Professional Documents
Culture Documents
Attitude Change in Face-To-Face and Computer-Mediated Communication: Private Self-Awareness As Mediator and Moderator
Attitude Change in Face-To-Face and Computer-Mediated Communication: Private Self-Awareness As Mediator and Moderator
2
0.002. No difference concerning attitude change was found between both communication media,
0.15, p 0.388,
2
0.022, but a trend towards an indirect effect of the communication media on
attitude change via PSA was found, t(35) 1.53, p 0.067, one-tailed,
2
0.063. The regression
weight of the communication media on attitude change is reduced from 0.15 to 0.003 by
including the PSA ( 0.33) into the regression (see Figure 1). The indirect effect of the
experimental condition via both interpersonal inuence measures on PSA was not signicant, both
Sobel ts <1. Similar results are found, when the CMC and the control condition are combined before
the mediation analyses are computed.
Figure 1. Path diagram (with standardized regression coefcients from multiple regression analysis) of the
mediating role of state private self-awareness. The impact of media on attitude change without including private
self-awareness into the regression is given in brackets (Experiment 1; CMC and face-to-face condition only)
2
All effect sizes for from multiple regressions have been computed using the General Linear Model module of SPSS12 and are
therefore partial
2
s.
Self-awareness and CMC 367
Copyright # 2005 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 35, 361374 (2005)
Discussion
The aim of Experiment 1 was to show that less interpersonal inuence in CMC compared to face-to-
face communication occurs due to the higher state private self-awareness produced by CMC. As
predicted group members disagreed more with their groups decision after CMC than after face-to-face
discussions. There was no direct effect of the experimental condition on the other measure of
interpersonal inuence (attitude change). One reason why we failed to nd this effect might be the
moderating impact of trait private self-awareness discussed in Experiment 2. However, there was a
trend towards the predicted indirect effect of the experimental condition on attitude change. The
absence of this pattern for disagreement, which at the same time was more strongly inuenced by the
communication media, suggests that different processes might underlie the effect of communication
media on the amount of disagreement. The impact of communication norms might be stronger for
disagreement than for attitude change (e.g. after agreeing on a group decision one does not diverge
from it). Moreover, the range of the disagreement was very restricted in our sample (01.33,
SD0.32) compared to the range of attitude change (03.11, SD0.70) and the possible range of
the index (04.44). This also works against the predicted indirect effect.
Compared to the face-to-face condition, the amount of disagreement was higher in the control
condition where private self-awareness was manipulated independently of the communication media.
Combined with the trend towards an indirect effect of communication media on attitude change via
private self-awareness, this result suggests that private self-awareness is the crucial psychological
effect of CMC leading to less interpersonal inuence than in the case of face-to-face communication.
One might object that the indirect effect was only marginal and that private self-awareness was not
manipulated during but after the communication in the control condition. However, the converging
evidence from the mediation analysis and the manipulation of the mediator in our opinion compensate
for these limitations.
In general, the results of Experiment 1 are in line with earlier research providing evidence that
private self-awareness lowers attitude change (Froming et al., 1982; Froming & Carver, 1981; Scheier,
1980). The current research extends earlier ndings because a trend towards mediation of the impact
of situational factors on attitude change via private self-awareness to our knowledge has not been
shown so far (see also Postmes & Spears, 1998). Froming et al. (1982) used the presence of a mirror
and an evaluative audience to manipulate private and public self-awareness. The manipulations
showed the predicted effects on compliance. However, it was not tested whether or not differences in
self-awareness mediated these effects. Scheier (1980) and Froming and Carver (1981) tested the
impact of dispositional self-awareness on the public expression of personal beliefs and compliance
respectively. Although they found effects of private and public self-awareness on their dependent
measures, they focused on long lasting interpersonal differences. This aspect of private self-awareness
will be addressed in Experiment 2.
EXPERIMENT 2PRIVATE SELF-AWARENESS AS A MODERATOR
The current experiment addresses the role of trait private self-awareness as a moderator of the impact
of communication media on interpersonal inuence. As explained above, results from Webb et al.
(1989) suggest that trait private self-awareness moderates the effect of communication media on
interpersonal inuence (i.e. lower interpersonal inuence resulting from CMC compared to direct
communication). More precisely, we predict that this effect is stronger, the higher trait private self-
awareness is.
368 Kai Sassenberg et al.
Copyright #2005 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 35, 361374 (2005)
The quasi-experimental test of this hypothesis overall was very similar to Experiment 1, but
some adjustments were made. Only a CMC and a face-to-face condition were included, because no
control condition was necessary to test the prediction. In some groups the CMC discussion in
Experiment 1 lasted more than 1 h. In order to be able to schedule sessions adequately and to run the
study in a more economical way, the rank order task was replaced by a dilemma similar to the most
frequently used dilemma in other research on interpersonal inuence (e.g. Postmes et al., 2001;
Sassenberg & Boos, 2003). Additionally, the participants were informed about the xed time frame for
their discussions. However, no group was actually told to stop before the members reached a common
decision.
Method
Design and Participants
A quasi-experiment with trait PSA and communication media (CMC vs. face-to-face) as independent
variables was conducted. Thirty-eight female and 15 male undergraduate students of the Georg-
August-University of Gottingen (Germany) with a mean age of 25 years (range 1943) took part in this
experiment.
Procedure
Participants were approached in a university building and asked to ll out a questionnaire. This
questionnaire consisted of 20 items including a six item scale measuring trait PSA. The items were
taken from Merz (1986), a German translation of Fenigstein et al. (1975; e.g. I watch signals from my
body carefully., 0.69; internal consistency reported by Merz, 1986: 0.91). Hence, the state
PSA measure used in Experiment 1 and the trait PSA measure administered in this experiment are
based on a similar set of items, but were phrased differently to capture the stable versus variable
component of PSA, respectively. Each item had to be rated on a 6-point scale ranging from 1 for I do
not agree to 6 I entirely agree. The mean PSA score was M4.61 (SD0.65). The remaining 14
items were distractor items asking for team work experience and computer knowledge.
After a time period of at least 1 week the experimenter called the participants in order to arrange the
main experimental session. Participants were invited in dyads or groups of three and randomly
assigned to the experimental conditions. The group size did not differ between communication media
conditions,
2
(1, N53) 0.01,
2
<0.001. Participants were asked to answer the following dilemma
item (administered in German):
A woman has cancer in an advanced state. There is no chance of recovery. Her physician Dr D.
knows that she will die in less than six months time. She suffers from indescribable pain but is so
weak that a strong pain killer would most likely lead to her immediate death. She has visions and
is crazed with pain. Several times she asks Dr D. for help in order to commit a medically assisted
suicide. She explains to him that she cannot stand the pain anymore and that she has to die soon
anyway. Dr D. knows that euthanasia is illegal; nevertheless he considers fulfilling the womans
desire. Please try to imagine that Dr D. asks you for advice. At which probability level of Dr D.s
help remaining undetected by others, would you suggest him to follow the womans wish?
Participants in the face-to-face condition rst noted down probability levels on different tables in
the same room. After the participants completed the dilemma item they were then seated around a
Self-awareness and CMC 369
Copyright # 2005 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 35, 361374 (2005)
common table and asked to introduce themselves to each other in order to make their personal identity
salient. Then they discussed their decision with the other participant(s). In the CMC condition
participants were seated in front of a computer in separate rooms. A chat tool was used to discuss the
dilemma item concerning the probability levels. Again, participants rst names were used as
nicknames. Participants were asked to come to an agreement after 40 min. In the face-to-face
condition participants were asked to stop the discussion as soon as possible with a maximal duration
of 10 min. The shorter time interval was chosen to compensate for the fact that typing is roughly four
times slower than speaking (e.g. Kiesler & Sproull, 1992). Face-to-face discussions were recorded by a
video camera. The video camera might have heightened public self-awareness, but the content of the
CMC was likewise logged and the study was said to be about communication processes in groups.
Hence, participants in both conditions most likely were aware to the same extent that researchers were
interested in their (communication) behaviour. After the discussion individual attitudes were assessed
once again. Finally, participants were debriefed and thanked.
Dependent Measure
The attitude change computed as the absolute difference between the attitude before and after the
discussion served as a measure of interpersonal inuence. Higher values indicate more interpersonal
inuence.
Results
To test the impact of communication media on attitude change, a t-test was computed. Both conditions
did not differ signicantly in the amount of attitude change, t(51) 1.14, p 0.261,
2
0.025. On a
descriptive level the difference was in the same direction as found in Experiment 1. Attitude change
was higher after face-to-face discussions (M26.85) than after CMC discussions (M20.19). The
attitude change was slightly higher in three person groups (M26.84) than in dyads (M15.63),
t(51) 1.79, p 0.080,
2
0.059. Therefore, group size was included as an additional factor in the
analyses reported below.
Concerning the impact of trait PSA, it was predicted that PSA would not directly impact attitude
change but that an interaction between PSA and communication media would occur. More precisely,
we expected that individuals with high trait PSA would react more strongly to the manipulation of
communication media (i.e. show less attitude change in CMC compared to face-to-face communica-
tion) than individuals with low trait private self-awareness. To test this prediction, a multiple
regression with attitude change as criteria and Media, Trait PSA and a Media Trait PSA interac-
tion-term as predictors was computed. To control for the impact of group size, Group Size, the Group
Size PSA interaction and the Group Size Media interaction were included in the analysis.
Following Aiken and West (1991) the interaction terms were computed as a product of the centred
PSA measure and the media variable (0 for CMC and 1 for face-to-face communication). Overall
the predictors explained a signicant amount of variance, adj. R
2
0.15, F(6, 46) 2.51, p 0.035.
Neither the Group Size ( 0.21, p 0.116,
2
0.053), nor the Group Size Self-Awareness
interaction ( 0.26, p 0.118,
2
0.052), nor the Group Size Media interaction ( 0.24,
p 0.100,
2
0.058) contributed signicantly to the explanation of the attitude change. Moreover,
there was neither a main effect of Media ( 0.06, p 0.678,
2
0.004), nor a main effect of PSA
( 0.02, p 0.917,
2
<0.001). Most importantly, the Trait PSAMedia interaction testing our
hypothesis was signicant ( 0.34, p 0.017,
2
0.118). Again following the suggestions of
370 Kai Sassenberg et al.
Copyright #2005 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 35, 361374 (2005)
Aiken and West (1991), we computed simple slopes of attitude change depending on communication
media for individuals with (a) mean PSA, (b) 1 SD above mean, and (c) 1 SD below mean. The pattern
in Figure 2 indicates that for participants with high PSA (b), face-to-face communication leads to more
attitude change than CMC ( 0.39, p 0.043,
2
0.086). However, for individuals with mean (a)
( 0.06, p 0.678,
2
0.004) or below mean (c) PSA ( 0.27, p 0.196,
2
0.036) no
signicant difference between face-to-face communication and CMC occurred. Thus, the hypothesis
that individuals with high PSA responded more strongly to media differences than individuals with
low or average PSA was supported.
Discussion
Experiment 2 sought to test whether trait private self-awareness moderates the impact of commu-
nication media on interpersonal inuence. As predicted we found that face-to-face communication
leads to more interpersonal inuence than CMC only for participants with high trait private self-
awareness. For participants with medium or low private self-awareness no signicant difference was
found. For participants with low trait private self-awareness the pattern even reversed on a descriptive
level. They show a slight trend towards more interpersonal inuence after CMC than after face-to-face
communication. Taken together, the trait private self-awareness moderates the impact of commu-
nication media on interpersonal inuence.
How can one explain the interaction between trait private self-awareness and communication
media? Trait private self-awareness is the tendency to pay attention to ones perceptions, thoughts, and
feelings in general. During face-to-face communication attention is distracted by the interaction
partner(s). However, during CMC this distraction by others behaviour is reduced and cognitive
resources are available to also pay attention to oneself. This opportunity will more likely be used by
individuals who generally have the tendency to pay attention to personal aspects of themselves than by
individuals who do not have this general tendency. In other words, only individuals with high trait
private self-awareness will allocate their attention to themselves in CMC. Hence, only for these
individuals does CMC lead to less interpersonal inuence than face-to-face communication.
Figure 2. Attitude change as a function of communication media and level of trait private self-awareness (PSA;
Experiment 2)
Self-awareness and CMC 371
Copyright # 2005 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 35, 361374 (2005)
CONCLUSION
The current research has shown that the interplay between the communication media on the one hand
and private self-awareness on the other hand, impacts interpersonal inuence in two ways. First, state
private self-awareness mediates the inuence of the communication media on interpersonal inuence.
CMC strengthens the private self-awareness of the communication partners compared to face-to-face
communication. There is some evidence that the heightened private self-awareness, in turn, might
reduce the interpersonal inuence. This nding can be explained by the fact that in CMC less attention
is used by the interaction partners because only written communication is available. Therefore, in
CMC more attention might be paid to ones own personal values, perceptions, and attitudes. Hence,
interpersonal inuence might become less likely to occur. Second, trait private self-awareness
moderates the impact of the communication media on interpersonal inuence. For individuals with
high private self-awareness CMC results in less interpersonal inuence than face-to-face commu-
nication. This effect does not occur for individuals low in trait private self-awareness.
Further research on interpersonal inuence in CMC should address the moderated mediation resulting
fromthe interplay of both effects reported here. The mediation of the impact of the communication media
on interpersonal inuence by state private self-awareness is most likely moderated by trait private self-
awareness. Research in personality psychology has argued that the general tendency to show a certain
behaviour (i.e. a trait) is a precondition for the impact of many situational variations. At the same time the
respective state will mediate the impact of the situation. Scaring someone mildly (situational variation)
will only make very anxious people (trait) run away and this is so because they feel anxiety (state)
(Spielberger et al., 1983). This state-trait interaction as well as the overall pattern of the ndings presented
here (i.e. a state mediating an effect that the respective trait moderates) has rarely been addressed in social
psychological research. However, this pattern may certainly also apply to other domains. Hence, the
current approach might also be a model for research in other domains.
In sum, the current results indicate that the impact of communication media on interpersonal
inuence is mediated by state private self-awareness and moderated by trait private self-awareness.
Earlier research has shown that media differences in norm-based inuence are mediated by the
salience of the social categorization (Lea et al., 2001; Postmes et al., 2001). These different mediators
point to the fact that different characteristics of CMC are responsible for the stronger norm-based
inuence and the weaker interpersonal inuence found in CMC compared to face-to-face commu-
nication. The salience of self-categorization is inuenced by the higher levels of anonymity in CMC,
whereas the differences in state private self-awareness are most likely elicited by the physical isolation
in CMC (Spears & Lea, 1994). All in all the current ndings contribute to the understanding of the
effect of CMC on social inuence.
ACKNOWLEDGEMENT
We are grateful to Kai J. Jonas, Andrew K. Woltin, Heather Smith, and three anonymous reviewers for
their extremely useful comments on earlier drafts of this paper.
REFERENCES
Abrams, D. (1990). How do group members regulate their behaviour? An integration of social identity and self-
awareness theories. In D. Abrams, & M. A. Hogg (Eds.), Social identity theory: Constructive and critical
advances (pp. 89112). London: Harvester Wheatsheaf.
372 Kai Sassenberg et al.
Copyright #2005 John Wiley & Sons, Ltd. Eur. J. Soc. Psychol. 35, 361374 (2005)
Abrams, D. (1994). Social self-regulation. Personality and Social Psychology Bulletin, 20, 473483.
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park:
Sage.
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, 11731182.
Bernstein, I. H., Teng, G., & Garbin, C. P. (1986). A conrmatory factoring of the Self-Consciousness Scale.
Multivariate Behavioral Research, 21, 459475.
Bottger, P. C., & Yetton, P. W. (1988). An integration of process and decision scheme explanations of group
problem solving performance. Organizational Behavior and Human Decision Processes, 42, 234249.
Britt, T. W. (1992). The Self-Consciousness Scale: On the stability of the three-factor structure. Personality and
Social Psychology Bulletin, 18, 748755.
Fenigstein, A., Scheier, M. R., & Buss, A. H. (1975). Public and private self-consciousness: Assessment and
theory. Journal of Consulting and Clinical Psychology, 43, 522527.
Franke, G. H. (1997). U