivity, which we constructed from the timesheets. It was the count of billable

client projects divided by the count of billable hours. These first two measures are considered

objective performance indicators in this paper. The third performance measure was

innovativeness. It is a subjective performance indicator, because it is based on colleague ratings

of promoters and sources of ideas in the organization (Appendix).

We use both objective and subjective performance ratings because they complement each

other, showing two different sides of employee performance in the organization. Objective

performance ratings show the direct and concrete benefits of networking to the employees’

performance. They are not widely used in network research, because in companies where the

profits flow from the sales of products in the market, objective performance of the firm cannot be

directly drawn from the performance of the employees. In knowledge-intensive, professional

service-type work, objective performance ratings can be used (cf. Huselid, 1995; Huselid,

Jackson and Schuler, 1997). Huselid et al. (1997) have used net sales per employee as a measure

of productivity in their studies. This indicator applies well in the situation where the performance

12
of an employee aggregates directly to the performance of the company as a whole, due to the low

physical capital intensiveness of the business.

The dependent variable of employee count of billable hours used in this study is similar

to the indicator of Huselid et al. (1997). According to the interviews with the company’s top

management, billable hours are generally considered as a good indicator of performance in the

company’s market, since the profit of the firms in the business depend on the ratio of billable

hours to overall hours, and the companies in the field grow by hiring new employees. Also,

different companies in the market are compared to each other according to their gross profit

margins.

During the data gathering the architects’ office was overbooked with projects and every

professional in the company had as much project work in their hands as they could or wanted to

work on. This is shown in the timesheets by the number of overall hours worked during 2007 –

out of the 83 employees who had worked in every month of the year, 66 had a number of total

working hours greater than the national average. The hourly billing rate was regulated by the

labor unions, and sometimes the best performing, high prestige, tenured designers marked down

more billable hours than they actually spent working in the client project.

The subjective performance ratings capture the instrumental and social benefits of

networking, and are most often used in network research. The subjective performance ratings are

usually based on superior ratings and opinions of colleagues (eg. Sparrowe, Liden, Wayne and

Kraimer, 2001; Cross and Cummings, 2004).

The architects’ office does not have any formal subjective performance measurement

practices, such as 360 degree evaluations. This is because the market environment where the

company functions is not highly competitive, there is a good supply of highly educated architects

13
available in the job market, and the architectural design work is regulated by labor unions and

employee unions who negotiate the salaries in the field. Also the turnover of employees is

generally low, the average tenure in the company in 2006 being 6.5 years. 17 employees had

worked in the company for more than 10 years.

The questionnaire used as the basis of a subjective innovativeness performance measure

in this paper mapped the most influential employees who were both sources and promoters of

ideas. According to theory, one definition of creativity is the production of ideas, solutions, or

products that are both novel and useful (Feist, 1998). The innovativeness questionnaire was

designed to find employees who are both capable of producing ideas and promoting them in the

organization according to colleagues working in the same office. There were no significant

differences between the scores of idea source and idea promoting in the organization, and if a

person was considered as a source of ideas, he/she was also considered as a promoter of ideas.

The performance indicators were all constructed from the timesheet and innovativeness

questionnaire data one year after the network survey. From the timesheets and from the

innovativeness questionnaire, complete performance data was constructed for 49 employees, out

of which 20 were working in routine tasks and 29 were working in non-routine tasks.

Performance measures were not constructed for the employees who had not worked every month

during 2007, employees with an administrative role, and employees with a tenure less than one

year at the time of the survey.

Billable hours and productivity

We constructed the dependent variable of billable hours from the 2007 timesheets by counting

and standardizing everyone’s hours billed from the client.

14
 We constructed the dependent variable of productivity by dividing the count of billable

projects with billable hours during 2007. Productivity is the ratio of client projects to the total

billable hours in the year. We took the natural logarithm of productivity values to adjust for its

skew.

Innovativeness

We constructed the innovativeness measure in the innovativeness questionnaire by adding the

scores of how many times an employee was named as a promoter of ideas and as a source of

ideas. We also standardized this value.

Independent variables

Our explanatory variables of interest were two centrality measures; centrality in a routine and

non-routing communication network. Centrality measures the degree to which an employee was

connected to many other employees in the networks. We calculated Freeman’s closeness

centrality measure for this variable (ie. Wasserman and Faust, 1994). The variable was

constructed for each employee in both routine and non-routine communication networks.

Closeness centrality describes how close an employee is to others in the network (for formula,

see Wasserman and Faust, 1994). A higher value of closeness centrality means that the employee

is in a closure position in a tightly knit communication network.

We constructed centrality measures from the communication network data that included

all the 93 employees in the office during the 2006 survey. The survey-based network data forms

a directed and valued graph of communication relationships in routine and non-routine tasks. The

15
Freeman closeness centrality measure can only be constructed from binary network data.

Therefore, the network data was dichotomized to describe the overall communication activity of

both routine and non-routine tasks, to include all means of communication, in every indicated

communication frequency of the questionnaire.

In order to increase the validity of the Freeman closeness centrality measure, the two-way

giving and getting knowledge-responses were combined from the surveys. The “getting”

responses were used to ensure the existence of the relationship, where as the “giving” responses

were used to determine the frequency of the relationship. This means that the relationship was

dropped if the employee had indicated giving knowledge to somebody but the other person had

not indicated getting knowledge from this employee. Also, if an employee had indicated getting

knowledge from someone, but that someone had not indicated giving knowledge to this

employee, the relationship was dropped. In the case of non-respondents, the relationship was

considered as existing if others indicated giving knowledge to the non-respondent.

Measure to indicate employee closure positions

Freeman closeness centrality measures for each employee in both networks were constructed

with UCINET VI (Borgatti, Everett and Freeman, 2002). Freeman closeness centrality

emphasizes the distance of an employee to all others in the network by focusing on the distance

from each employee to all others (Hannemann and Riddle, 2005). The farness of an employee is

the sum of all lengths of the geodesics to every other employee, and the reciprocal farness is the

closeness centrality measure (Borgatti, Everett and Freeman, 2002). All possible frequencies of

communication were included in the network to construct the closeness centrality measure. The

closeness centrality was constructed from dichotomized data and normalized by default.

16
Control variables

We controlled other variables that may affect employee performance. The control variables used

are tenure, gender, education level, language skills and formal role.

Demographic factors of the employees have been found to influence the network

relationships and performance of the employees in organization. Ahuja et al. (2003) highlight the

effect of the formal role as a mediator between the network relationships and performance,

Reagans and McEvily (2003) show the importance of education, and Reagans and Zuckerman

(2001) argue that organizational tenure is an important variable. The control variables were

obtained from the personnel records of the company at the end of 2007.

In this study, the formal role of the employees has been identified as the most important

variable affecting how network positions will predict the performance of employees. The formal

role has been controlled in the research setting and in testing the hypotheses by separating the

employees in routine roles (professionals in the company) from the employees in non-routine

roles (managers in the company).

In terms of education level the company was quite homogeneus – most of the employees

held a Master’s degree in architecture. There were, however, a group of employees with

Bachelor level degrees among the professionals, which was taken into account in the analysis.

The education level was controlled by dividing the employees in four classes based on their

degree (1= vocational school, 2) Bachelor degree, 3) Master’s degree, and 4) PhD).

Language skills may also have an effect on performance, since the company works in

design projects in a market environment where multiple languages are spoken. Usually,

according to the personnel records, the employees can work in two languages, but there were a

17
few employees who could work in up to six different languages. The company executives valued

highly the language skills of their professionals and kept record of the languages available in the

company.

18
 Network description

Before discussing the results of our analysis, we present descriptions of the network to support

our assumption that the position in a network depends on the network type. This paper aims to

show that the centrality measures are inherently different in routine and non-routine networks.

The networks were compared at various levels of communication frequency. The counts of

relationships, densities, weighted overall clustering coefficients, and Freeman degree

centralization measures were constructed for each communication frequency separately in

routines and non-routines. The densities were constructed by dividing the number of relations in

each category of communication frequency with the total number of possible relationships. The

Freeman degree centralization as well as the weighted overall clustering coefficient measures

were constructed with UCINET VI (Borgatti, Everett and Freeman, 2002).

The density, centralization and clustering coefficients show the connectedness of the

communication networks as a whole. Density is the mean number of ties per actor in the

network, and thus it describes the overall level of interaction among the employees.

Centralization shows whether communication is concentrated in a small number of employees

rather than distributed equally, by showing the variance in the number of network ties. The

clustering coefficient shows the aggregate measure of how well the actor’s connections are

connected back to the actor, thus indicating the overall level of clustering in the network. The

weighted overall graph clustering coefficient was used, because it is the average of the densities

of the neighborhoods of all of the actors weighted with the neighborhood size, and is thus

suitable for comparing networks with different densities (see Cross and Cummings, 2004;

Hannemann and Riddle, 2005).

19
 Descriptive statistics of the two networks reveal that the inherent differences between

routine and non-routine networks are relevant. As suggested, the network of routine tasks shows

features of high density and high centralization, whereas the non-routine network is low in

density and low in centralization. Centralization shows the variance in network ties per actor, and

when the variance is low, an actor does not enjoy substantially more ties than any other actors.

The routine network is dense and highly cohesive, and the cohesion is organized around

particular focal points, whereas the non-routine network is sparse, un-cohesive and decentralized.

Also the weighted overall graph clustering coefficient measures of the networks show that the

routine network has more clustering compared to the non-routine network.

Generally, communication between employees in non-routine tasks happens less often

than in routine task communication. Metric multidimensional scaling (MDS) of the routine and

non-routine networks show and confirm the observation made of the clustering coefficients that

the non-routine communication is generally more spread across the organization. In routines,

there are more clusters in the organization compared to non-routines, which means that the

employees communicate in their non-routine tasks with more distant others compared to routine

tasks.

According to the data, not everyone in the organization indicated communication with

everyone else within the last one year prior to the survey. In the first question of the survey, the

respondents were asked to indicate who they had been communicating with during the past year.

There were 5549 links of 8556 possible links (density of 0.648), and the respondents named on

average 65 others. When examining how many times the person was recalled to be

communicated with by others (in-degree of the first question), it was noted that the employees

with a shorter than a two-year tenure had an average in-degree of 48.

20
 The average in-degree of the employees tenured between two and ten years, and also

employees with over ten years of tenure were exactly the same, 66 others indicated

communication with them during the past year. This tells roughly that it takes about two years to

work in the company to reach some kind of average, established level of communication with

others in the same office, but after reaching the certain normal level, the number of others

communicated with does not increase along with time.

Table 1 shows the number of ties in the routine and non-routine networks in different

frequencies of communication. The number of links and the density of the routine and non-

routine networks decrease when the frequency of communication increases. The non-routine

network is substantially less dense compared to the routine network – there are only 72 daily

relations in communication related to non-routine tasks in the company. Besides the notable

difference in density, the communication network in non-routine tasks is also less centralized and

less clustered compared to the routine tasks, measured with Freeman degree centralization and

weighted overall clustering coefficient (Table 1).

------------------------------------------

Insert Table 1 about here

-------------------------------------------

Figure 1 presents a two-dimensional metric multidimensional (MDS) scaling of the routine and

non-routine networks in the communication frequency of weekly or more often. MDS presents

the similarities and dissimilarities of the relationships of the actors compared to each other. The

closer the actors are in the MDS scale, the more similar structure of ties they have (Scott, 2000).

21
It can be seen in the figure that the routine network is more equally distributed, which suggests

that there is more closure among the actors. In the non-routine network there are more

dissimilarities on the network ties among the employees, which illustrates that some individuals

are more connected than others.

------------------------------------------

Insert Figure 1 about here

-------------------------------------------

22
 Results

We used an ordinary least squares (OLS) regression model to test our hypotheses. Table 2

presents descriptive statistics and correlations of the variables. Correlations between dependent

and independent variables that are significant are indicated. Regression results are shown in

Table 3. We present three sets of models for the three performance outcomes. The first model of

each of the three sets shows the effect with the control variables. Models 2 and 3 of each set

show the main effects of centrality in routine and non-routine network types. Finally, Models 4

and 5 show the interactions with the routine and non-routine roles.

------------------------------------------

Insert Table 2 about here

-------------------------------------------

The models explain notable degrees of variance in the dependent variables (Table 3). The results

indicate that the effect of centrality on performance increases for an individual in a non-routine

role greater than in a routine role.

------------------------------------------

Insert Table 3 about here

-------------------------------------------

Our analysis shows mixed results (Table 3). We found support for Hypotheses 1a and 1b only for

two of the three performance measures. An increase in central position in non-routine and

23
routine communication networks increases performance in productivity and innovativeness

(p<0.1 in productivity and billable hours and p<0.05 in innovativeness). There is a non-

significant negative effect on the billable hours performance. We found support for Hypothesis 2

only for productivity and innovative performance measures. Increased centrality in non-routine

tasks increases the performance of employees in non-routine roles more than centrality in a

routine network. For productivity, the coefficient for non-routine*non-routine role is 0.0641

(p<0.05) and the coefficient for routine*non-routine is 0.0441 and non-significant. For

innovativeness, the coefficient for non-routine*non-routine role is 0.105 (p<0.01), and the

coefficient for routine*non-routine is 0.0565, with marginal significance. We were surprised to

find that the interaction of the centrality measures with the roles was negative. This indicates that

centrality in both non-routine and routine networks and having a non-routine role reduces

performance when measured by billable hours. The results do not support Hypothesis 2. The

match between non-routine centrality and non-routine role decreases performance more than the

mismatch between routine centrality and the non-routine role.

24
 Discussion

According to the results of this paper, the routine or non-routine nature of the task affects the

network structure. Furthermore, centrality affects employee performance, depending on the

network context, as well as the employee’s role. The finding is in line with the existing theory on

external contingencies, such as market uncertainties, authority power, cultural traditions or

institutions having an effect on the network structures (Lincoln, Hanada and McBride, 1986;

Podolny, 1994; Podolny and Baron, 1997). Ahuja (2000) concludes that the structural

characteristics of networks depend on the objectives of the network members. The contingency

presented in this paper is in line with previous network research at the employee and

organizational level.

The results bring new light to the debate between dense (Coleman, 1988) and sparse

(Burt, 1992) network structures. The debate has been exaggerated in the network theory, and the

task-contingent view shows that they are both right in their own respect. Burt has repeatedly

stated that the employees occupying structural holes in a highly sparse network perform better in

innovative tasks (Burt, 1992; Burt, 2004). This is contrary to Coleman’s (1988) arguments where

reciprocity and trust are created in dense, centralized networks that are good for solving

unambiguous problems quickly.

It is logical that the employees whose formal role deals with non-routine tasks, benefit

from a central position in a sparse and decentralized non-routine communication network as an

increase in their performance. Furthermore, following this logic, employees with a routine role

should benefit from a central position in routine tasks networks. In this study, we found support

to the first argument – employees in non-routine role benefit more from centrality in a non-

routine network compared to a routine network. However, we found only partial support to the

25
latter contingency argument. The congruence can be explained as in the following two-by-two

table where “ns” stands for non-significant effect:

------------------------------------------

Insert Table 4 about here

-------------------------------------------

The research result also strengthens the traditional assumptions about the strata of professional

service firms: the manager and professional roles are separated. Those employees whose role is

to perform routine type, billable work to clients under the leadership of managers, communicate

more in routine tasks and less in non-routine tasks, and do not seem to benefit from the central

position in the communication network. Those employees whose tasks are related to non-routine

work communicate more about non-routine work, and also benefit from the central position in

that network in the form of increased productivity and increased status in the organization.

Increased centrality in non-routine tasks seems to lower the performance of those in non-

routine roles when measured with billable hours. The increase in centrality leading towards

decline in performance is supported by Krackhadt’s (1994) research, in which the density of

communication relationships was negatively related to performance after a certain threshold,

which suggests that there is a curvilinear relation between the degree of interaction and

performance. In Krackhardt’s studies excess communication was a result of increased demands

of coordination of work due to problems in the workplace that prevented the actors from

performing their actual tasks.

26
Limitations

The research has some limitations related to the subjectivity of network analysis research, the

problem of endogeneity in the research design and low number of observations in the OLS

regression analysis. Furthermore, the operationalization of the constructs of routine and non-

routine poses a challenge. Despite that theoretically the difference between the two seem clear,

their meaning differs from company to another. Intuitively, the definition of a routine task means

factory work in the 1920’s, but according to the Nelson and Winter’s (1982) definition, routines

are simply standardized procedures and rules, no matter the industry. Therefore, it can be argued

that even in a highly creative type of work such as architecture, there are certain routines in the

firm that are performed by employees in routine role. In this study, professionals were defined as

performing routine work of the company because they produced the drawings, the output that the

company sold in the market. The managers were defined as performing non-routine work, since

their work involved managing the projects and the professionals, sales and marketing and

business development.

When relationships are studied on the basis of subjective evaluations of employees, the

structure of the network is likely to vary according to the type of relationships. It can be

presumed that if the network questionnaire were designed to measure such relationships as

advice, friendship or trust, the resulting Freeman closeness centrality measures used as the

independent variable could have been different.

It can be argued, however, that communication relationships are the most useful when

investigating task-related network structures in an organization. Such relationships as friendship

or advice, important as such, are more related to social, not task-related ties between employees.

It can be expected that task-related communication ties reveal underlying social relationships –

27
in expert work, employees communicate with those they like the most, and choose not to

communicate with some others. Measuring task-related communication is, however, easier than

measuring social relationships, because employees answer a task-related questionnaire more

willingly than questions about their friendship ties. It can be presumed that especially non-

routine task-related communication is more likely to happen among people who already know

each other at a personal level. This would mean that other types of relationships create

possibilities for work-related communication in the organization - personal relationships create

opportunities for cooperation (cf. Ouchi, 1980; Larson, 1992; Eisenhardt and Schoonhoven,

1996).

The structure of communication relationships reveals the flow of knowledge in the

organization, because communication can be thought to be an opportunity to share knowledge

with other employees. Transfer of knowledge is discretionary, and it follows the path of least

resistance, therefore the more communication ties there are, the more likely the employees are to

share their knowledge (Reagans and McEvily, 2003).

In the network survey, the respondents were given two months to answer the network

questionnaire. This was considered suitable because the frequency scales in the survey were

designed to map perceived, well established communication relationships between the

employees. The long response time made possible a high response rate, but due to the broad

frequency scale, employees with a low tenure were not able to provide adequate answers to the

survey. The two-way, directed questionnaire design improved the validity of the survey results

(Cross and Cummings, 2004), because the relationships were counted as relationships only if

both parties indicated the existence of the tie.

28
 The testing of the hypotheses included both objective and subjective performance

measures. The subjective performance measures, such as the perceived innovativeness used in

this study, cause a problem of endogeneity in network research. This is because the employees

who are in a network position to control the flow of information to and from other employees

(employees with high centrality) may be able to use their position to influence the performance

evaluations (Brass, 1995). Central employees will also be more visible for others in the

organization, and they will naturally be perceived to have a higher status than less visible

employees. However, communication networks are a social phenomenon, and as the survey-

based network presents subjective evaluations of the frequencies of communication among

employees, also the performance variable can be constructed similarly. Then, the subjective

performance indicator will actually reveal how well an employee is able to utilize his/her

network position to gain a social status in the organization.

The problem of endogeneity was controlled in this study by constructing the dependent

variables on the basis of data gathered one year after the network survey. However, as several

professionals had left the company between the network survey and constructing the

performance measures, the number of observations remained fairly low in the testing of the

hypotheses (routine role = 20, non-routine role = 29). In the analysis, the effect of centrality on

performance in different roles was tested with interaction term that allowed the inclusion of full

sample of N=49 in the models.

Implications for practice and future research

Despite of the limitations, this paper advances the network theory and management of

professional service firms. First, this study shows links between the characteristics of task-related

29
communication networks and employees’ performance. Communication in non-routine tasks is

highly important in the contemporary knowledge economy, which puts increasing weight on

intangible assets and innovation. Future applications of the research approach introduced in this

paper could include the structures of R&D teams in the early phases of the innovation processes.

According to the network theory, the communication structure among employees makes a

difference, and by influencing the structure with right leadership, the time to market of new

innovations could be improved. It can be proposed that in different phases of the innovation

process, some phases benefit from dense, interlocked communication networks, while others

benefit from sparse and decentralized network structures. Future research could include

longitudinal observation-based research on teams performing inherently routine or non-routine

types of projects and comparisons of them. This way, it would be possible to determine the

structural characteristics of communication networks that influence team performance.

Secondly, besides structural characteristics, future research should include network

studies on employee relationships in knowledge intensive organizations. As stated in this paper

based on existing theory, but left unclear in terms of the research results, highly performing

employees are embedded in the network positions of high clustering and reach. The future

research could concentrate on the demographics of those employees who are both in favorable

network positions and also perform well. What are their personalities and demographics like?

How do they perform their routine and non-routine tasks? This would have an impact on the

management of professional service firms, since the success of those firms relies heavily on

employee competence, expertise and virtuoso talent. A successful professional service firm

should aim to identify, encourage and retain highly performing employees in favorable network

positions.

30
 Finally, this research is important for managers in professional service firms in terms of

performance measurement. As shown in the results of the research, the objective performance

measures do not always indicate the managerial performance in the organization. Excess

embeddedness in the communication network may decrease performance in the billable client

work, even if the employees’ productivity is high and is considered as important in

innovativeness. Highly central employees in the communication network are likely to be the ones

who coordinate others’ work in multiple projects, disseminate knowledge and solve conflicts,

which results in lower client performance. This means that those professional service firms that

use only billable client work as a measure of managerial performance, should be aware that it

may not tell the whole truth about the manager, and consider also other types of performance

measures.

31
 References:

Ahuja, G., 2000. Collaboration Networks, Structural Holes, and Innovation: A Longitudinal
Study. Administrative Science Quarterly, 45, 425-455.

Ahuja, M., Galletta, D., Carley, K. M., 2003. Individual Centrality and Performance in Virtual
R&D Groups: An Empirical Study. Management Science, 49, 1 21-38.

Bavelas, A., 1951. Communication patterns in task-oriented groups. In: D. Lerner and H. D.
Lasswell (Eds.), The Policy Sciences. Stanford University Press, Stanford, CA.

Becker, M., 2004. Organizational routines: a review of the literature. Industrial & Corporate
Change, 13, 4 643-677.

Becker, M., 2005. A framework for applying organizational routines in empirical research:
linking antecedents, characteristics and performance outcomes of recurrent interaction
patterns. Industrial & Corporate Change, 14, 5 817-846.

Blau, P. M., Scott, W. R., 1962. Formal Organizations: The Concept. In: J. M. Shafritz and J. S.
Ott (Eds.), Classics of Organization Theory. Wadsworth Pub. Co., Belmont, 19: 214.

Borgatti, S. P., Everett, M. G., Freeman, L. C. (2002), Ucinet 6 for Windows: Software for
Social Network Analysis, Analytic Technologies, Harvard.

Brass, D. J., 1995. A Social Network Perspective on Human Resources Management. Research
in Personnel and Human Resources Management, 13, 1 39-79.

Burt, R. (1992), Structural holes: The social structure of competition., Harvard University Press,
Cambridge.

Burt, R., 2004. Structural Holes and Good Ideas. American Journal of Sociology, 110, 2 349-
399.

Carley, K. M., 1991. A theory of group stability. American Sociological Review, 56, 331-354.

Cohendet, P., Llerena, P., 2003. Routines and incentives: the role of communities in the firm.
Industrial & Corporate Change, 12, 2 271-297.

Coleman, J., 1988. Social Capital in the Creation of Human Capital. The American Journal of
Sociology, 94, 1 95-120.

Coleman, J. S. (1990), Foundations of Social Theory, Harvard University Press, Cambridge,

MA.

32
Cross, R., Cummings, J. N., 2004. Tie and Network Correlates of Individual Performance in
Knowledge-Intensive Work. Academy of Management Journal, 47, 6 928-937.

Cyert, R. M., March, J. G. (1963 [1992]), A Behavioral Theory of the Firm, Blackwell, Malden,
MA.

Eisenhardt, K. M., Schoonhoven, C. B., 1996. Resource-Based View of Strategic Alliance

Formation: Strategic and Social Effects in Entrepreneurial Firms. Organization Science,
7, 2 136-150.

Feist, G., 1998. A Meta-Analysis of Personality in Scientific and Artistic Creativity. Personality
and Social Psychology Review, 2, 4 290-309.

Granovetter, M., 1973. The Strength of Weak Ties. American Journal of Sociology, 78, 6 1360-
1380.

Hannemann, R. A., Riddle, M. (2005), Introduction to Social Network Methods, University of

California, Riverside, Riverside, CA.

Hansen, M., Nohria, N., Tierney, T., 1999. What's Your Strategy for Managing Knowledge.
Harvard Business Review, 77, 2 106-117.

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.

Hansen, M. T., 2002. Knowledge Networks: Explaining Effective Knowledge Sharing in

Multiunit Companies. Organization Science, 13, 3 232.

Huselid, M., 1995. The Impact of Human Resource Management Practices on Turnover,
Productivity, and Corporate Financial Performance. The Academy of Management
Journal, 38, 3 635-672.

Huselid, M., Jackson, S. E., Schuler, R. S., 1997. Technical and Strategic Human Resource
Management Effectiveness as Determinants of Firm Performance. The Academy of
Management Journal, 40, 1 171-188.

Ibarra, H., 1993. Personal Networks of Women and Minorities in Management: A Conceptual
Framework. Academy of Management Review, 18, 1 56-87.

Krackhardt, D., 1994. Constraints on the Interactive Organization as an Ideal Type. In: C.
Heckscher and A. Donnellon (Eds.), The Post-Bureaucratic Organization: New
Perspectives on Organizational Change. Sage Publications, Thousand Oaks, CA: 211-
222.

Larson, A., 1992. Network Dyads in Entrepreneurial Settings: A Study of the Governance of
Exchange Relationships. Administrative Science Quarterley, 37, 1 76-104.

33
Lillrank, P., 2003. The Quality of Standard, Routine and Nonroutine Processes. Organization
Studies, 24, 215

Lincoln, J. R., Hanada, M., McBride, K., 1986. Organizational Structures in Japanese and U.S.
Manufacturing. Administrative Science Quarterly, 31, 3 338-364.

March, J. G., Simon, H. A., Guetzkow, H. S. (1958 [1993]), Organizations, Blackwell,

Cambridge, Mass., USA.

Nahapiet, J., Ghoshal, S., 1998. Social Capital, Intellectual Capital, and the Organizational
Advantage. Academy of Management Review, 23, 2 242-266.

Nelson, R. R., Winter, S. G. (1982), An Evolutionary Theory of Economic Change, Belknap,

Cambridge, MA.

Ouchi, W. G., 1980. Markets, bureaucracies and clans. Administrative Science Quarterly, 23,
293-317.

Pava, C., 1983. Designing managerial and professional work for high performance: A
sociotechnical approach. National Productivity Review, 126-135.

Podolny, J. M., 1994. Market uncertainty and the social character of economic exchange.
Administrative Science Quarterly, 39, 3 458.

Podolny, J. M., Baron, J. N., 1997. Resources and Relationships: Social Networks and Mobility
in the Workplace. American Sociological Review, 62, 5 673-693.

Reagans, R., McEvily, B., 2003. Network Structure and Knowledge Transfer: The Effects of
Cohesion and Range. Administrative Science Quarterly, 48, 2 240-267.

Reagans, R., Zuckerman, E. W., 2001. Networks, Diversity, and Productivity: The Social Capital
of Corporate R&D Teams. Organization Science, 12, 4 502.

Salter, A., Gann, D., 2003. Sources of Ideas in engineering design. Research Policy, 32, 8 1309-
1324.

Scott, J. (2000), Social Network Analysis. A Handbook, Second Edition, SAGE Publications,
London.

Scott, W. R. (2003), Organizations: Rational, Natural and Open Systems, Prentice Hall, Upper
Saddle River, NJ.

Shaw, M., 1964. Communication networks. In: L. Berkowitz (Eds.), Advances in Experimental
Social Psychology. Academic Press, New York: 111-147.

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, 2 316.

34
Uzzi, B., 1996. The Sources and Consequences of Embeddedness for the Economic Performance
of Organizations: The Network Effect. American Sociological Review, 61, 4 674-698.

Wasserman, S., Faust, K. (1994), Social Network Analysis: Methods and Applications,
Cambridge University Press.

35
 APPENDIX

Translated outline of the network questionnaire and the innovativeness survey

2006 network survey

Background information:
Name:
Year of birth:
Tenure in company x:
Department:
Title:
Describe your current job with a couple of sentences:

Name those that you have been communicating during the past year (from the roster of names)

With whom of these people have you been communicating with during the last year? Communication includes all means of communication (face-
to-face, phone, email…)

Below is a list of all people working in Company X. Depending on your tenure and work description, you may know some people well and
somebody hardly at all. From the list below, pick the people that you have been communicating with over the last year.
You may pick as many people as you like.

Routine task questions:

How often do you give information related to routine work to the following people?

Below is a list of all those people you communicate with based on the first question. If you notice that you forgot to mention somebody, you can
add him/her before answering this question.

With this question, we map those persons that you exchange information with related to routine –like work tasks that you encounter. Routine
work is:
- something that has been defined in advance
- has a repetitive nature
- must be done in a given timeframe
- routine work can be related to internal affairs, such as to timesheets or reports
- routine work can be something done for the customer that you feel is routine of your own expertise
- information related to routine work can be for example standardized information, instructions, documents or schedules. You can give
or receive this information with all communication means, i.e. face-to-face, email or by phone.

There are no correct answers to this question and the answers vary according to the work description. It is perfectly normal if you don’t recall
anybody when you think about routines. It is also normal to recall several people.

From the list below, pick those people that you give information related to routines to.

If you give information related to routines only to one person, pick him/her. If you give information related to routines to many persons, pick
them. If you don’t give information related to routines to anybody, don’t pick anyone. All of your entries are naturally just estimates.

How often do you get information related to routine work from the following people?

From the list below, pick those people that you receive information related to routines from.

These people might be the same that you give information related to routines to, but they can also be other people.

If you receive information related to routines only from one person, pick him/her. If you receive information related to routines from many
people, pick them. If you don’t receive information related to routines from anybody, don’t pick anybody. All your entries are naturally just
estimates.

Non-routine task questions:

How often do you give information related to ideas to the following people?

Below is a list of all those people you have been communicating with during the past year according to the first question. If you notice that you
forgot to mention somebody, you can add him/her before answering this question.

These questions map those people that you exchange ideas with in your work. With ideas we mean:
- the feeling when you get those “light bulb moments” in your work

36
 - an idea is something new and you are not aware that anybody had thought about it before
- everybody has their own ways and places to come up with ideas. You may get ideas whenever and wherever at work, at home or in
your freetime
- ideas can be born or transmitted in informal occasions, such as during lunch or coffee breaks

There are no correct answers to this question and the answers vary according to the work description. It is perfectly normal if you don’t recall
anybody when you think about ideas. It is also normal to recall several people.

From the list below, pick those people that you present your ideas to.

If you present your ideas only to one person, pick him/her. If you present your ideas to many people, pick them. If you don’t present your ideas to
anybody, don’t pick anybody. All your entries are naturally just estimates.

How often do you get information related to ideas from following people?

From the list below, pick those people that you get ideas from (who present their ideas to you). These people may be the same individuals that
you tell your ideas to, or they can be other people.

If only one person presents his/her ideas to you, pick him/her. If many people present their ideas to you, pick them. If nobody presents their ideas
to you, don’t pick anybody. All your entries are naturally just estimates.

2007 Innovativeness survey

Ideas: important individuals that promote ideas

Question – Who are the most important individuals in your organization in terms of promoting ideas?

Please write in the gaps below the names of five individuals in your office that you feel are the most important in terms of promoting your own
ideas. With this question we are looking for individuals whose time (or action) you would most probably use if/when you would like to get your
own idea forward in the organization. Without the help of these individuals ideas would be considerably more difficult to implement.

Person 1:
Person 2:
Person 3:
Person 4:
Person 5:

Ideas: important individuals that create ideas

Question – Who are the most important individuals in your organization in terms of creating ideas?

Please write in the gaps below the names of five individuals in your office that you feel are the most important in terms of creating new ideas.
This question we are looking for individuals who are known to be rich in ideas. Without these individuals, there would be considerably less new
ideas created.

Person 1:
Person 2:
Person 3:
Person 4:
Person 5:

37
 TABLE 1

Numbers of ties in routine and non-routine communication networks (max = 93*92 = 8556)

< Less than

Daily or more < Weekly < Once per month
once per month

Routines

Number of relations 347 1080 1890 2817

Density 0,041 0,126 0,221 0,329

Weighted overall graph clustering coefficient 0.255 0.266 0.35 0.454

Freeman degree centralization 14.80 % 78.55 % 67.43 % 53.58%

Non-Routines

Number of relations 72 275 585 1168

Density 0,008 0,032 0,068 0,137

Weighted overall graph clustering coefficient 0.053 0.162 0.22 0.244

Freeman degree centralization 11.75 % 13.41 % 18.86 % 79.14%

38
 TABLE 2

Means, Standard Deviations and Correlations

VARIABLE Mean s.d. 1 2 3 4 5 6 7 8 9 10 11

1. Tenure 9.04 5.452542

2. Education level 2.469388 0.8441475 0.1897

3. Gender 0.387755 0.4922875 -0.195 -0.2967*

4. Language skills 3.102041 1.084837 -0.0292 0.1059 -0.1537

5. Role task 0.591837 0.496587 0.1442 0.6654*** -0.0209 0.2723

6. Routine closeness centrality 67.68545 9.444632 0.142 0.0422 -0.2045 0.137 0.2709

7. Non-routine closeness centrality 57.70269 8.653174 0.0838 0.073 -0.1998 0.1093 0.122 0.7385***

8. Routine closeness cent. X Role task 41.3032 35.42254 0.1758 0.6284*** -0.068 0.2931* 0.9784*** 0.4256** 0.2185

9. Non-routine closeness cent. X Role task 34.66408 29.7714 0.1602 0.6547*** -0.0545 0.2953* 0.977*** 0.3727** 0.2759 0.9846***

10. Productivity -4.70243 1.01691 -0.2415 0.0564 0.0831 0.1956 0.3302* 0.3075* 0.2956* 0.3781** 0.3981**

11. Innovativeness -1.60E-09 1 0.0234 0.3422* -0.277 0.2081 0.4115** 0.4283** 0.5294*** 0.4876*** 0.5416*** 0.6145***

12. Billable hours -2.14E-10 1 0.2037 -0.0986 0.0905 -0.2143 -0.2162 -0.2839* -0.2924* -0.2788 -0.2992* -0.7976*** -0.5691***

Sample size = 49
* p<0.05
** p<0.01
*** p<0.001

39
 TABLE 3

Results of OLS Regression Analysis Predicting Productivity, Innovativeness and Billable

Hours
Productivity Innovativeness Billable hours
VARIABLE Model 1 Model 2 Model 3 Model 4 Model 5 Model 1 Model 2 Model 3 Model 4 Model 5 Model 1 Model 2 Model 3 Model 4 Model 5

Tenure -0.0505* -0.0545** -0.0526** -0.0596** -0.0562** -0.0161 -0.0213 -0.0194 -0.0278 -0.0253 0.0460* 0.0499* 0.0478* 0.0567** 0.0522**

Education level -0.300 -0.173 -0.257 -0.165 -0.291 0.00532 0.170 0.0743 0.181 0.0180 0.114 -0.0104 0.0742 -0.0221 0.116

Gender -0.0456 0.112 0.0815 0.143 0.0560 -0.561* -0.356 -0.357 -0.316 -0.399* 0.292 0.138 0.175 0.0959 0.206

Language skills 0.0659 0.0677 0.0556 0.0436 0.0175 0.0493 0.0516 0.0326 0.0207 -0.0297 -0.109 -0.111 -0.0997 -0.0786 -0.0538

Role 1.055** 0.780* 0.949** -2.141 -2.665 0.807** 0.449 0.637* -3.293* -5.278*** -0.566 -0.296 -0.468 3.627* 3.888**

Routine centrality 0.0272* -0.000431 0.0354** -1.70e-05 -0.0267 0.0105

Non-routine centrality 0.0329** -0.00203 0.0527*** -0.00440 -0.0302* 0.0119

Routine centrality X Role 0.0441 0.0565* -0.0592*

Non-routine centrality X Role 0.0641** 0.105*** -0.0773**

R-squared 0.233 0.285 0.307 0.318 0.377 0.252 0.342 0.447 0.399 0.643 0.141 0.192 0.205 0.254 0.311
F 2.62 2.79 3.10 2.74 3.55 2.90 3.64 5.65 3.89 10.53 1.41 1.66 1.80 2.00 2.64

Tenure is in years, Gender is coded 1 for "female", Role is coded 1 for "manager"
* p<0.1
** p<0.05
*** p<0.01

40
 TABLE 4

Impact of high centrality on employee performance

Non-routine role Routine role

Centrality in non-routine + ns
network
Centrality in routine ns ns +
network

41
 FIGURE 1

Illustrations of MDS scaling of routines (left) and non-routines (right), in a frequency of

weekly or more often

2 2

1 1

0 0

1 1

-2 -2
-2 -1 0 1 2 -2 -1 0 1 2

42

View publication stats