Content-Type: text/html

This paper was presented at the Association for Education in Journalism and
Mass Communication in San Antonio, Texas August 2005.
         If you have questions about this paper, please contact the author
directly. If you have questions about the archives, email
rakyat [ at ] eparker.org. For an explanation of the subject line, 
send email to
<a href="/cgi-bin/wa?LOGON=A3%3Dind0602A%26L%3DAEJMC%26E%3D8bit%26P%3D2461642%26B%3D--%26T%3Dtext%252Fplain%3B%2520charset%3Diso-8859-1" >[log in to unmask]</a> with just the four words, "get help info aejmc," in the
body (drop the "").

(Jan 2006)
Thank you.
Elliott Parker
====================================================================

Effects of Interactivity on Attitude Formation on Political Websites:
A Test of Mediation Effect of Perceived Interactivity
Given the development of networked communication technology in recent decades,
interactivity has emerged as a key concept in the discussion of new 
media (Bucy, 2004a, 2004b;
Chung & Zhao, 2004; McMillan, 2002b). Indeed, in defining what is new 
about new media,
researchers have used the terms "interactive media" and "new media" 
interchangeably (Vorderer,
2000), implicitly distinguishing two-way communication channels from 
one-way platforms that
lack opportunities for immediate feedback, user control, or audience 
participation (e.g., Morris &
Ogan, 1996; Rafaeli, 1988; Sims, 1997).
After a period of typologizing and classification, interactivity 
research has begun to focus
on the effects of interactivity on various outcome measures, 
including emotional and evaluative
responses to online news sites (Bucy, 2004b), advertising 
effectiveness (Chung & Zhao, 2004;
Liu & Shrum, 2002), and political attitudes (Sundar, Hesser, 
Kalyanaraman, & Brown, 1998;
Sundar, Kalyanaraman, & Brown, 2003). However, there have not been 
consistent results. Some
studies have found positive effects of interactivity, for instance, 
when measuring attitudes toward
advertisements (Cho & Leckenby, 1999; Chung & Zhao, 2004; McMillan & 
Hwang, 2002;
McMillan, Hwang, & Lee, 2003), while other studies have found mixed 
or even negative effects
of interactivity (Bucy, 2004b; Massey & Levy, 1999; McMillan, 2000; 
Sundar, 2000; Sundar et
al., 1998; Sundar et al., 2003).
These conflicting results are almost certainly due to inadequate, or 
competing,
conceptualizations of interactivity. Different researchers have 
conceptualized, operationalized,
and measured interactivity in myriad ways—all but guaranteeing 
inconclusive results about the
effects of interactivity (see Bucy, 2004a; Liu, 2003). Therefore, it 
is not surprising that there
Mediation Effect of Perceived Interactivity 2
have not been consistent results about the effects of interactivity. 
Given this history of
conceptual inconsistency, this study seeks, firstly, to redefine the 
concept of interactivity by
conceptually relocating the locus of interactivity further into 
user's perception and, secondly, to
test a hypothesized model of website interactivity by examining the 
effect of user's perception on
the relationship between interactive technology and various attitudes 
on political websites.
Conceptualizing Interactivity: Objective vs. Perceived Interactivity
As reviews of the literature attest, there has been little agreement 
among researchers
about the conceptualization of interactivity (Bucy, 2004a). In terms 
of the locus of interactivity,
some researchers assert that interactivity resides in the media 
system or in the process of
message exchange, while others argue that it primarily resides in 
user perception. These three
approaches are named the functional, contingency, and perceptual 
views, respectively in this
study.
In the functional view, interactivity is often used to describe 
technological features of the
new media (Vorderer, 2000) and defined as "a measure of a medium's 
potential ability to let the
user exert an influence on the content and/or form of the mediated 
communication" (Jensen,
1998, p. 201; see also Steuer, 1992). This view assumes that "the 
audience is not a passive
receiver of information, but rather an active co-creator" (McMillan, 
2002a). Thus, many
researchers have defined interactivity based on how many and what 
types of features are
available for users to fulfill interactive communication. In the 
context of the Internet, these
features might include bulletin boards, search engines, registration 
and online ordering forms
(McMillan, 1998); curiosity-arousal devices, games, user choice 
options, and surveys (Ha &
James, 1998); or, e-mail links, feedback forms, and chatrooms (Massey 
& Levy, 1999). In this
view, interactivity is thought to be objectively constant across 
people (Lee, 2000), while "the
Mediation Effect of Perceived Interactivity 3
degree to which these functions are used and the extent to which they 
actually serve the dialogue
or discourse function do not appear to be part of the concept's 
definition" (Sundar et al., 2003, p.
33).
Definitions in the contingency view (see Sundar et al., 2003) are 
based on the ideal model
of interpersonal communication. In this tradition, interactivity is 
conceptualized "as a process
involving users, media, and messages, with an emphasis on how 
messages relate to one another"
(Sundar et al., 2003, pp. 34-35); in other words, their degree of 
responsiveness. In this view
interactivity often refers to a process of reciprocal influence 
between two users or between a user
and a medium in a mediated communication. For example, Rafaeli (1988) 
defined interactivity
as "an expression of the extent that in a given series of 
communication exchanges, any third (or
later) transmission (or message) is related to the degree to which 
previous exchanges referred to
even earlier transmission" (p. 111). Ha and James (1998) also 
proposed that "[i]nteractivity
should be defined in terms of the extent to which the communicator 
and the audience respond to,
or are willing to facilitate, each other's communication needs" (p. 
461). More recently, Sundar et
al. (2003) suggested that "for interactivity to be perceived, not 
only should the interface possess
the functionality needed for mutual discourse, it should also ensure 
that the resulting messages
are interconnected. For interactivity to be realized, messages should 
be exchanged between
senders and receivers in a manner that leads to a thread of 
interdependent messages" (p. 34).
A third body of literature has addressed the concept of interactivity 
from a perceptual
approach. In this tradition, researchers have defined interactivity 
as a subjective state within the
user that may be evoked by technological features (e.g., Bucy & 
Newhagen, 1999; Lee, 2000;
McMillan, 2002b; Vorderer, 2000; Wu, 1999). Along these lines, Hwang 
and McMillan (2002)
suggest that "interactivity should not be measured by counting 
features, but rather by
Mediation Effect of Perceived Interactivity 4
investigating how users perceive and/or experience those features." 
 From this perspective, a
medium's technological features, while remaining objectively 
constant, are assumed to
subjectively vary according to individual differences among users 
(Heeter, 2000). Indeed, users
may perceive their communications to be interactive, and anticipate a 
response from the
recipient, even though the communication process promises no actual 
feedback (Newhagen,
Cordes, & Levy, 1995).
Although various definitions of interactivity in these three 
traditions provide a good
framework to identify different aspects of interactivity, "in many 
ways distinctions among these
traditions are arbitrary" (McMillan, 2002a, p. 166), especially the 
distinction between systembased
interactivity in the functional view and message-based interactivity 
in the contingency
view (Lee, 2000). Jensen (1998) argued that "in most specific cases, 
it would be difficult to
determine whether the 'interactivity' is directed toward a document 
or toward a platform" (p.
196). A more serious problem of conceptualizing interactivity in 
either functional or contingency
view is that it is almost impossible to operationalize and measure 
interactivity in a consistent
way due to the fact that a myriad number of interactive technologies 
continues to develop and
message exchange behaviors are different among people.
Accordingly, it seems better to focus on the psychological process 
occurred within user's
mind than to categorize and quantify interactive technologies and 
messages. In addition, the fact
that users may perceive a communication or a medium as interactive 
even when it lacks requisite
qualities of interactivity, which have been identified in the 
functional and contingency views,
suggests an important role for perceived interactivity (Bucy, 2004a). 
However, this should not be
interpreted as objective aspects of interactivity including media 
system and message exchange
have nothing to do with user perceptions. Instead, both technological 
features of a medium and
Mediation Effect of Perceived Interactivity 5
the way in which messages are exchanged influence user perceptions of 
that medium's
interactivity in disparate ways, depending on the various individual 
factors such as, for example,
gender, expertise, or experience of individual users. To summarize, 
the objective aspect of
interactivity affects the subjective (or perceptual) aspect of 
interactivity but the relationship
between these two differential aspects is not necessarily linear—an 
important proposition
overlooked in most studies.
Based on this consideration, this study suggests that interactivity 
be conceptually
understood and defined as residing in the relationship between 
available features and user
perceptions during the process of message exchange in the context of 
mediated communication.
Therefore, this study makes a distinction between objective and 
perceived interactivity in order
to conceptually define interactivity. Objective interactivity is 
conceptualized as the technological
potential of a medium to allow users to directly engage with the 
system or content. By contrast,
perceived interactivity is defined as the degree to which users 
actually experience a sense of
interactivity (regardless of the amount of technological features) 
during the communication
process.
As aforementioned, if a communication medium does not have enough potential of
interactivity, in other words if the medium does not provide 
sufficient objective interactivity
through technological features that may facilitate reciprocal and 
synchronous message exchange,
the user will not experience perceived interactivity that much. 
However, it should be also kept in
mind that increasing objective interactivity does not necessarily 
guarantee higher levels of
perceived interactivity (Bucy, 2004a). For example, it might be 
possible that one perceives
communication through asynchronous e-mail as more interactive than synchronous
communication through Instant Messenger (IM) even though objectively 
(technologically) the
Mediation Effect of Perceived Interactivity 6
opposite appears true. In this sense, it can be said that objective 
interactivity is a necessary
condition but not a sufficient condition for perceived interactivity.
A Mediation Model of Interactivity
A shortcoming of previous interactivity research has been a 
reluctance to empirically
investigate the relationship between the concept's objective and 
perceived aspects. It has been
naïvely assumed that there would be a positive relationship between 
objective and perceived
interactivity: the more a medium provides interactive features with 
various technological
applications, the more users would perceptually experience 
interactivity. In this view, the sheer
presence of certain interactive technologies is sufficient evidence 
of perceived interactivity
(Sundar et al., 2003).
Unfortunately, this groundless assumption about the relation between 
objective and
perceived interactivity has been adopted by most researchers, even 
when they define interactivity
as a perceptual variable. Sundar and his colleagues (2003), for 
example, proposed such a
hypothesis that "[p]articipants' ratings of a Web site's 
interactivity [i.e., a form of perceived
interactivity] will be a direct positive function of the degree of 
message contingency present in
the site [i.e., a form of contingent interactivity]" (p. 36), and 
then manipulated message
contingency on political Websites by providing different numbers of 
Webpage layers (i.e., a
form of functional interactivity). In Bucy's (2004b) study, 
interactivity on news Websites was
manipulated as to be either interactive or non-interactive by having 
subjects to vote in a poll,
view a slide show, and e-mail the news organization (i.e., a form of 
functional interactivity) or by
asking subjects only to read three stories on the Websites.
In both studies (Bucy, 2004b; Sundar et al., 2003), these 
manipulations of interactivity
were checked by a single-item measure of perceived interactivity that 
asked, for example, "how
Mediation Effect of Perceived Interactivity 7
interactive would you rate this Website" (Sundar et al., 2003, p. 
41). However, use of the singleitem
measure of perceived interactivity leaves much room for doubt in 
order to support the
correspondence between objective and perceived interactivity. In 
addition, in analyzing the
effects of interactivity on dependent variables based upon the 
assumption that website features
would directly influence outcomes independent of user perceptions, 
they did not utilize the
perceptual measures in their statistical analyses, downplaying the 
distinctive role of perceived
interactivity.
The real problem in previous interactivity research, which can be 
directly attributed to the
misunderstanding of the relationship between objective and perceived 
interactivity, is that the
very mechanism of mediated communication through which the medium's 
objective aspects such
as interactive technology result in certain outcomes has not been 
explained in a logical and
theoretical way. This theoretical inability to explain how 
interactive technologies affect certain
outcome variables becomes more conspicuous especially when 
researchers try to explain the socalled
threshold effects (Bucy, 2004a) or interactivity paradox (Bucy, 
2004b) found in recent
interactivity studies.
For example, in the previous studies that examined the effects of 
interactivity on various
outcomes including memory (Massey & Levy, 1999), emotions (Bucy, 
2004b), and political
attitudes (Sundar et al., 1998; Sundar et al., 2003), it was found 
that, contrary to the optimistic
expectation that positive outcomes would be the direct function of 
interactivity, increasing
interactivity did not always result in positive outcomes. Sundar and 
his colleagues (2003) found
that moderate interactivity on a political website enhanced the 
political candidate's appeal,
character, and the level of voter agreement with the candidate's 
position on various policy issues,
but high interactivity seemed to detract from those positive attitude 
formation effects. They
Mediation Effect of Perceived Interactivity 8
simply inferred that "interactivity at higher levels may impose 
greater navigational demands on
users, which serve to counteract its positive effects on users' 
impression of the site" (p. 49)
without suggesting any theoretical mechanism in order to explain how 
cognitive demands cancel
out positive effects of interactivity.
The lack of theoretical explanation about such an unexpected result 
(i.e., threshold effects
of interactivity) may be due to the fact that interactivity 
researchers have examined only the
effects of objective interactivity, while not paying enough attention 
to the role of perceived
interactivity that may significantly affect the direct relationship 
between objective interactivity
and outcome variables. That is, the measures of perceived 
interactivity have been seldom utilized
for the statistical analyses in previous interactivity research.
However, as Norman (1998) has usefully observed, the gap between the 
set of possible
actions, or affordances, that a medium makes possible and what users 
actually perceive to be
available (and know how to use) can be sizeable. According to Norman, 
an affordance "is not a
property, it is a relationship" (p. 123) between the medium and the 
user—and the same medium
may have different affordances for different users. Here, it is 
useful to distinguish between real
and perceived affordances. Perceived affordances inform the user what 
actions can be performed
and, to some extent, how to perform them. If a technology's actual 
properties or real affordances
are not detected and understood, then, like a hidden program icon in 
an applications folder, they
will have little value. Not recognizing real affordances because they 
are hidden or obscured, or
wrongly perceiving a false affordance as real, may lead to mistakes. 
Perceived affordances, then,
distinguish usable interfaces from those that are mysterious and 
unfathomable (Norman, 1998).
Overall, interface designers contend, "what makes a system successful 
is how well the design
model is communicated to the user" (Mohnkern, 1997 n.p.).
Mediation Effect of Perceived Interactivity 9
The point is that, since perceived interactivity is not the direct 
function of objective
interactivity, it must be separated from objective interactivity both 
conceptually and
operationally. Therefore, perceived interactivity must be measured as 
another variable
independent from objective interactivity and then included in the 
analysis with objective
interactivity in order to better explain the effects of 
interactivity. If such is done, for instance, a
possible explanation about the case of the people who perceive e-mail 
as more interactive than
IM in the aforementioned example is that objective interactivity may 
overload the amount they
can accept, expect, or need so that the residual objective 
interactivity afforded by the IM
application may be processed negatively and can not be internalized 
into perceived interactivity.
Similarly, if perceived interactivity had been measured in a more 
stringent way by using a welldeveloped
scale instead of a single-item question and put into the statistical 
analyses, some of the
subjects in previous studies might have been found to experience more 
perceived interactivity in
the low objective interactivity condition than in more objectively 
interactive conditions, which
may hint why increasing objectivity does not necessarily guarantee 
more positive outcomes.
This, in turn, suggests that perceived interactivity might mediate 
the relationship between
objective interactivity and outcomes.
Considering the effects of interactivity on political attitudes, it 
can be hypothesized that
the formation of positive attitudes in mediated communication may be 
the indirect function of
objective interactivity through perceived interactivity rather than 
the direct function of objective
website features. In other words, perception of interactivity 
experienced on a political website
will mediate the direct effects of interactive features embedded in 
the website on political
attitudes formed on the website.
Mediation Effect of Perceived Interactivity 10
Summarizing the arguments presented thus far, objective interactivity 
does not directly
result in corresponding outcomes. Instead, objective interactivity is 
first translated into perceived
interactivity. However, perceived interactivity is not the simple 
direct function of objective
interactivity because there are many third variables that may 
moderate the process of translation
from objective interactivity into perceived interactivity—the very 
mechanism overlooked by
most of interactivity researchers, especially, who fails to explain 
the so-called threshold effects
of interactivity. To better understand the effects of interactivity 
in the (computer) mediated
communication, the role of perceived interactivity as a potential 
mediator between objective
interactivity and outcome variables should be more fully appreciated.
Method
Experimental Design and Participants
To test the mediation model of interactivity, the present study 
employed a two-betweengroup
(low vs. high objective interactivity) experiment. The participants 
(N = 78) were
undergraduate students enrolled in three communications classes at a 
large mid-western
university who participated in this study for extra credit. They were 
randomly assigned and
exposed to either low (n = 39) or high (n = 39) objective 
interactivity condition.
Stimulus Material and Experimental Conditions
In discussing the results of their manipulation check, Sundar et al. 
(2003) admitted that
interactivity cannot be easily operationalized by the number of 
hyperlinks present on a website
and proposed that "[t]he actions initiated by the website (by guiding 
the user through a series of
hierarchical hyperlinks) are as important as the organization of 
message content in determining
the degree of interactivity of a website, especially as perceived by 
a user" (p. 48). Considering
Mediation Effect of Perceived Interactivity 11
this suggestion and adopting Bucy's (2004b) activity-oriented 
manipulation of interactivity, two
different versions of a stimulus site were adapted (with permission) 
from the actual campaign
site of incumbent Governor John E. Baldacci of Maine. The site of an 
out-of-state office holder
was selected to minimize familiarity effects and avoid activating 
partisan sentiments associated
with a more familiar candidate.
The two stimulus sites were designed to have identical content but 
were structured
differently via hyperlinks to individual pages. In addition, the 
stimulus sites were designed to
maximize the variance between the low and high interactivity 
conditions by offering various
interactive features and applications only to the high interactivity 
condition. However, to control
for ideological orientation, the issue content was carefully chosen 
to represent a moderate
viewpoint, so as not to be seen as either too liberal or too conservative.
As a result, in the low interactivity condition, the site featured 
just one layer of content,
without any hyperlinks or applications (i.e., a one-layer, 
reading-only condition). A brief
biography of the candidate was presented at the top of the page and, 
along with a few
photographs, was followed by his positions on various policy issues, 
including the economy,
health care, education, the environment, and defense/veterans issues 
(in that order).
On the other hand, in the high interactivity condition, the site 
featured hyperlinks that
connected three layers of web pages and provided users with an e-mail 
link to the candidate and
an interactive budget calculator, plus the online poll application 
(i.e., a three-layer browsing,
polling, e-mailing, and budget calculating condition). The main page 
on the first layer showed
the same content and format as did the low condition but also 
included an e-mail link to the
candidate and a link to a budget calculator. Five pages on the second 
layer were hyperlinked for
a total of 16 sub-pages that provided the most specific agendas 
related to the main policy issues:
Mediation Effect of Perceived Interactivity 12
four for the economy; three for health care; three for education; 
three for environment; and three
for defense/veterans.
Variables
Objective Interactivity. For the purpose of statistical analysis, 
dummy coding was used so
that the low objective interactivity condition was coded as "0" and 
the high objective
interactivity condition as "1."
Perceived Interactivity. Participants' perceived interactivity was 
measured by utilizing a
scale developed by Liu (2003), which includes 15 items that tap three 
subdimensions of
perceived interactivity: Two-Way Communication, Active Control, and 
Synchronicity. However,
for the purpose of statistical analysis for the mediation test, all 
the responses on the15 items were
averaged and treated as a single-dimension variable. Each item was 
measured using a sevenpoint
Likert-type scale ranging from 1 = "strongly disagree" to 7 = 
"strongly agree." The overall
reliability of the scale was high (_ = .97).
Attitudes toward the Website. To measure participants' attitude 
toward the political
website after exposure, a scale developed by Chen, Clifford, and 
Wells (2002) was utilized.
Through a series of scale-development studies (Chen et al., 2002; 
Chen & Wells, 1999), the scale
for measuring attitudes toward websites has been found reliable and 
robust across various kinds
of websites and respondent groups. The scale is composed of a single 
dimension that includes six
items, such as "I would like to visit the website again in the 
future" and "I feel surfing the
website is a good way for me to spend my time." These items were also 
measured on the sevenpoint
Likert-type scale, ranging from 1 = "strongly disagree" to 7 = 
"strongly agree." Scale
reliability was acceptable (_ = .88).
Mediation Effect of Perceived Interactivity 13
Attitudes toward the Politician. To measure users' attitudes toward 
the candidate for
whom the website was designed, a set of questions borrowed from 
Sundar et al's (1998) study
was utilized. The scale is composed of a total of 16 items depicting 
various aspects of the
politician such as qualities as a politician, the politician's 
attention to constituent concerns, and
charisma. Each item was measured on the seven-point Likert-type scale 
ranging from 1 =
"strongly disagree" to 7 = "strongly agree." The reliability of this 
scale was also high (_ = .92).
Attitudes toward Policies. To measure participants' attitudes toward 
the candidate's issue
positions, including opinions about the economy, health care, 
education, the environment, and
defense/veterans, a set of items asked users the extent to which they 
agreed on each of the
proposed policies. Each item was measured on the seven-point 
Likert-type scale ranging from 1
= "strongly disagree" to 7 = "strongly agree" and then all five items 
were averaged into a single
score to represent the general attitude toward the policies proposed 
by the politician on the
website. Scale reliability was acceptable, too (_ = .85).
Procedure
The experiment was administered to small groups of participants in a 
communication
research laboratory equipped with five lap-top computers connected to 
the Internet. Upon arrival,
participants were asked to complete the informed consent process, 
after which they were given a
laptop computer and completed an online pre-experiment questionnaire 
designed to obtain basic
demographic information. Then, participants were randomly assigned to 
one of two experimental
conditions: a low- or high-interactivity website. They were asked to 
browse the site and engage
in a series of designated activities for 15 minutes. To prevent 
contamination, participants who sat
on the same side of the table were assigned to different experimental 
conditions and were asked
not to talk to each other during the experiment. After exposure to 
the stimulus site, participants'
Mediation Effect of Perceived Interactivity 14
perceived interactivity and attitudes were measured via an online 
post-experiment questionnaire.
Participants were then debriefed, thanked for their participation, 
and dismissed.
Statistical Analysis
Mediation implies a causal relationship in which an independent 
variable causes a
mediator which then causes a dependent variable. Figure 1 illustrates 
the mediation model of
interactivity where perceived interactivity (M) mediates the 
relationship between objective
interactivity (X) and attitudes (Y).
To test the mediation model of interactivity proposed in Figure 1, 
this study conducted a
series of regression analyses, following the steps suggested by many 
researchers (Baron &
Kenny, 1986; Judd & Kenny, 1981; MacKinnon & Dwyer, 1993). In each 
step, the following
regression analyses were conducted and significance of the 
coefficients was examined:
Step 1: Simple regression with X predicting Y to test for path c 
alone, Y = C1 + Bc1X + e1
Step 2: Simple regression with X predicting M to test for path a 
alone, M = C2 + Ba2X + e2
Step 3: Simple regression with M predicting Y to test for path b 
alone, Y = C3 + Bb3M + e3
Step 4: Multiple regression with X and M predicting Y to test for 
indirect path b through a,
Y = C4 + Bc4 + Bb4 + e
The purpose of the simple regression analyses in Step 1 through 3 is 
to examine whether
there is a zero-order relationship between two variables. In other 
words, mediation can be said to
occur when (1) the independent variable X significantly affects the 
dependent variable Y in Step
1 regression, (2) the independent variable X significantly affects 
the mediator M in Step 2
regression, and (3) the mediator M has a significant effect on the 
dependent variable Y in Step 3
Mediation Effect of Perceived Interactivity 15
regression. After all these conditions are met, some form of 
mediation is supported (4) if the
direct effect of mediator M (i.e., Bb4) remains significant after 
controlling for the independent
variable X in Step 4 regression. Note that, among these four steps 
suggested, Step 1 and 3 are not
required because the significances of paths b and c are implied if 
Step 2 and 4 are met. So, the
essential steps in analyzing mediation effect are Step 2 and Step 4.
In Step 4, if the direct effect of independent variable X on the 
dependent variable Y (i.e.,
Bc4) is no longer significant when the mediator M is controlled, the 
finding supports full
mediation. If both direct effect of independent variable X (i.e., 
Bc4) and direct effect of mediator
M (i.e., Bb4) are significant, the finding indicates partial mediation.
To calculate the indirect effect of independent variable X through 
the mediator M, Judd
and Kenny's (1981) differences of coefficients approach was adopted. 
This approach involves
subtracting the partial regression coefficient obtained in Step 4 
(Bc4) from the simple regression
coefficient in Step 1 (Bc1):
Bindirect = Bc1 – Bc4
And, the significance of the indirect effect was tested by conducting 
the Goodman (I) version of
the Sobel test which is recommended by Baron and Kenny (1986). This 
test enables us to have a
z-value based on the following equation:
Here, s = standard error of the coefficient
The reported p values are drawn from the unit normal distribution 
under the assumption of a twotailed
z-test of the hypothesis that the mediated effect equals to zero in 
the population. Therefore,
z = ±1.96 is the critical value of the test at p = .05.
Mediation Effect of Perceived Interactivity 16
Results
Attitude toward Website
Step 1. Simple regression was conducted to investigate how well 
objective interactivity
(X) predict attitude toward the political website (Y). The results 
were statistically significant F(1,
76) = 20.53, p < .001. The identified equation to understand this 
relationship was attitude toward
website (Y) = 3.27 + 1.21*(X: objective interactivity). The adjusted 
R2 value was .20. This
indicates that 20% of the variance in attitude toward website was 
explained by objective
interactivity.
Step 2. Simple regression was conducted to examine the relationship 
between objective
interactivity (X) and perceived interactivity (M). The results were 
statistically significant F(1,
76) = 136.94, p < .001. The identified equation to understand this 
relationship was perceived
interactivity (M) = 3.06 + 2.58*(X: objective interactivity). The 
adjusted R2 value was .64,
indicating that 64% of the variance in perceived interactivity was 
explained by objective
interactivity.
Step 3. The results of simple regression with perceived interactivity 
(M) predicting
attitude toward website (Y) were statistically significant F(1, 76) = 
69.46, p < .001. The
identified equation to understand this relationship was attitude 
toward website (Y) = 1.43 +
.56*(M: perceived interactivity). The adjusted R2 value was .47, 
which indicates that 47% of the
variance in attitude toward website was explained by perceived interactivity.
Step 4. Multiple regression was conducted to examine the linear combination of
independent variable (X: objective interactivity) and moderator (M: 
perceived interactivity) for
predicting the dependent variable (Y: attitude toward website). The 
combination of objective
interactivity (X) and perceived interactivity (M) significantly 
predicted attitude toward website
Mediation Effect of Perceived Interactivity 17
(Y), F(2, 75) = 37.77, p < .001. The identified equation was attitude 
toward website (Y) = 1.03 -
.68*(X: objective interactivity) + .73*(M: perceived interactivity). 
However, the beta weights,
presented in Table 1, indicate that only perceived interactivity (M) 
significantly contributes to
predicting attitudes toward website (Y), while objective 
interactivity (X) does not. The adjusted
R2 value was .49, indicating that 49% of the variance in attitude 
toward website was explained by
the combination of objective interactivity and perceived interactivity.
To summarize the results from Step 1 to Step 4 (see Table 1), (1) 
objective interactivity
(X) significantly affected attitude toward website (Y) in the absence 
of perceived interactivity
(M), (2) objective interactivity (X) significantly affected perceived 
interactivity (M), and (3)
perceived interactivity (M) significantly affected attitude toward 
website (Y). Therefore, all the
basic conditions for mediation model were met. And, (4) the direct 
effect of perceived
interactivity (M) remained significant after controlling for 
objective interactivity (X), while the
direct effect of objective interactivity (X) on attitude toward 
website (Y) was no longer
significant when perceived interactivity (M) was controlled. The 
coefficient of the indirect effect
of objective interactivity through perceived interactivity on 
attitude toward website was
statistically significant, Bindirect = 1.886, z = 5.726, p < .001. 
Therefore, in the case of attitude
toward website as the dependent variable, full mediation was supported.
Attitude toward Politician
Step 1. Simple regression was conducted to examine the relationship 
between objective
interactivity (X) and attitude toward politician (Y). The results 
were statistically significant F(1,
76) = 4.97, p < .05. The identified equation to understand this 
relationship was attitude toward
politician (Y) = 4.58 + .43*(X: objective interactivity). The 
adjusted R2 value was .05, indicating
that 5% of the variance in attitude toward politician was explained 
by objective interactivity.
Mediation Effect of Perceived Interactivity 18
Step 2. The results of simple regression with objective interactivity 
(X) predicting
perceived interactivity (M) were the same with those of Step 2 for 
attitude toward website
described above.
Step 3. Simple regression was conducted to investigate how well 
perceived interactivity
(M) predict attitude toward politician (Y). The results were 
statistically significant F(1, 76) =
24.14, p < .001. The identified equation to understand this 
relationship was attitude toward
politician (Y) = 3.65 + .26*(M: perceived interactivity). The 
adjusted R2 value was .23. This
indicates that 23% of the variance in attitude toward politician was 
explained by perceived
interactivity.
Step 4. Multiple regression was conducted to examine the linear combination of
independent variable (X: objective interactivity) and moderator (M: 
perceived interactivity) for
predicting the dependent variable (Y: attitude toward politician). 
The combination of objective
interactivity (X) and perceived interactivity (M) significantly 
predicted attitude toward website
(Y), F(2, 75) = 16.13, p < .001. The identified equation was attitude 
toward politician (Y) = 3.24
- .70*(X: objective interactivity) + .44*(M: perceived 
interactivity). The beta weights, presented
in Table 2, indicate that both objective (X) and perceived 
interactivity (M) significantly
contribute to predicting attitudes toward politician (Y). The 
adjusted R2 value was .28, indicating
that 28% of the variance in attitude toward politician was explained 
by the combination of
objective and perceived interactivity.
To summarize the results (see Table 2), (1) objective interactivity 
(X) significantly
affected attitude toward politician (Y) in the absence of perceived 
interactivity (M), (2) objective
interactivity (X) significantly affected perceived interactivity (M), 
and (3) perceived interactivity
(M) significantly affected attitude toward politician (Y). Therefore, 
all the basic conditions for
Mediation Effect of Perceived Interactivity 19
mediation model were met. And, (4) the direct effect of perceived 
interactivity (M) remained
significant after controlling for objective interactivity (X). The 
coefficient of the indirect effect
of objective interactivity through perceived interactivity on 
attitude toward politician was
statistically significant, Bindirect = 1.251, z = 4.64, p < .001. 
However, the direct effect of objective
interactivity (X) on attitude toward website (Y) was also significant 
even though perceived
interactivity (M) was controlled. Therefore, in the case of attitude 
toward politician as the
dependent variable, partial mediation was supported.
Attitude toward Policies
Step 1. Simple regression was conducted to investigate how well 
objective interactivity
(X) predict attitude toward policies (Y). The results were not 
statistically significant F(1, 76) =
1.95, ns.
Step 2. The results of simple regression with objective interactivity 
(X) predicting
perceived interactivity (M) were the same with those of Step 2 for 
attitude toward website
described above.
Step 3. The results of simple regression with perceived interactivity 
(M) predicting
attitude toward policies (Y) were not statistically significant F(1, 
76) = .15, ns.
Step 4. Multiple regression was conducted to examine the linear combination of
independent variable (X: objective interactivity) and moderator (M: 
perceived interactivity) for
predicting the dependent variable (Y: attitude toward policies). The 
combination of objective
interactivity (X) and perceived interactivity (M) was not 
statistically significant in predicting
attitude toward policies (Y), F(2, 75) = 1.73, ns.
To summarize the results, although the most important condition among 
three basic
conditions for mediation was met in Step 2, the direct effect of 
perceived interactivity (M) was
Mediation Effect of Perceived Interactivity 20
not significant after controlling for objective interactivity (X). In 
addition, the direct effect of
objective interactivity (X) on attitude toward website (Y) was not 
significant, either, when
perceived interactivity (M) was controlled. Therefore, in the case of 
attitude toward policies as
the dependent variable, mediation was not supported.
Discussion
This study proposed a new approach to interactivity research based on 
the idea that
perceived interactivity might mediate the nuanced relationship 
between website structure
(objective interactivity) and various outcomes such as political 
attitudes. The reason for dividing
the concept of interactivity into objective and perceived 
interactivity is that objective aspects of
interactivity or technological aspects of interactivity do not always 
correspond to the perceptual
or subject aspects of interactivity. And, more importantly, if this 
is true, the direct relationship
between objective interactivity and an outcome is a spurious one that 
previous interactivity
research has in vain tried to explain. In this situation, it is hoped 
that this study may provides a
new start point for researchers to study the effects of interactivity 
by examining the mediating
role of perceived interactivity.
The results supported the mediation model for predicting political 
attitudes formed on a
political website. The influence of the subjective variable 
(perceived interactivity) mediated the
effect of objective interactivity on attitudes toward the political 
website as well as attitude toward
the politician for whom the website was designed. However, in the 
case of attitude toward
policies proposed on the political website, the mediation effect of 
perceived interactivity was not
found. Rather, attitude toward policies was found to be affected 
neither by objective interactivity
nor by perceived interactivity. This suggests that formation of 
attitude toward policies may not
be formed through emotional or perceptual experiences. Indeed, 
considering that the stimulus
Mediation Effect of Perceived Interactivity 21
website and the politician were new to the subjects but all the 
policy issues discussed in the
website were very general topics which might be familiar to the 
subjects, attitudes toward the
website and the politician were likely to be newly formed mainly 
through emotional and
perceptual experiences, whereas attitude toward policies were likely 
to be simply expressed
based on the subjects' existing opinions while less affected by 
interactivity. We will discuss
more about this later in relation to another third variable, interest 
in politics.
Interestingly, when the mediation model was supported, objective 
interactivity was found
to have positive impact on attitudes. However, when it was examined 
with perceived
interactivity together, objective interactivity was found to 
negatively affect attitudes. These
results support that interactive presentations do not uniformly 
elicit positive evaluations and that
too many interactive functions embedded in a website, rather than 
having an salutary affect on
outcomes, may instead provoke negative attitudes—especially when they 
are offered to users
who do not internalize their online experience with a strong sense of 
perceived interactivity.
Considering that the participants in this study were college 
students, who are generally more
practiced online than the general public to which political websites 
are targeted, the significance
of design considerations for political websites becomes evident.
However, still unknown is what makes some users perceive their 
mediated experience to
be more interactive than other users, even when the same interface 
features are presented. That
is, what factors determine each individual's threshold point up to 
which objective interactivity is
well translated into perceived interactivity but after that point 
increasing objective interactivity
negatively affects certain outcomes. A potential variable, for 
example, is Internet self-efficacy.
According to Bandura's (1997) social cognitive theory, an individual 
possesses a self-system that
enables him to exercise a measure of control over his thoughts, 
feelings, motivations, and
Mediation Effect of Perceived Interactivity 22
actions. The self-system provides reference mechanisms and a set of 
sub-functions for
perceiving, regulating, and evaluation behavior, which results from 
the interplay between the
system and environmental sources of influence. In this self-referent 
process, self-efficacy plays a
key role in mediating subsequent behaviors (Bandura, 1994a, 1994b, 
1997, 2001).
The importance of self-efficacy in explaining computer use has been 
well documented.
Hill, Smith, and Mann (1987) found that computer self-efficacy 
affected whether individuals
chose to use computers or not. Similarly, individuals with high 
self-efficacy used computers
more than those with low self-efficacy (Compeau & Higgins, 1995). 
Hill and Hannafin (1997)
studies the influence of perceived self-efficacy on strategies 
employed in an Internet search.
Participants' self-efficacy beliefs in using computer technologies 
and information searching
systems affected both the number and types of strategies they 
employed. Those with high selfefficacy
explored the systems more vigorously while those with low 
self-efficacy retreated or
concentrated on simply locating information. In previous research on 
uses and gratifications of
the Internet, it was also that Internet self-efficacy was positively 
related to expected gratification
(Ebersole, 1999; LaRose, Mastro, & Eastin, 2001). These results 
suggest that Internet selfefficacy
may be a key factor that moderates the relationship between objective 
interactivity and
perception about it, which results in the threshold effects. That is, 
it can be hypothesized that
people with high Internet self-efficacy will more likely to perceive 
interactivity than those with
low self-efficacy.
Another potential variable that may also moderate the relationship 
between objective and
perceived interactivity, causing individual differences in the 
location of threshold points, is
interest in the subject matter. In the context of political 
communication, for instance, it was
found that for politically apathetic people greater (objective) 
interactivity tended to result in
Mediation Effect of Perceived Interactivity 23
higher levels of affinity to the candidate but not for the 
politically savvy people (Sundar et al.,
1998). Interestingly, politically savvy participants in the high 
(objective) interactivity condition
tended to have most negative perceptions of the candidate, while the 
medium condition enhanced
their affinity to the candidate.
To interpret these results based upon the mediation model suggested 
in this study,
politically savvy people might not internalize high objective 
interactivity into perceived one
because they were likely to spend most of their cognitive capacity in 
careful scrutiny of the
issue-based contents. Therefore, their low levels of perceived 
interactivity in high objective
interactivity condition might lead to negative attitudes toward the 
politician. On the contrary,
politically apathetic people might experience higher levels of 
perceived interactivity as the level
of objective interactivity increased, because they were more likely 
to be susceptible to the
inherent appeal of interactive features of the website while not 
paying enough attention to the
issue-based contents. Therefore, their high levels of perceived 
interactivity increased by the
direct function of objective interactivity might positively affect 
the impression about the
politician.
Under similar reasoning, the finding in this study that neither 
objective interactivity nor
perceived interactivity affected formation of attitude toward 
policies can be explained. Since the
policy issues presented in the stimulus website included very broad 
and general topics such as
economy, health care, education, environment, and defense, when the 
subjects express their
positions about the policies suggested on the stimulus website they 
might be based on their own
existing opinions regardless of the level of interactivity rather 
than based on their perceptual
experiences of interactivity.
Mediation Effect of Perceived Interactivity 24
Although this study proposed here only two potential variables 
(Internet self-efficacy and
interest in subject matter) that may affect the effects of 
interactivity, future research should
consider more possible factors to better understand the paradoxical 
relationship between
objective and subjective aspects of interactivity as an important 
source of persuasive influence.
As well, it is necessary to validate the mediation model by utilizing 
more stringent and advanced
statistical methods such as path analysis and structural equation 
modeling. However, most
importantly, the role of perceived interactivity as a mediator 
between objective interactivity and
outcome variables should be fully grasped first before including 
other third variables.
Mediation Effect of Perceived Interactivity 25
References
Bandura, A. (1994a). Self-efficacy. In V. S. Ramachaudran (Ed.), 
Encyclopedia of human
behavior (Vol. 4, pp. 71-81). New York: Academic Press.
Bandura, A. (1994b). Social cognitive theory of mass communication. 
In J. Bryant & D.
Zillmann (Eds.), Media effects: Advances in theory and research (pp. 
61-90). Hillsdale,
NJ: Lawrence Erlbaum Associates.
Bandura, A. (1997). Self-efficacy: The exercise of control.New York: Freeman.
Bandura, A. (2001). Social cognitive theory of mass communication. 
Mediapsychology, 3, 265-
299.
Baron, R. M., & 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.
Bucy, E. P. (2004a). Interactivity in society: Locating an elusive 
concept. Information Society,
20(5), 373-383.
Bucy, E. P. (2004b). The interactivity paradox: Closer to the news 
but confused. In E. P. Bucy &
J. E. Newhagen (Eds.), Media access (pp. 47-72). Mahwah, NJ: Lawrence Erlbaum
Associates.
Bucy, E. P., & Newhagen, J. E. (1999). The emotional appropriateness 
heuristic: Processing
televised presidential reactions to the news. Journal of 
Communication, 49(4), 59-79.
Chen, Q., Clifford, S. J., & Wells, W. D. (2002). Attitude towrad the 
Site II: New information.
Journal of Advertising Research, 42(2), 33-45.
Chen, Q., & Wells, W. D. (1999). Attitude toward the site. Journal of 
Advertising Research,
39(5), 27-37.
Cho, C.-H., & Leckenby, J. D. (1999). Interactivity as a measure of 
advertising effectiveness.
Paper presented at the American Academy of Advertising, Gainesville, 
FL: University of
Florida.
Chung, H., & Zhao, X. (2004). Effects of perceived interactivity on 
Web site preference and
memory: Role of personal motivation. JCMC, 10(1),
http://jcmc.indiana.edu/vol10/issue11/chung.html.
Mediation Effect of Perceived Interactivity 26
Compeau, D. R., & Higgins, C. A. (1995). Computer self-efficacy: 
Development of a measure
and initial test. MIS Quarterly, June.
Ebersole, S. E. (1999). Adolescents' use of the World Wide Web in ten 
public schools: A uses and
gratifications approach. Unpublished Ph. D. dissertation, Regent University.
Ha, L., & James, E. L. (1998). Interactivity reexamined: A baseline 
analysis of early business
Web sites. Journal of Broadcasting & Electronic Media, 42(4), 457-474.
Heeter, C. (2000). Interactivity in the context of designed 
experiences. Journal of Interactive
Advertising, 1(1), http://jiad.org/vol1/no1/heeter/.
Hill, J. R., & Hannafin, M. J. (1997). Cognitive strategies and 
learning from the World Wide
Web. Educational Technology Research and Development, 45(4), 37-64.
Hill, T., Smith, N. D., & Mann, M. F. (1987). Role of efficacy 
expectations in predicting the
decisions to use advanced technologies: The case of computers. 
Journal of Applied
Psychology, 72(2), 307-313.
Hwang, J.-S., & McMillan, S. J. (2002). The role of interactivity and 
involvement in attitude
toward the Web site. Paper presented at the 2002 Conference of the 
American Academy
of Advertising, Auburn, AL.
Jensen, J. F. (1998). Interactivity: Tracking a new concept in media 
and communication studies.
Nordicom Review, 1, 185-204.
Judd, C. M., & Kenny, D. A. (1981). Process analysis: Estimating 
mediation in treatment
evaluations. Evaluation Review, 5(5), 602-619.
LaRose, R., Mastro, D., & Eastin, M. S. (2001). Understanding 
Internet usage: A socialcognitive
approach to uses and gratifications. Social Science Computer Review, 19(4),
395-413.
Lee, J.-S. (2000, August). Interactivity: A new approach. Paper 
presented at the annual meeting
of the Association for Education in Journalism and Mass Communication (AEJMC),
Phoenix, AZ.
Liu, Y. (2003). Developing a scale to measure the interactivity of 
websites. Journal of
Advertising Research, 43(2), 207-216.
Liu, Y., & Shrum, L. J. (2002). What is interactivity and is it 
always such a good thing?
Implications of definition, person, and situation for the influence 
of interactivity on
advertising effectiveness. Journal of Advertising, 31(4), 53-64.
Mediation Effect of Perceived Interactivity 27
MacKinnon, D. P., & Dwyer, J. H. (1993). Estimating mediated effects 
in prevention studies.
Evaluation Review, 17(2), 144-158.
Massey, B. L., & Levy, M. R. (1999). Interactivity, online 
journalism, and English-language
Web newspapers in Asia. Journalism and Mass Communication Quarterly, 
76(1), 138-
151.
McMillan, S. J. (1998). Who pays for content? Funding in interactive 
media. Journal of
Computer Mediated Communication, 4(1),
http://www.ascusc.org/jcmc/vol4/issue1/mcmillan.html.
McMillan, S. J. (2000). Interactivity is in the eye of the beholder: 
Function, perception,
involvement, and attitude toward the web site. Paper presented at the 
conference of the
American Academy of Advertising, East Lansing, MI: Michigan State University.
McMillan, S. J. (2002a). Exploring models of interactivity from 
multiple research traditions:
Users, documents, and systems. In L. A. Lievrouw & S. Livingstone 
(Eds.), Handbook of
new media: Social shaping and consequences of ICTs (pp. 163-182). London: SAGE
Publications.
McMillan, S. J. (2002b). A four-part model of cyber-interactivity: 
Some cyber-places are more
interactive than others. New Media and Society, 4(2), 271-291.
McMillan, S. J., & Hwang, J.-S. (2002). Measures of perceived 
interactivity: An exploration of
the role of direction of communication, user control, and time in 
shaping perceptions of
interactivity. Journal of Advertising, 31(3), 29-42.
McMillan, S. J., Hwang, J.-S., & Lee, G. (2003). Effects of 
structural and perceptual factors on
attitudes toward the website. Journal of Advertising Research, 43(4), 400-409.
Mohnkern, K. (1997). Visual interaction design: Beyond the interface 
metaphor. SIGCHI
Bulletin, 29(2) Retrieved April 1, 2003, from
http://www.acm.org/sigchi/bulletin/1997.2/vid.html
Morris, M., & Ogan, C. (1996). The Internet as mass medium. Journal 
of Communication, 46(1),
39-50.
Newhagen, J. E., Cordes, J. W., & Levy, M. R. (1995). 
<a href="/cgi-bin/wa?LOGON=A3%3Dind0602A%26L%3DAEJMC%26E%3D8bit%26P%3D2461642%26B%3D--%26T%3Dtext%252Fplain%3B%2520charset%3Diso-8859-1" >[log in to unmask]</a>: Audience scope and
the perception of interactivity in viewer mail on the Internet. 
Journal of Communication,
45(3), 164-175.
Norman, D. A. (1998). The invisible computer (pp. 113-133).Cambridge, 
Massachusetts: The
MIT Press.
Mediation Effect of Perceived Interactivity 28
Rafaeli, S. (1988). Interactivity: From new media to communication. 
In R. P. Hawkins, J. M.
Wiemann & S. Pingree (Eds.), Advancing communication science: Merging mass and
interpersonal processes (pp. 110-134). Newbury Park, CA: Sage Publications.
Sims, R. (1997). Interactivity: A forgotten art? Retrieved Sep. 28, 2003, from
http://www.gsu.edu/~wwwitr/docs/interact/
Steuer, J. (1992). Defining virtual reality: Dimensions determining 
telepresence. Journal of
Communication, 42(4), 73-93.
Sundar, S. S. (2000). Multimedia effects on processing and perception 
of online news: A study
of picture, audio, and video downloads. Journalism and Mass 
Communication Quarterly,
77(3), 480-499.
Sundar, S. S., Hesser, K. M., Kalyanaraman, S., & Brown, J. (1998, 
July). The effects of Website
interactivity on political persuasion. Paper presented at the the 
Sociology & Social
Psychology section at the 21st General Assembly & Scientific Conference of the
International Association for Media and Communication Research 
(IAMCR), Glasgow,
UK.
Sundar, S. S., Kalyanaraman, S., & Brown, J. (2003). Explicating Web 
site interactivity:
Impression formation effects in political campaign sites. 
Communication Research,
30(1), 30-59.
Vorderer, P. (2000). Interactive entertainment and beyond. In D. 
Zillmann & P. Vorderer (Eds.),
Media entertainment: The psychology of its appeal (pp. 21-36). 
Mahwah, NJ: Lawrence
Erlbaum Associates.
Wu, G. (1999). Perceived interactivity and attitude toward Website. 
Paper presented at the
annual conference of American Academy of Advertising, Albuquerque, New Mexico.
Mediation Effect of Perceived Interactivity 29
Tables and Figures
Figure 1. Mediation Model of Interactivity
M
b a
Y X
c
X = Objective Interactivity
M = Perceived Interactivity
Y = Attitudes
Table 1: Regression Analyses Summary for the Mediation Model 
Predicting Attitude toward
Website
Step Outcome Variable Predictor Variable
X Y 1
Note. R2 = .21; F(1, 76) = 20.53, p < .001.
X M 2
Note. R2 = .64; F(1, 76) = 136.94, p < .001.
M Y 3
Note. R2 = .48; F(1, 76) = 69.46, p < .001.
X
Y 4
B
1.21
2.58
.56
-.68
.73 M
Note. R2 = .50; F(2, 75) = 37.77, p < .001.
SEB
.27
.22
.07
.36
.11
X = Objective Interactivity; M = Perceived Interactivity; Y = 
Attitude toward Website.
*p < .001.
_
.46*
.80*
.69*
-.26
.90*
Mediation Effect of Perceived Interactivity 30
Table 2: Regression Analyses Summary for the Mediation Model 
Predicting Attitude toward
Politician
Step Outcome Variable Predictor Variable
X Y 1
Note. R2 = .06; F(1, 76) = 4.97, p < .05.
X M 2
Note. R2 = .64; F(1, 76) = 136.94, p < .001.
M Y 3
Note. R2 = .24; F(1, 76) = 24.14, p < .001.
X
Y 4
B
.43
2.58
.26
-.70
.44 M
Note. R2 = .30; F(2, 75) = 16.13, p < .001.
SEB
.19
.22
.05
.28
.09
X = Objective Interactivity; M = Perceived Interactivity; Y = 
Attitude toward Website.
*p < .05; **p < .001.
_
.25*
.80**
.49**
-.41*
.82**