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

net/publication/220079742

Informational Influence in Organizations: An Integrated Approach to Knowledge

Adoption

Article  in  Information Systems Research · March 2003

DOI: 10.1287/isre.14.1.47.14767 · Source: DBLP

CITATIONS READS
1,122 11,189

2 authors, including:

Stephanie Watts Sussman

Boston University
29 PUBLICATIONS   3,088 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Stephanie Watts Sussman on 14 January 2014.

The user has requested enhancement of the downloaded file.

 INFORMATIONAL INFLUENCE IN ORGANIZATIONS:

AN INTEGRATED APPROACH TO KNOWLEDGE ADOPTION

By
Stephanie Watts Sussman
Boston University School of Management
595 Commonwealth Avenue
Boston, MA 02106
[email protected]
Phone: (617) 358-2330
Fax: (617) 353-5003

And

Wendy Schneier Siegal

Boston University School of Management
595 Commonwealth Avenue
Boston, MA 02106
[email protected]
Phone: (617) 353-6145
Fax: (617) 353-4098

Forthcoming in Information Systems Research

The authors wish to thank Lee Sproull and John C. Henderson for their many contributions to the

development and execution of this work as members of the first author’s dissertation committee.

This material is based upon work supported by the National Science Foundation under grant

number SBR-9433384 and the Boston University Systems Research Center. Any opinions,

findings, conclusions or recommendations expressed in this material are those of the authors and do

not necessarily reflect the views of the National Science Foundation.

 INFORMATIONAL INFLUENCE IN ORGANIZATIONS:

AN INTEGRATED APPROACH TO KNOWLEDGE ADOPTION

ABSTRACT

This research investigates how knowledge workers are influenced to adopt the advice that

they receive in mediated contexts. The research integrates the Technology Acceptance Model

(Davis, 1989) with dual process models of informational influence (e.g., Petty and Cacioppo, 1986;

Chaiken and Eagly, 1976) to build a theoretical model of information adoption. This model

highlights the assessment of information usefulness as a mediator of the information adoption

process. Importantly, the model draws on the dual process models to make predictions about the

antecedents of informational usefulness under different processing conditions.

The model is investigated first qualitatively using interviews of a sample of 40 consultants,

and then quantitatively on another sample of 63 consultants from the same international consulting

organization. Data reflect participants’ perceptions of actual e-mails they received from colleagues

consisting of advice or recommendations. Results support the model, suggesting that the process

models used to understand information adoption can be generalized to the field of knowledge

management, and that usefulness serves a mediating role between influence processes and

information adoption. Organizational knowledge work is becoming increasingly global. This

research offers a model for understanding knowledge transfer using computer-mediated

communication.

1
 INTRODUCTION

A consultant is designing a new distribution process for a client in an unfamiliar industry.

She asks several people in her formal and informal networks for some advice concerning a

particularly complex part of the process. She also posts the question on the organization-wide

electronic bulletin board system. In response to these inquiries she receives half a dozen different

e-mails, from people with varying degrees of expertise in the area and industry, some of whom she

knows better than others. How does she sort through the advice in these e-mails to decide which

action to take?
This scenario exemplifies one problem of knowledge transfer in organizations, specifically,

knowledge utilization or adoption. While engaged in knowledge work, workers must comprehend

the knowledge they receive, discern the complexities and subtleties of it, and incorporate it into

their schemas and mental models. Meaningful learning and insightful problem solving require that

learners reorganize and integrate new information into their cognitive structures (Ausubel, 1963).

Only then can they determine a course of action that seems to best serve their needs and the needs

of the organization.

This research empirically investigates how people in organizations integrate received

knowledge into their ongoing work. A field study was conducted which examines how individuals

are influenced to follow certain courses of action, based on actual advice, recommendations and

suggestions they received via e-mail. In this study we develop and test hypotheses based on

theories of adoption (Technology Acceptance Model; Davis, 1989) and informational influence

(e.g. Petty and Cacioppo, 1986; Eagly and Chaiken, 1993), integrating the theories to reflect field-

based knowledge transfer work. We employ computer-mediated communication (CMC) as a

context to examine how individuals incorporate received advice into their actions in order to

understand the broader phenomenon of organizational knowledge transfer. We believe the resultant

model of knowledge adoption has the potential to inform more general processes of organizational

knowledge management.

2
 THEORETICAL DEVELOPMENT

Knowledge Transfer in Organizations

In organizations, knowledge is transferred between and among individuals, groups,

communities of practice, and other organizations. Knowledge is embedded in the flows of

communication and information that move across these multiple arenas. Theorists conceive of

knowledge flows as consisting of the following five elements (Gupta and Govindarajan, 1996): 1)

perceptions as to the value of the source knowledge, 2) the willingness of the source to share that

knowledge, 3) the availability of rich information channels, 4) the willingness of the recipient to

acquire the knowledge, and 5) the absorptive capacity of the recipient(s) (Cohen and Levinthal,

1990). We can conceptualize these elements as reflecting characteristics or perceptions of the

source (# 1 and 2), the channel (# 3), and the recipient (# 4 and 5). To understand knowledge

transfer, we need to understand each of these elements and the way they work together to influence

how people learn and behave in organizations.

For knowledge transfer to occur, learning must transpire in the mind of the recipient. This

reflects the widely shared view that knowledge involves cognitive structures and processes, and

cannot be embodied as texts or other explicit representations (Nonaka, 1994; Brown and Duguid,

1991; Davenport and Prusak, 1998). Knowledge recipients' perceptions of transferred information

are situated in the particular place and time of the learner. For example, received content may be

taken to heart by one recipient and ignored by another, depending on these recipients’ experience of

the learning content, structure, and process. The same content can engender very different

responses in different recipients (Eagly and Chaiken, 1976). This makes the knowledge transfer

process very complex to research empirically, particularly in the field where there is little about

recipient processes that researchers can control or even see. And in organizations, knowledge is

3
transferred among multiple entities, at multiple levels of analysis. Knowledge can be tacit or

explicit, procedural or declarative, social, pragmatic, and equivocal. Clearly, any fruitful

investigation of knowledge transfer must limit the scope of this broad phenomenon.

In most ways learning is more effective when it occurs in a face-to-face context rather than a

mediated one (Lave and Wenger, 1991; Brown and Duguid, 1991; Zack and McKenney, 1995).

However, global knowledge work is increasingly reliant on online document transfer, underscoring

the need to investigate the knowledge transfer process at the interface between explicit information

(i.e. texts) and personal knowledge. Nonaka (1994) describes this process as the internalization

phase of knowledge transfer, in which explicit information is transformed into internalized

knowledge and meaning. This study investigates this internalization phase of organizational

knowledge transfer as it transpires from received text, calling this process knowledge adoption. We

focus on one particular manifestation of knowledge internalization – information adoption in

response to advice-bearing e-mail.

E-mail is a form of asynchronous computer-mediated communication (CMC) that has

become commoditized in knowledge-based organizations. More than 10 billion e-mails are now

exchanged daily, and this quantity is expected to increase to 35 billion by 2005 (Levitt and

Mahowald, 2001). In a study of managerial e-mail use, 27% of managers reported using e-mail for

equivocal communication such as exchanging subjective views, and 14% would use e-mail to

influence or sell an idea (Markus, 1994). This evidence suggests that knowledge sharing via e-mail

is a rapidly emerging phenomenon.

Organizational use of e-mail has been widely studied for its effects on social context cues

(Sproull and Kiesler, 1986) and its relationship to media richness (Dennis and Kinney, 1998; Lee,

1994; Rice, 1992; Markus, 1994; Fulk, 1993; Zack, 1993). In this research, we focus on one

particular type of e-mail - e-mail which recipients perceive to contain advice.

4
 The assessment of advice is beginning to be recognized as an important managerial process

(Harvey, Harries and Fischer, 2000; Jungermann, 1999; Sniezek and Buckley, 1995). Advice is

generally viewed as information that communicates an opinion about what could or should be done

about a problem or issue. In our earlier example, any opinion or recommendation that the

consultant receives in response to her query would be construed as advice. Importantly, advice is

viewed as advisory rather than mandatory, as the most that advice-givers can realistically expect is

to have a degree of influence, given the multiple, often-conflicting responses advice-seekers receive

(Harvey, Harris and Fischer, 2000). As organizations become less hierarchical, the need for non-

impositional forms of knowledge transfer such as advice giving and receiving is likely to increase.

In the following discussion, we draw upon two bodies of theory - adoption and

informational influence - to build a process model of information adoption as it occurs via advice-

based e-mails in organizational contexts. In presenting this model, our discussion is organized as

follows: First, we describe how theories of adoption are relevant to the transfer of ideas. Second,

we present theories of informational influence as a means for understanding how people are

influenced to adopt ideas. Third, we present our model which integrates these two theoretical

perspectives, a necessary step for describing adoption and influence in the particular domain of

knowledge transfer. Finally, we describe and report on an empirical study performed to explore and

validate the model.

Theories of Adoption
Adoption theories describe the processes people undergo when deciding to perform an

activity for the first time (as distinguished from ongoing usage behavior). According to the Theory
of Reasoned Action (TRA) (Fishbein and Ajzen, 1975; Ajzen and Fishbein, 1980) and its

derivative, the Technology Acceptance Model (TAM) (Davis, 1989), people form intentions to

5
adopt a behavior or technology based on their beliefs about the consequences of adoption and their

evaluation of these consequences. An information adoption-based view of knowledge transfer

assumes that, just as people form intentions toward adopting a behavior or a technology, they

similarly form intentions toward adopting particular advocated ideas and behaviors. As such,

factors that influence the adoption of behaviors or technologies can be used to understand the

adoption of advice as well.

Research using TAM has found that beliefs about the usefulness of adopting an advocated

behavior are particularly consequential in determining user adoption intentions (Davis,1989). By

applying a TRA/TAM-based model of adoption to the domain of knowledge transfer, we propose

that perceptions of the usefulness of received advice should similarly predict intentions toward

adopting that advice. This reflects accruing evidence that perceived usefulness is a fundamental

predictor of user adoption, with significant correlations to both current and future self-reported

system usage (Davis, 1989). Additional support for the centrality of message usefulness comes

from social cognition research that has found that the relevance of a stimulus is the key factor

determining whether managers ignore the stimulus or go on to interpret it during problem sensing

(Kiesler and Sproull, 1982). Information that is likely to produce useful task resolutions receives

precedence and weight in judgment and choice processes (Feldman and Lynch, 1988). In this

sense, the advice assessment and adoption process can be conceived of as a form of informational

influence, whereby individuals are influenced by information received from others to the degree

that they assess it as useful evidence about reality (Eagly & Chaiken, 1993; p. 630).

Information Adoption and the ELM

While adoption models are useful first steps in understanding how intentions toward a

message are formed, these models were not designed to answer questions about the influence

process itself. For instance, returning to our earlier example, when a consultant receives advice

from several colleagues regarding the design of a new distribution process, what aspects of a

received message are more or less consequential in influencing her to follow a piece of advice?

6
And further, how is it that different people can be influenced by the same message in different

ways? In what situations does the source of the message have a major impact? When are people

most likely to ignore the quality of the argument made? While TAM and the Theory of Reasoned

Action (TRA) are useful first steps in understanding how behavioral intentions toward adopting a

message are formed, these models were not designed to answer these types of questions. This study

suggests an alternative model for teasing out some of the important effects of this extremely

complex knowledge adoption process.

To understand the process by which individuals will be influenced by the messages that they

receive, we draw on the Elaboration Likelihood Model (ELM) of informational influence. The

ELM was developed by Petty and Cacioppo (Petty and Cacioppo, 1981; 1986) in order to account

for reported differences in influence results across individuals and contexts. The ELM states that in

different situations, different message recipients will vary in the extent to which they cognitively

elaborate on a particular message, and these variations in elaboration likelihood affect the success

of an influence attempt, along with other factors. Elaboration involves attending to the content of

the message, scrutinizing and assessing its content, and reflecting on issues relevant to the message.

Because of the cognitive effort involved, receivers don't elaborate on every message they receive,

and some receivers elaborate on fewer messages than others do.

According to ELM, informational influence can occur at any degree of receiver elaboration,

but as the result of very different influence processes; high levels of elaboration represent a central

route to influence, while low levels result in a peripheral route. The central route occurs when

recipients carefully consider the issues presented by the message, whereas the peripheral route

occurs when recipients use simple decision rules to evaluate the message rather than analyzing its

content. As elaboration likelihood decreases, peripheral cues have increasingly important effects on

recipient attitude, belief and consequent influence, since recipients use these cues as heuristics or

decision rules rather than undertaking the greater cognitive effort of elaboration (Petty and

Cacioppo, 1986). The central and peripheral routes are viewed as the extremes of a single

7
underlying elaboration dimension. Thus at moderate elaboration likelihood levels, influence

processes involve a complex mixture of both central and peripheral route processes.

The Role of Argument Quality and Peripheral Cues

According to the ELM, when an individual is able and willing to cognitively elaborate on a

persuasive communication, the quality of the arguments contained within the communication will

determine the degree of informational influence (Petty and Cacioppo, 1986). Thus the ELM

identifies argument quality as the critical determinant of informational influence under conditions

of high elaboration likelihood. When an individual is either unable or unwilling to process the

arguments presented in a message, peripheral cues will play a more critical role in the influence

process.
Peripheral cues are informational indicators that people use to assess content other than the

content itself (Petty and Cacioppo, 1986). A potentially infinite number of these heuristics exist in

interpersonal communication contexts (Gergen, 1982). For instance, in groups people are

influenced by consensus cues and attributes of the group leader such as charisma. Individually,

people often use cues pertaining to the message’s source when they are unable or unwilling to

expend the effort to elaborate on the message’s content (Petty and Cacioppo, 1986). Research

indicates that individuals following the peripheral route can be influenced by the source’s

attractiveness, likeability, and credibility. In order to understand how individuals are influenced by

advice transmitted via e-mail in the workplace, we are interested in the peripheral cues and that are

most salient in such contexts.

Early research on the use of peripheral cues in CMC contexts found that the social presence

of the communicator varies by media and affects persuasive communication by attenuating potential

peripheral cues (Short, Williams and Christie, 1976). For example, research on rich versus lean

media (Worschel, Andreoli and Eason, 1975; Andreoli and Worschel, 1978) and vividness effects

8
(Pallak, 1983) supports the notion that lack of peripheral cues and heuristics for informing attitude

judgments serves to force message recipients to elaborate on the nature of the arguments presented,

especially when the arguments are complex (Chaiken and Eagly, 1976). In this stream of research,

source likeability was the peripheral cue investigated, a cue that is clearly attenuated by text-based

media. But not all peripheral cues are attenuated by computer-mediated communication. While

face-to-face interaction provides the many peripheral cues that enable us to establish a shared

context (Zack 1993), we know that peripheral cues do operate in CMC contexts (Walther, 1992).

For example, people use cues to delete e-mails they receive without scrutinizing the arguments they

contain or even reading them at all.

While there are many potential cues which may operate within a CMC context, this research

focuses on one peripheral cue, source credibility, that the literature suggests may be relevant both

within the context of CMC and within knowledge work more generally construed. In the context of

text-based messages, source credibility refers to a message recipient's perception of the credibility

of a message source, reflecting nothing about the message itself. Early laboratory experiments on

the role of credibility in informational influence found significantly more opinion change in the

direction advocated by the communicator when the material was attributed to a high credibility

source than when it was attributed to a low credibility source (Hovland, 1951; Hovland et al.,

1953). These results were thought to be due to the association of high credibility sources with

favorable outcomes. More recently, ELM researchers have taken a cognitive response approach to

source credibility. In this view, higher levels of source credibility can interact with other variables

to produce patterns quite different from the simple enhancement effect produced by Hovland

(Heesacker, Petty and Cacioppo, 1983). For example, when people are highly involved in a

message topic, source credibility has little impact on attitude change since individuals will

scrutinize the argument rather than assess this peripheral cue. In contrast, when individuals are not

9
involved in a topic, source credibility has been found to be an important predictor of attitude change

in general (Petty, Cacioppo and Goldman, 1981). In organizations, source credibility has been

found to influence collaborative filtering (Ehrlich and Cash, 1996), and to affect members' attitudes

towards organizational change implementation processes (Ellis, 1992). In information systems

research, source credibility has been shown to positively influence low-participation users’

acceptance of recommendations made by knowledge-based systems (Mak and Lyytinen, 1997;

Mak, Schmitt and Lyytinen, 1997). On the Internet, perceptions of source credibility play an

important role in our judgments of cognitive authority (Rieh and Belkin, 1998).

In addition to its role as a peripheral cue under low elaboration likelihood conditions, source

credibility has been conceptualized as a variable which might bias message processing - changing a

message recipient's propensity to support or counter argue an advocated position. Further, research

suggests that, under conditions of high elaboration likelihood, a credible source may potentially

serve as an additional argument in favor of an advocated position (Chaiken and Maheswaran,1994).

Thus, although it is known that source credibility plays an important role in informational influence,

the mechanisms underlying the role that it plays appear to be complex.

Conceptual Model
The ELM enables us to make predictions about the relative impact of various factors on information

adoption under different levels of elaboration likelihood, but it excludes the role of information usefulness

in the influence process. Yet we know from the adoption literature that usefulness is a key construct in

adoption behaviors. Empirically, the amount of adoption variance explained by TAM generally exceeds

levels of persuasion variance explained by theories of informational influence. We therefore expect that

usefulness will be more strongly associated with information adoption than the peripheral cues and

argument factors studied under ELM. Further, we posit that perceptions of information usefulness can be

explained by theories of informational influence. For these two reasons, in the context of information

adoption, we expect that perceptions of information usefulness will mediate between influence outcomes

10
(i.e., intentions toward an advocated behavior) and the constructs of argument quality and source

credibility. In this way, ELM processes are influential to the extent that they contribute to the perceived

usefulness of the message.

Our integrated model of the information adoption process is presented in Figure 1. From

TRA/TAM, we know that a critical aspect of the process by which individuals in organizations act on an

advocated issue or behavior is the extent to which they believe the information contained within a message

is useful for the task at hand. From ELM, we understand how this process depends on elaboration

likelihood, and have suggested two likely antecedents of usefulness from this stream of research. In this

way, our model integrates key internal validity factors from the informational influence literature with

theory from TRA and TAM that has established the need to ascertain the external validity of a particular

message. Thus, we view organizational knowledge transfer in terms of the cognitive structures involved in

assessing the usefulness of received knowledge and the influence processes involved in changing them.

Argument Quality

Information Information
Usefulness Adoption
Source Credibility

Figure 1. Model of Information Adoption

The discussion in the preceding sections and consequent relationships as depicted in Figure

1 are summarized in the following hypotheses, and apply to the context of received, computer-

mediated advice.

H1. The higher the perceived argument quality of a message, the more useful the message

will be perceived to be.

H2. The higher the perceived credibility of a message source, the more useful the message

will be perceived to be.

11
 H3. Messages perceived to contain information of high usefulness will be associated with

higher levels of information adoption than messages perceived to contain information of low

usefulness.

H4. Usefulness will mediate the effects of source credibility on information adoption, and

argument quality on information adoption.

Moderators of the Informational Influence Process

According to the ELM, two factors influence the degree of elaboration that a receiver is

likely to engage in – ability and motivation – and both must be present for extensive elaboration to

occur (Petty and Cacioppo, 1986). Researchers in ELM manipulate these moderators to understand

information processing under different contexts. By measuring differences in factors that are likely

to alter levels of elaboration likelihood, we can find evidence of interactions among central and

peripheral information processing activities. We therefore include them in our in situ model to

confirm the basic tenets of ELM and to learn more about how central and peripheral processes work

during actual interpretation processes.

Prior expertise of the message recipient alters elaboration likelihood by affecting

individuals’ ability to process: Higher levels of receivers' prior knowledge and comprehension

about the message topic increase the quantity and depth of issue relevant thoughts that occur to

them and their ability to understand them, which in turn increases the likelihood of elaboration and

decreases reliance on peripheral cues (Ratneshwar and Chaiken, 1991). Thus, drawing on the ELM,

we would predict that in forming attitudes toward received advice, experts and gurus will attend

less strongly to peripheral cues such as source credibility than will message recipients with lower

levels of expertise, and will attend more strongly to the quality of arguments. Source credibility

will be most important for messages received by non-experts.

Motivation levels also alter recipients’ elaboration likelihood. High levels of recipient

involvement – operationalized as the personal relevance of the issue to the reader – tend to motivate

increased elaboration on the message: As issues become increasingly important to the receiver, he

12
or she is more likely to undertake the cognitive effort to thoughtfully consider the message (Petty

and Cacioppo 1986, Petty, Cacioppo and Goldman, 1981). Receivers that are highly involved with

the message issue are likely to engage in high elaboration, while those that are not involved will be

less likely to engage in elaboration and more likely to be influenced by peripheral cues such as

source credibility. (Petty, Cacioppo and Goldman,1981; Stamm and Dube, 1994). We expect to

find that argument quality is more salient than source credibility under such conditions, while less

involved recipients will be influenced by high levels of perceived source credibility.

Figure 2 presents the overall model of information adoption. Recipients’ perceptions of

message usefulness are the direct determinant of information adoption, mediating the ELM-based

informational influence processes described above. Message recipients’ domain-based expertise

and involvement serve as moderators of the credibility-usefulness and argument quality-usefulness

relationships by affecting elaboration likelihood through mechanisms of ability and motivation.

RecipientExpertise
Recipient Expertise
RecipientExpertise
Recipient Involvement
Argument Quality

Information Information
Usefulness Adoption
Source Credibility

Figure 2. Moderated Model of Information Adoption

The preceding discussion and consequent relationships as depicted in Figure 2 are

summarized in the following hypotheses:

H5a. The greater the recipient's expertise in the topic of the message, the more argument

quality affects perceived message usefulness.

H5b. The greater the recipient's expertise in the topic of the message, the less source

credibility factors affect perceived message usefulness.

13
 H6a. The greater the recipient's involvement in the topic of the message, the more argument

quality affects perceived message usefulness.

H6b. The greater the recipient's involvement in the topic of the message, the less source

credibility factors affect perceived message usefulness.

METHODS

Overview
The theoretical model above was assessed through a study of the e-mail behavior of

consultants in the consulting division of a major multinational public accounting firm. It employed

both qualitative and quantitative data collection and analysis in a two-phase approach. First, forty

semi-structured interviews were undertaken to gather rich qualitative data about the phenomenon.

Based on the findings from the interviews, a survey was developed and conducted with a second

sample of consultants regarding particular e-mails they had received.

The unit of analysis for the study is the individual e-mail message. Informants were asked

to select actual e-mail messages they received that contain advice, recommendations, and

suggestions for how to solve problems. This is consistent with the definition of knowledge as

professional intellect involving knowing how, what, and why (Quinn, Anderson and Finkelstein

(1996)).

Below we briefly describe the qualitative phase of this study. We then discuss the methods

and results of the quantitative phase of this research. Before we begin these descriptions, we

discuss the research site, since it is common to both phases of the study.

Research Site
This research took place at the city office of a multinational public accounting firm

comprised of 103,000 professionals worldwide. However, the study was limited to their North

American operations, henceforth referred to as ICBM, a pseudonym used to ensure confidentiality.

ICBM revenues were $12 billion in 1999. ICBM is organized into three divisions – assurance, tax,

14
and consulting. Consulting is the smallest but fastest growing of the three divisions. Each division

is structured into multiple lines of business. While several of the top consulting firms have

specialized in information technology consulting within the past decade, ICBM is not one of these.

ICBM has geographic headquarters throughout the U.S., each with its own consulting division

serving the five lines of business (finance, manufacturing, IT & knowledge management, consumer,

and strategy). Informants were all members of the Consulting division and headquartered at the city

office where this research was conducted, but varied across lines of business.

Consulting organizations typically have mobile workforces, and ICBM is no exception.

Only administrative staff and senior partners have offices. The rest reserve office space as needed

on a daily basis. Consultants spend most of their time at their client sites, on assignments ranging

from a few days to several months or more. The mobile nature of their work makes them

extremely reliant on e-mail and other telecommunications technologies for getting their work done.

All informants were officially headquartered in one city but they worked at client sites all over the

U.S.. So while interviews took place in the office of one city, informants were geographically

disbursed to client sites, and this minimized the likelihood of regional effects.

The e-mail system used by ICBM is a commercial e-mail system manufactured by a

Canadian company. It has standard e-mail system features such as file attachment capabilities and

the ability to track sent e-mails.

Preliminary Qualitative Study: Methods and Analyses

Since the model integrates two theoretical perspectives that together have not been applied

to this domain, a qualitative study was deemed necessary as a first step in order to understand the

phenomenon and refine survey instruments. This study consisted of hour-long face-to-face

interviews with the researcher.

Permission was attained and arrangements were made for gathering data in ICBM-Boston in

Summer 1997. At that time an e-mail was sent to all city office consultants (218) under the two

senior partners who were supporting the study. This e-mail described the study generally as one on

15
knowledge management, and requested hour-long interviews with each participant. In this way a

comprehensive sampling strategy was used and every case was solicited. Forty-six consultants

responded to this e-mail by scheduling an interview. Of these 46, six ultimately canceled, resulting

in an interview response rate of 18%.

Interviews consisted of two parts and required informants to bring with them preselected e-

mail message printouts. First, an ethnographic, open-ended interview was conducted to elicit

informants’ perceptions regarding how they process advice-based e-mails in general, and how they

processed the e-mails they preselected, in particular. (See Appendix I for the interview guide used.)

Second, informants completed quantitative measures regarding their perceptions of each of their

preselected e-mail messages. While they did this, the researchers took note of the objective

characteristics of each e-mail message. This second part of the interview was primarily for the

purpose of instrument refinement and validation of the quantitative measures to be used in the

second phase of the study.

Prior to each interview, informants were instructed to save the next two to four e-mail

messages that they received that consisted of work-related but non-administrative advice or

recommendations. Informants were instructed to select messages that contained advice, as opposed

to those that were actually status reports, fact lists or logistical coordination. (See Appendix II for

actual instructions sent to informants.) Several e-mails are presented below to illustrate the type of

advice-based message we collected. All names and identifying information have been altered to

protect confidentiality.

John,
Just a couple of thoughts…You should give them an example of the type of answer
you are looking for. I think you should explain that the (Organization X) is considering
allowing XYZ to reimburse them for their expenditures… (47b)

Dana,
I was thinking about your request and even though I sent you myXYZ Test Mgt
handout files it doesn’t make sinse (sic) that you would be using them in the renovation

16
 methodology. I believe that your renovation methodology should just refer to the testing
methodology. What do you think?… (112ma)

Frank,
On the Y2K conference call the other week when we discussed the merger
implications to our practice it was stated that XYZ already had a good process for the
renovation/ conversion stage. Should we be spending a lot of resources to develop one when
we will be able to share theirs in January? (03f)

Analyses of the interview data sought to understand the extent to which they reflected the a

priori theoretical model and to explore ways in which the theory might not be applicable. Content

analyses of these data was performed following the first-order analysis procedures outlined by

Spradley (1979). First-order analysis seeks to represent the data within domains as they are

conceived of by the informants. Interview tapes were transcribed and then memoed, and a domain

analysis was conducted to identify major themes or domains. Once the domains were specified and

coded, they were then mapped onto the theoretical constructs of interest. The domains mapped

sufficiently well to support the next step of undertaking a confirmatory study. Interviews also

informed development of the quantitative survey instrument, since interviewees were asked to

complete the survey instrument in addition to answering interview questions.

From these and similar data we concluded that the model was sufficiently supported to

proceed with the quantitative, confirmatory phase of the study, as follows:

Quantitative Study Methods

Following completion of the qualitative data collection, quantitative data were gathered by

soliciting survey responses from the 178 consultants who had not been interviewed in the first phase

of the study. We sent these 178 consultants a second e-mail from the two top partners requesting
their participation and informing them that they would soon be receiving a survey packet in the

mail. A survey packet was then sent to the homes of this group of 178. The packet included a

cover letter from the two senior partners sponsoring the study, stating their endorsement of the

17
study and their selection of the recipient as a participant. It also included a letter from us explaining

in general terms the nature of the study (knowledge management), the instructions for survey

completion and e-mail selection, and an explanation of the procedures we would be using to ensure

confidentiality and proper handling of responses. Informants were asked to complete the

demographics questionnaire once, and the e-mail assessment instrument twice (two copies were

included) as they pertained to each of the two e-mails selected, and to send these to us along with

the actual e-mails they had used as the basis for their responses.

As above, survey respondents were asked to select the next two actual e-mails they received

that fit the definition of advice or recommendations concerning a non-administrative task. Other

than demographics, all survey questions pertained to these two e-mails. As in the first phase of the

study, instructions were designed so that informants selected messages that were advice-related.

Recipients’ self-response estimation of the percent received of these type of e-mails was 36%. By

asking informants to complete instruments for the next two e-mails they received in time of the type

we were interested in, we minimized the threat of e-mail selection bias by the informants, since this

would occur randomly. The date on all e-mails received was compared to the date of survey

mailing to verify that the e-mail had been received by the respondent within two weeks of the

mailing, as opposed to having been selected from archives.

Out of these 178 informants solicited, three had left the organization, two were on long term

assignment out of the country, and two requested that they be removed from the study due to time

constraints. Of the remaining consultants, 37 returned the completed survey packet. To increase

the sample size, a second packet was mailed to the 144 consultants who did not respond to the

initial survey as mailed. In this second mailing, the ordering of questions was rearranged within the

questionnaire to allow for testing of order effects. Twenty-two consultants returned this packet.

Researchers then telephoned those consultants that had not returned either packet to solicit

additional responses. This resulted in an additional 4 surveys being returned. In total, surveys

were returned by 63 ICBM consultants for a 35% response rate. Fifty-four of these completed two

instruments as requested – one for each of two separate e-mails received – and sent in these two e-

18
mails along with their corresponding surveys. Five completed only one survey and sent in the one

e-mail that corresponded to it, and the remaining three completed one e-mail instrument but did not

include the actual e-mails.

While a 35% response rate is not high, it is reasonable considering the demands placed on

the time of these consultants. These respondents are representative of the city office as a whole in

terms of their age and division, however they do reflect a slightly higher than average percentage of

senior managers and above (23% of this sample are senior managers, while only 18% of employees

of this office are senior managers). Perceptual measures of source credibility, argument quality,

usefulness, adoption, expertise, and involvement were collected, and these items are listed in

Appendix III. All actual e-mail messages except three were also collected. The focus on actual

messages during interview and survey completion allowed us to target the informant on the specific

stimuli that elicited the focal attitudes.

Measures
Each of the constructs in the theoretical model are operationalized using validated measures

or measures adapted from validated instruments. All measures were pre-tested prior to use on the

sample of interviewees.

The section on demographics and electronic communication behavior was adapted from

those measures used by Sproull and Kiesler (1986). Items adapted from Wu and Shaffer (1987) are

used to measure source credibility and information adoption. The dependent measure was

operationalized utilizing three items which assessed self-reported information adoption.

Information usefulness and argument quality items are a subset of those used and validated by

Bailey and Pearson (1983). The measures of issue involvement and expertise are from Stamm and

Dube (1994).

A number of authors have studied the underlying structure of source credibility, both at the

individual and organizational level (see Newell, 1993 for a review). While many dimensions have

been identified, two sub-dimensions originally identified by Hovland (1951) have consistently

19
emerged -- competence and trustworthiness (Weiner and Mowen, 1986; McGinnies and Ward,

1980; Mowen, Wiener and Joag, 1987). Competence-based source credibility refers to expertise

and expertness, while trustworthiness-based source credibility is a function of the perceived

character and integrity of the source. We utilized this sub-dimensional structure in our

operationalization of source credibility.

QUANTITATIVE ANALYSIS AND RESULTS

Demographics and Descriptive Statistics

Of the 63 ICBM survey respondents, 68% are male, reflecting the higher percentage of male

consultants employed by the organization. Respondents’ mean job experience is 4.67 years, mean

age is 36, and mean organizational tenure is one year, reflecting the high turnover typical in

consulting organizations. As for educational level attained, 100% of consultants have Bachelors

degrees, 40% of these have Masters-level degrees, and 8% have some post-Masters. Respondents

worked in various industrial sectors, as indicated in Table 1:

Line of Business % Respondents

Financial Services 19%
Manufacturing 30%
IT & Knowledge Management 33%
Consumer Markets/Retail 6%
Strategy/Value Management 8%
Other 4%

Table 1. Industries in which Respondents Consult

Due to the spread of expertise across both technical and non-technical domains, we found

wide disparities in technical skill levels across informants.

Respondents were also asked about their mediated communication behavior. On the

average, respondents reported receiving 14.4 e-mails per day and sending 8.4 per day. This is in

addition to the 7.5 voice mails they receive and the 6.5 voice mails they send each work day. They

check their e-mail an average of 2.7 times per day when they are at their client site, and 4.7 times

20
per day at headquarters. On the whole, this mobile work force relies on mediated communication to

perform their everyday work, however, for the majority, the content of their everyday work is not

IT-based. T-tests were used to test for effects of respondent characteristics. There were no

significant differences by age, education, line of business, organizational tenure, or any of the

telecommunications characteristics measured.

The Measurement Model and Preliminary Analyses

For reasons described below, results are reported based on the analysis of the first e-mail

cohort only, for a sample size of 59. Data were assessed for normality visually and by the skewness

and kurtosis of each. All measures are below the critical value for skewness at alpha=.01 and n=59.

Kurtosis for a normal distribution equals 3. Kurtosis values are less than three, but not significantly

so. This indicates slight platykurtosis, resulting in actual power being less than nominal power.

Table 2 below presents the means and standard deviations of all constructs in the model for

the reduced sample of cohort-one e-mails (n=59), along with construct correlations. The largest

correlation (.42) between independent constructs is between usefulness and argument quality, but is

not enough to cause a multicollinearity problem, as would occur if the correlation were greater than

.8 (Billings and Wroten, 1978). As is predicted by theory, involvement and expertise are correlated,

but these two constructs are not analyzed together in any model. All variables are measured using a
standard 7-point Likert scale.

Variable Mean S.D. Scale 1 2 3 4 5

Range
1. Argument Quality 5.42 1.25 1-7
2. Source Credibility 4.83 1.27 1-7 .24
3. Usefulness 5.12 1.44 1-7 .42 .34
4. Information Adoption 3.52 1.03 1-9 .25 .32 .49
5. Expertise 4.81 1.46 1-9 .17 .33 .31 .34
6. Involvement 4.40 1.79 0-1 .07 ..39 .33 .34 ..52
Note: N=59.
Table 2. Descriptive Results and Correlation Matrix

21
 The measurement model was tested by examining the reliability of individual items, the

internal consistency of constructs, and discriminant validity. An acceptable measurement model

was built using reliability and factor analyses of the first message reported on by each respondent

(n=59), and hypotheses were tested based on this model using standard regressions, with results

presented below. However, an attempt to build an acceptable measurement model using the second

of the two e-mails reported on by these respondents failed, as the Cronbach alpha for the dependent

variable was unacceptably low (.52). It is not clear why this is the case, but we believe that the

heavy time constraints placed on these consultants may have created fatigue effects as they hurried

to finish the survey task. Because of this, a repeated measures analysis was not possible with the

second cohort of e-mail cases. Results are therefore reported based on the analysis of the first e-

mail cohort only.

Discriminant validity was assessed using factor analysis on cohort-one e-mails. Internal

consistency was assessed using Cronbach’s alpha, and the reliabilities reported in Table 3 are well

within accepted levels (Nunnally, 1967, p. 226).

Construct No. of Cronbach’s Alpha –

Items e-mail – Cohort One
N=59
Information 3 .6412
adoption
Source 4 .7905
Credibility
Argument 3 .8225
Quality
Usefulness 3 .8601
Involvement 2 .7306
Expertise 2 .7426

Table 3. Internal Consistency of Model Constructs

Hypothesis Testing Using OLS

22
 Hypothesis 1 states that higher levels of perceived argument quality will be associated with

significantly higher reported levels of perceived message usefulness. Similarly, hypothesis 2 states

that higher levels of perceived source credibility will be associated with significantly higher levels

of perceived message usefulness. To test for these effects, usefulness was regressed onto both

argument quality and source credibility, for the first cohort of e-mails only (n=59). Results were

significant as predicted by the model (F=8.25, p<.001), with an adjusted r2 of .20 (d.f.=2,57). Both

independent contructs were significant in this model: The beta coefficient for argument quality was

.34 (t=2.73, p<.01), while that of source credibility was .24 (t=1.94, p<.05).

Hypothesis 3 states that higher levels of perceived usefulness will be significantly associated

with higher levels of reported information adoption. When perceived information adoption was

regressed onto perceived usefulness, results were significant as predicted (F=18.91, p<.001), with

an adjusted r2 of .22 (d.f.=1,58). Hypothesis three is confirmed.

Hypothesis 4 posits that usefulness mediates the relationships between argument quality and

information adoption. Three regressions are performed to confirm mediation (Baron and Kenny,

1986). First, the mediator - in this case usefulness - is regressed onto the independent variables of

argument quality and source credibility. We know this relationship to be significant as hypothesis 1

above (an adjusted r2 of .20, F= 8.25 (d.f.=2,57; p<.001). Second, the dependent variable of

information adoption is regressed onto the independent variables of argument quality and source

credibility. Results are also significant, with an adjusted r2 of .11, F= 4.80 (d.f.=2,57; p<.05). Both

independent contructs were also significant in this model, although less so for source credibility

than for argument quality: The beta coefficient for argument quality was .30 (t=3.36, p<.01), while

that of source credibility was .14 (t=1.57, p<.10).

Finally, the dependent variable of information adoption is regressed onto both the

independent variables of argument quality and source credibility, and the mediator, usefulness.

23
Results are again significant, with an adjusted r2 of .22, F= 6.95 (d.f.=3,56; p<.001). The effect of

argument quality on information adoption is significant in the second regression (without

usefulness), but is non-significant in the third regression which includes usefulness (Β=.11, t=1.15).

The effect of source credibility on information adoption is also significant in the second regression

(without usefulness), but is non-significant in the third regression which includes usefulness

(Β=.08, t=.98). Usefulness is highly significant in this last regression (Β=.40, t=4.18 Sig. T<.001).

These findings, taken together, indicate pure mediation of the both the independent variables in the

model, argument quality and source credibility, by the mediator usefulness.

Hypothesis 5 states that expertise will moderate the effects of argument quality and source

credibility on usefulness. Our model suggests that individuals with higher levels of expertise will

perceive the quality of the message argument to be of high utility, while those with lower levels of

expertise will find credibility cues more useful. Hypothesis 6 repeats this theoretical argument for

the moderator of involvement, since involvement also increases elaboration likelihood. To test for

these moderating effects, moderated multiple regression models were built in which all data were

standardized, and product terms were built by multiplying the indicators of the independent

construct with those of the hypothesized moderator (Jaccard, Turrisi, and Wan, 1990). The

independent constructs of argument quality and source credibility were analyzed separately since

the hypothesized moderating effects are in opposing directions and would therefore obscure one

another. Table 4 below presents results of these analyses. In it we see that the beta values of the

interaction terms are significant for argument quality, but only marginally so for source credibility.

The F-statistic of the change in R2 for each overall model indicates the extent to which whether the

addition of the moderator significantly increased the variance explained by the model over that

explained without the moderator.

24
 Overall Main Moderator Interaction Term F-stat for
Model Effect ∆R2
H5a: Adj. R2 = .34 β=3.75 β=2.92 β=5.44 F=4.55**
Arg Quality F=20.14 t=2.64 t=2.90 t=3.01
X Expertise Sig. < .001 Sig. T= .01 Sig. T= .004 Sig. T= .003**
H5b: Adj. R2 = .09 β=3.74 β=2.16 β=4.95 F=2.87
Source F=4.71 t=1.53 t=1.57 t=1.65
Cred. Sig. < .01 Sig. T= .13 Sig. T= .12 Sig. T= .10
X Expertise
H6a: Adj. R2 = .35 β=2.97 β=2.15 β=4.41 F=8.22***
Arg Quality F=21.23 t=2.10 t=2.25 t=2.47
X Sig. < .001 Sig. T= .04 Sig. T= .03 Sig. T= .02*
Involvement
H6a: Adj. R2 = .12 β=2.56 β=1.29 β=3.47 F=3.47*
Source F=6.02 t=1.41 t=1.32 t=1.54
Cred. Sig. = .001 Sig. T= .16 Sig. T= .19 Sig. T= .13
X
Involvement
Table 4. Results of Regressions Testing for Moderation Effects, H5 and H6.

These analyses indicate that argument quality significantly interacts with both expertise and

involvement to affect perceptions of message utility, in support of hypotheses 5 and 6. Message

recipients who are experts in the topic or highly involved in it will be find argument quality more

useful than will novice recipients, or those not involved in the topic at hand. We see weaker

evidence of moderation of source credibility by expertise and involvement, but only the effect of

involvement is significant. Figure 3 below presents these results graphically for the moderator of

expertise, by splitting the sample (n=59) at the median expertise. This technique provides visual

support for the consequences of moderation by elaboration likelihood. We see that usefulness is

significantly correlated with source credibility for those recipients reporting lower levels of

expertise, but this correlation is not significant at higher reported levels of expertise. This is

consistent with theory, since the influence of the peripheral cue of source credibility decreases as

elaboration likelihood increases. Usefulness is significantly correlated with argument quality only

at high levels of reported expertise, as is also predicted by theory. These findings support the

25
theory that elaboration likelihood moderates the relationships among argument quality, source

credibility, and usefulness.

0.6

Strength of 0.5
Correlation
with 0.4
Usefulness
0.3 Credibility
Argument Quality
0.2

0.1

0
Low Expertise High Expertise

Note: Pearson correlation coefficients greater than .45 are significant.

Figure 3. The Moderating Effect of Expertise

DISCUSSION
This study lends support to an informational influence view of information adoption as it

relates to receiving advice in organizational contexts. In the study, it was found that consultants

varied in the degree to which they were influenced to adopt specific pieces of advice based on their

perceptions of the usefulness of that advice. Importantly, perceptions of usefulness were

26
differentially affected by elements of the received messages, depending on the elaboration

likelihood of the situation.

According to the ELM, individuals following the peripheral route are not able or motivated

to process the arguments in a message and thus are influenced primarily by peripheral cues. In our

study, the peripheral cue selected for investigation was source credibility. The rationale for

choosing this cue from among the many cues potentially operating in CMC contexts was as follows:

First, early research suggests the importance of source credibility in learning and knowledge

transfer (Hovland, 1953). Second, many of the other peripheral cues typically examined in studies

of ELM, such as perceptions of source attractiveness and likeability, are less likely to be salient in a

CMC context than in face-to-face or other visually-rich situations. Finally, the heuristic "trust an

expert" seemed plausible for the task at hand - determining which pieces of advice to follow in a

work context.

Source credibility is significantly associated with information usefulness in our model, and

tends to function as a peripheral cue as we had predicted based on the ELM. However, empirical

support for this function is weak, consistent with the notion that source credibility plays a more

complex role in the ELM than that of a simple heuristic cue (Heesacker, Petty and Cacioppo, 1983).

As noted, in previous research, source credibility has been found to function as an additional

argument factor for those following the central route, and some have claimed that it may bias

people's perceptions of argument quality (Chaiken and Maheswaran, 1994). Ex post facto, it is

difficult to test for these alternative hypotheses regarding the role of source credibility. While our

findings could be an indication that central processors were using source credibility as an additional

argument for advice usefulness, further research is needed which is specifically designed to

examine the impact of source credibility under different elaboration conditions in the information

adoption process.

Without strong results explaining how source credibility functions in the information

adoption process, the model is not very good at accounting for the behavior of peripheral processes.

It seems likely that some additional cues played a role in the influence of peripheral processes in

27
this study. While prior research has identified several cues that have an impact under the peripheral

route, it may be that cues specific to a CMC-context are most critical in CMC-based information

adoption behavior. For instance, cues relating to the relationship of the source to the recipient (e.g.,

relative hierarchical status) or characteristics of particular e-mail themselves (e.g., the subject

heading; number of people copied in the message; status of people copied in the message) may be

particularly fruitful as the basis for further investigation in order to gain a better understanding of

the heuristics that operate in CMC-based knowledge work.

Perhaps the most important result of our study is the finding that information utility

assessments mediate influence processes in knowledge adoption contexts. This was found to be the

case for both central and peripheral processes. In these data, argument quality (and to a lesser

extent source credibility) were found to be associated with usefulness but not information adoption.

The influence of these two factors occurs as a function of their transformation into perceptions of

usefulness, a transformation that appears to be affected by the elaboration likelihood of the advice

context. One may wonder why this important role of usefulness has not been acknowledged in

previous work under the ELM. Two possible explanations could account for this "oversight." First,

most studies conducted under the rubric of the ELM have been tested in situations where the

usefulness of messages has not been allowed to vary in the experimental task. For instance, many

ELM studies have been conducted as paper-and-pencil tasks where participants receive a single

communication about an issue (e.g., Senior Year Comprehensive Exams) and then are asked to form

a judgment. Obviously, in such tasks, usefulness is held constant., making it difficult to assess a

mediating role played by usefulness. Alternatively, it may be that information usefulness is

particularly relevant to the domain investigated in this paper. In organizational advice giving and

receiving, the objective may not be to convince others to change their attitudes toward a given

behavior as much as it is to convince others that the advice given is credible. While this is a subtle

distinction, it leads to a different model of the influence process - one in which assessments of

usefulness (i.e., attitudes toward the information) rather than attitudes toward the behavior

determine adoption intentions.

28
 We believe that our conceptualization of the information adoption process, with usefulness

playing a major mediating role, offers interesting avenues of research both for ELM researchers

interested in advice adoption and for others interested more generally in knowledge transfer. For

example, we need to understand how well this model reflects information adoption in other

organizational contexts. Perhaps knowledge workers that are afforded more slack time for research

and innovation take a less instrumental approach to information adoption, one in which usefulness

figures less prominently. Also, this model may be more or less valid for explaining information

adoption via technologies other than e-mail. For instance, how would a “pull” technology such as a

discussion board compare to the “push” technology of e-mail studied here?

This study contributes to theory in information systems in several ways. When knowledge

sharing takes place across time and distance it necessarily transpires via various media rather than

face-to-face. For this reason knowledge transfer is increasingly a problem of interest to information

systems researchers, and this study offers a theoretically grounded approach to this complex

problem. Here, knowledge transfer has been operationalized as the adoption of advice, and

informants report that a significant amount of their received e-mail falls into this category. These

findings support the accruing evidence that CMC is being used for ambiguous and complex

communication in organizations (Markus, 1994). Knowledge work is becoming increasingly

mobile, global and reliant on telecommunications, and this study presents a theory of how one form

of knowledge is transferred in such contexts.

The central contribution to information systems research here is the integration of ELM with

TAM (Davis, 1989) and the light this may shed on our understanding of the antecedents of

usefulness. In this respect, this study builds on the growing body of literature on the antecedents of

usefulness in the context of TAM (Agarwal and Prasad, 1997; Karahanna, Straub, and Chervany,

1999; Venkatesh and Davis (2000). These studies have generally taken an inductive approach to

understanding predictors of usefulness, as is appropriate given the lack of available theory in the

technology domain. By focusing on adoption of information rather than adoption of technology, we

have been able to apply a well-developed body of theory to the problem of understanding

29
antecedents of usefulness. It remains to be seen whether this integrative model can contribute to

our understanding of technology adoption processes.

It should be noted that this research is limited to non-interactional recipient behaviors, in

response to a single mediated message. Yet knowledge work is an interactional process (Nonaka,

1994) that transpires in organizations over all available media and face-to-face. We chose to bound

this study in this way in order to reduce some of the variance inherent in this highly complex

phenomenon. By narrowing the scope as such, we have produced results that we hope will inform

broader research that does take relational factors into account. By limiting the investigation to a

single medium channel (CMC-based e-mail), we were able to minimize confounding channel

factors, but this also prevents us from making any claims about the media, since no comparisons

have been made. It also raises the problem of whether the novel findings here - the mediation of

usefulness and the weak source credibility results - are due to the CMC media or the knowledge

adoption domain. Since no manipulation was performed to investigate this, we can only claim

generalization of these findings to knowledge adoption that results from e-mail content and not any

other form of knowledge transfer. However, we believe that it is the context of ongoing

organizational work and the knowledge domain that requires modification of ELM to take

perceptions of information utility into account, not the mediated nature of the knowledge. This

problem is a key research question that remains to be addressed.

Practically speaking, by understanding how people adopt mediated advice, we can design

processes that support effective information adoption. And since knowledge transfer and

information adoption are embedded in key organizational processes such as new product

development, systems implementation, organizational restructuring, and change management, such

design issues present an important avenue for this type of research. Practitioners can benefit from

understanding the information adoption component of knowledge transfer as dual-process

informational influence that reflects complex tradeoffs among central and peripheral processing

routes. For example, it may sometimes be appropriate to design knowledge work so that

elaboration likelihood is increased when central route processes are optimal. For important tasks

30
and decisions, distractions and multi-tasking can be minimized to induce elaboration. This leads us

to ask why and how some practitioners are able to assess advice quickly and effectively despite

multitasking and other job distractions. And by viewing knowledge transfer from the perspective of

information adoption, practitioners can understand why it may be important to take non-message

factors - such as recipient and context characteristics - into account when sending advice-based

messages.

Practitioners will also be interested in the implications of this research for the design of

knowledge management systems. Our findings suggest that the external validity of potentially

shared knowledge - how useful it is for the problem at hand - is actually more important than but

also encompassing of the internal validity of that knowledge. This provides empirical support for

the knowledge management precept that effective knowledge transfer systems should be designed

to balance these two attributes of knowledge. We now have a theoretical explanation for why

knowledge management systems with a central validation function risk sacrificing external validity

and hence the usefulness of that knowledge for adoption assessment. The same can be said for

validation processes that strip off so much of the knowledge context (such as the source) that few

cues are available for peripheral processing. At the same time, valid knowledge is useful

knowledge, so organizations cannot neglect validation processes. For mediated knowledge transfer,

we find that perceptions of internal validity contribute greatly to our ability to assess external

validity. But it is the external validity that ultimately determines whether some advice will be

adopted or not.

31
APPENDIX I. QUALITATIVE INTERVIEW GUIDE

1. Do you receive work-related e-mails that consist of advice and recommendations? If so, how

many in what period of time? In general, are they helpful to you? Do you use them? Do you

solicit them?

2. When you receive information from a colleague via e-mail, how do you decide whether or not to

act on its recommendations? Probe if credibility is mentioned: What you mean when you say that

someone is credible? Probe if argument quality is mentioned: What do you mean when you say the

advice is well-argued? Probe if usefulness or relevant is mentioned: What do you mean when you

say the advice is useful/relevant?

2. When trying to solve a problem, what things do you look for on an e-mail to decide whether or

not to follow the advice in the e-mail?

3. When you receive advice or a recommendation via e-mail, how do you decide whether or not the

source is credible? If you have some preconceptions in this regard, do you rely on them? If not,

what else do you do?

4. Can you think of an instance in which you received advice electronically from someone credible

but then chose not to follow that advice? If so, please describe the situation and talk about some of

the reasons why you disagreed with or chose not to follow the advice it contained.

5. Can you think of an instance in which you received advice electronically from someone you did

not think was very credible, but chose to follow it anyway? If so, please describe the situation and

talk about some of the reasons why you chose to follow that advice.

32
6. Can you think of an instance in which you received advice electronically from a credible source

about a topic which you did not know very much? If so, please describe the situation and talk about

some of the reasons why did or did not chose to follow that advice. Probe: Does it make a

difference how much you know about the topic?

7. Can you think of an instance in which you received advice electronically from a credible source

about a topic which you were extremely interested in? If so, please describe the situation and talk

about some of the reasons why did or did not chose to follow that advice. Probe: Does it make a

difference how interested you are in the topic?

8. Now lets talk about each of the e-mails you brought with you today. Starting with the first,

please tell me why you received it, and what thoughts occurred to you upon first reading it? Did

your view of it change over time? Ultimately, what did you do with it and why? (probe on why).

When did you act on it, relative to when you received it?

33
APPENDIX II. SURVEY INSTRUCTIONS TO INFORMANTS

Dear ICBM Consultant, November 12, 1997

My research investigates how people perceive and use work-related knowledge that is transferred
electronically. This study investigates “electronic knowledge" in the form of e-mail. Enclosed is a
survey that consists of two parts: 1) Demographics, and 2) Two identical sheets of survey
questions, each pertaining to an e-mail that has been sent to you for work-related purposes.

Over the next few weeks, please print out and set aside the first two e-mails you receive that advise
you regarding a non-administrative action. The action can be anything, small or large (e.g. make a
decision, provide a response, give feedback, form a plan of attack, etc.). It doesn't matter whether
you have acted (or plan to act) on them or not. I am as interested in e-mails you have not acted on
as I am in those you have.

Once you have two of these, please complete the attached material as follows:

1. Complete the cover sheet asking about demographics and telecommunications

history.
2. Complete both sides of the e-mail survey for each of the two e-mails you have
set aside. Some of the questions are repetitive for statistical purposes – please go
though it fairly quickly.
3. Staple the e-mail surveys to the actual e-mails you printed out, and mail them
directly to me along with the demographic sheet in the enclosed stamped, self-
addressed envelope.

As an academic researcher, I hold the confidentiality of your responses in highest regard: I have
signed a ICBM nondisclosure agreement. You will be assigned a code number so that your answers
to my questions will be known only to me, and the key to this coding sheet will be destroyed as
soon as the data is entered into my computer (within 48 hours of receipt). All e-mails will also be
assigned numerical codes. You may also choose to white-out any specifics on them as you see fit.

I am confident that my findings will provide value to ICBM in its knowledge management
initiatives, and will keep you informed of the results of my analyses as I complete them. Please e-
mail me at [email protected] or telephone (617) 321-4321 if you have any questions. Your
participation is greatly appreciated!

34
APPENDIX III. QUANTITATIVE MEASURES AND INDICATORS

Source credibility (a=.8296): Adapted from Wu and Shaffer (1987)

Competence-based (2):

How knowledgeable is the person who wrote this message, on the topic of the message?
1 2 3 4 5 6 7
Not knowledgeable Knowledgeable

To what extent is the person who wrote this message an expert on the message topic?
1 2 3 4 5 6 7
Not expert Expert

Trustworthiness-based (2):

How trustworthy is the person who wrote this message, on the topic of the
message?
1 2 3 4 5 6 7
Not trustworthy Trustworthy

How reliable is the person who wrote this message, on the topic of the message?
1 2 3 4 5 6 7
Not reliable Reliable

Involvement (a=.6727): (Stamm and Dube, 1994).

How involved are you in the topic of this e-mail?

1 2 3 4 5 6 7
Not at all A great deal

How much has the issue discussed in this e-mail been on your mind lately?
1 2 3 4 5 6 7
Not at all A great deal

Expertise (a=.8049): (Stamm and Dube, 1994).

How informed are you on the subject matter of this issue? (Adapted from
Constant et. al., 1996)
1 2 3 4 5 6 7
Novice Expert

35
 To what extent are you an expert on the topic of this e-mail?
1 2 3 4 5 6 7
Not at all To a great extent

Argument Quality (a=.8444): (From Bailey and Pearson, 1983).

Please rate the content of this e-mail on the following scales, as in: “The information in this
e-mail is_______”

Complete 1 2 3 4 5 6 7 Incomplete

Consistent 1 2 3 4 5 6 7 Inconsistent

Accurate 1 2 3 4 5 6 7 Inaccurate

Usefulness (a=.8699): (From Bailey and Pearson, 1983).

Please rate the content of this e-mail on the following scales, as in: “The information in this
e-mail is_______”

Valuable 1 2 3 4 5 6 7 Worthless

Informative 1 2 3 4 5 6 7 Uninformative

Helpful 1 2 3 4 5 6 7 Harmful

Information adoption (a=.6520):

Have you acted on the content of this e-mail? (Y/N) If yes, How closely did you follow its
suggestions?
0 1 2 3 4 5 6 7 (0 = Not at all)
With major modifications To the letter

To what extent does the content of the e-mail motivate you to take action?
1 2 3 4 5 6 7
Not motivated Highly motivated

To what extent do you agree with the action suggested in the message?
1 2 3 4 5 6 7
Completely disagree Completely agree

36
References

Agarwal, R. and Prasad, J. (1997). The role of innovation characteristics and perceived

voluntariness in the acceptance of information technologies. Decision Sciences, 28, 557-

582.

Ajzen, I. And Fishbein, M. (1980). Understanding Attitudes and Predicting Social Behavior. New

Jersey: Prentice-Hall, Englewood Cliffs.

Andreoli, V., and Worschel, S. (1978). Effects of media, communicator, and message position on

attitude change. Public Opinion Quarterly, 42; 59-70.

Ausubel, D.P. (1963). The Psychology of Meaningful Learning. New York: Grune and Stratton.

Bailey, J.E. and Pearson, S.W. (1983). Development of a tool for measuring and analyzing

computer user satisfaction. Management Science, 29, 5, 530-545.

Baron, R.M. and Kenny, D.A. (1986). The moderator-mediator variable distinction in social-

psychological research: Conceptual, strategic and statistical considerations. Journal of

Personality and Social Psychology, 51:6, 1173-1182.

Billings, R. S. and Wroten, S. P. (1978). Use of path analysis in industrial/organizational

psychology: Criticisms and suggestions. Journal of Applied Psychology, 63, 677-688.

Carmines, E.G. and Zeller, R.A. (1979). Reliability and Validity Assessment, Sage University

series on quantitative applications in the social sciences, No. 07-017, Beverly Hill, CA:

Sage.

Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus

message cues in persuasion. Journal of Personality and Social Psychology, 39, 5, 752-766.

Chaiken, S. and A. H. Eagly (1976). Communication modality as a determinant of message

persuasiveness and message comprehensibility. Journal of Personality and Social

Psychology, 34, 605-614.

37
Chaiken, S. and Maheswaran, D. (1994). Heuristic processing can bias systematic processing:

Effects of source credibility, argument ambiguity, and task importance on attitude judgment.

Journal of Personality and Social Psychology, 66:3, 460-473.

Chin, W., Marcolin, B.L., and Newsted, P.R., (in submission). A partial least squares latent variable

modeling approach for measuring moderating effects: Results from a Monte Carlo

simulation study and voice-mail emotion/adoption study. Available on Dr. Chin’s web site,

Univ. of Houston, at http://disc-nt.cba.uh.edu/chin/icis96.pdf

Chin, W.W. (1998). The partial least squares approach to structural equation modeling. Chap. 10

in Modern Methods for Business Research, G.A. Marcoulides, Ed., Lawrence Erlbaum

Associates, London.

Cohen, W.M. and Levinthal, D.A. (1990). Absorptive capacity: A new perspective on learning and

innovation. Administrative Science Quarterly, Vol. 35, 128-152.

Constant, D., Sproull, L. and Kiesler, S. (1996). The kindness of strangers: The usefulness of

electronic weak ties for technical advice. Organization Science, 7:2, 119-135.

Davis, F.D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information

technology. MIS Quarterly, 13:3, 319-340.

Dennis, A.R. (1996). Information exchange and use in group decision making: You can lead a

group to information, but you can't make it think. MIS Quarterly, Dec., 433-457.

Dennis, A.R. and Kinney, S.T. (1998). Testing media richness theory in the new media: The effects

of cues, feedback, and task equivocality. Information Systems Research, 9:3, 256-274.

Duxbury, L.E. and Higgins, C. A. (1991). Gender differences in work-family conflict. Journal of

Applied Psychology, 76:1, 60-74.

Eagly, A.H. and Chaiken, S. (1993). The Psychology of Attitudes. Fort Worth, TX: Harcourt Brace

Jovanovich.

38
Ehrlich, K. and Cash, D. (1996). Credibility, trust and information filtering. Lotus Development

Corporation Publication.

Ellis, B.H. (1992). The effects of uncertainty and source credibility on attitudes about

organizational change. Management Communication Quarterly, 6:1, 34-57.

Feldman, J.M. and Lynch, J.G. (1988). Self generated validity and other effects of measurement on

belief, attitude, intention, and behavior. Journal of Applied Psychology, 73, 421-435.

Fishbein, M. and I. Azjen (1975). Belief, attitude, intention and behavior: An introduction to theory

and research. Reading, MA: Addison Wesley.

Fornell, C. and Bookstein, F. (1982). Two structural equation models: LISREL and PLS applied to

consumer exit-voice theory. Journal of Marketing Research, 19, 440-452.

Fulk, J. (1993). Social construction of communication technology. Academy of Management

Journal, 36, 921-950.

Gergen, K. (1982). Toward Transformation of Social Knowledge. New York: Spriger-Verlag.

Gupta, A. and Govindarajan, V. (1996). Determinants of knowledge outflows from and inflows into

foreign subsidiaries in multinational corporations. Working paper, Smith School of

Business, University of Maryland, College Park, MD.

Harvey, N., Harries, and Fischer, I. (2000). Using advice and assessing its quality. Organizational

Behaviors and Human Decision Processes, 81:2, 252-273.

Heesacker, M., Petty, R.E. and Cacioppo, J.T. (1983). Field dependence and attitude change:

Source credibility can alter persuasion by affecting message relevant thinking. Journal of

Personality, 51, 653-666.

Hovland, C.I. (1951). Changes in attitude through communication. Journal of Abnormal and Social

Psychology, 46; 424-437.

39
Hovland, C.I. , Janis, I.L., and Kelley, H.H. (1953). Communication and Persuasion, New Haven,

CT; Yale University Press.

Jaccard, J., Turrisi, R. and Wan, C.K. (1990). Interaction Effects in Multiple Regression, Series:

Quantitative applications in the social sciences, 72. Sage Publications, Inc.

nd
Jungermann, H. (1999). Advice giving and taking. In Proceedings of the 32 Hawaii International

Conference on Systems Sciences (HICS-32), Maui, HI. IEEE, on CD-ROM.

Karahanna, E., Straub, D., Chervany, N.L. (1999). Information technology adoption across time: A

cross-sectional comparison of pre-adoption and post-adoption beliefs. MIS Quarterly, 23:2,

183-213.

Kiesler, S. and Sproull, L. (1982). Managerial response to changing environments: Perspectives on

problem sensing from social cognition. Administrative Science Quarterly, 27, 548-570.

Lave, J. and Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge

University Press.

Lee, A.S. (1994). Electronic mail as a medium for rich communication: An empirical investigation

using hermeneutic interpretation. MIS Quarterly, June, 143-157.

Levitt, M. and Mahowald, R. (2001). Email usage and forecast analysis. IDC Report #W25335.

Mak, B., Schmitt, B.H., and Lyytinen, K. (1997). User participation in knowledge update of expert

systems. Information and Management, 32:2, 55-63.

Mak, B. and Lyytinen, K. (1997). A model to assess the behavioral impacts of consultative

knowledge based systems. Information Processing and Management, 33:4, 539-550.

Markus, M.L. (1994). Electronic mail as the medium of managerial choice. Organization Science,

5, 502-527.

McGinnies, E. and Ward, C.D. (1980) Better liked than right: Trustworthiness and expertise as

factors in credibility. Personality and Social Psychology Bulletin, 6:3, 467-472.

40
Mowen, J.C., Wiener, J.L., and Joag, S. (1987). An information integration analysis of how trust

and expertise combine to influence source credibility and persuasion. Advances in

Consumer Research, 14; p.564.

Newell, S. J. (1993). Developing a measurement scale and a theoretical model defining corporate

credibility, Unpublished PhD. dissertation, Florida State University College of Business,

Tallahassee.

Nonaka, I. (1994). A dynamic theory of knowledge creation. Organization Science, 5: 14-37.

Nunnally, J.C., (1967). Psychometric Theory, McGraw-Hill, New York, N.Y.

Pallak, S.R. (1983). Salience of a communicators' physical attractiveness and persuasion: A

heuristic versus systematic processing interpretation. Social Cognition, 2, 122-141.

Petty, R.E. and Cacioppo, J.T. (1986). Communication and Persuasion: Central and peripheral

routes to attitude change. New York: Springer-Verlag.

Petty, R.E., Cacioppo, J.T. and Goldman, R. (1981). Personal involvement as a determinant of

argument-based persuasion. Journal of Personality and Social Psychology, 41, 847-855.

Quinn, J.B., Anderson, P., and Finkelstein, S. (1996). Managing professional intellect: Making the

most of the best. Harvard Business Review, 74, 71-80.

Ratneshwar, S. and Chaiken, S. (1991). Comprehension's role in persuasion: The case of its

moderating effect on the persuasive impact of source cues. Journal of Consumer Research,

18, 52-62.

Rieh, S.Y. and Belkin, N.J. (1998). Understanding judgment of information quality and cognitive

authority in the WWW. Proceedings of the ASIS Annual Meeting, 35, 279-289.

Rice, R.E. (1992). Task analyzability, use of new media, and effectiveness: A multi-site exploration

of media richness. Organization Science, 3:4, 475-500.

41
Sharma, S., Durand, R.M. and Gur-Arie, O. (1981). Identification and analysis of moderator

variables. Journal of Marketing Research, 18, 291-300.

Short, J.A., Williams, E. and Christie, B. (1976). The Social Psychology of Telecommunications.

London: Wiley International.

Sniezek, J.A. and Buckley, B. (1995). Cueing and cognitive conflict in judge-advisor decision

making. Organizational Behavior and Human Decision Processes, 62, 159-174.

Spradley, J. 1979. The Ethnographic Interview. New York: Holt Reinhart.

Sproull, L. and Kiesler, S. (1986). Reducing social context cues: Electronic Mail in Organizational

Communication. Management Science, 32:11, 1492-1512.

Stamm, K. and Dube, R. (1994). The relationship of attitudinal components to trust in media.

Communication Research, 21:1; 105-123.

Venkatesh, V. and Davis, F. (2000). A theoretical extension of the technology acceptance model:

Four Longitudinal Field Studies. Management Science, 46:2; 186-204.

Walther, J.P. (1992). Interpersonal effects in computer-mediated interaction. Communication

Research, 19:1, 52-90.

Weiner, J.L. and Mowen, J.C. (1986). Source Credibility: On the independent effects of trust and

expertise. Advances in Consumer Research, 13; 306-310.

Worschel, S. Andreoli, V. and Eason, J. (1975). Is the medium the message? A study of the effects

of media, communicator, and message characteristics on attitude change. Journal of Applied

Social Psychology, 5, 157-172.

Wu, C. and Shaffer, D.R. (1987). Susceptibility to persuasive appeals as a function of source

credibility and prior experience with the attitude object. Journal of Personality and Social

Psychology, 52:4, 677-688.

42
 Zack, M. H. (1993). Interactivity and communication mode choice in ongoing management groups.

Information Systems Research, 4:3, 207-239.

Zack, M.H. and McKenney, J.L. (1995). Social context and interaction in ongoing computer-

supported management groups. Organization Science, 6:4, 394-422.

43

View publication stats