Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The effect of massed versus spaced arguments on persuasion
(USC Thesis Other)
The effect of massed versus spaced arguments on persuasion
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
THE EFFECT OF MASSED VERSUS SPACED
ARGUMENTS ON PERSUASION
by
Brian M. Monroe
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)
August 2008
Copyright 2008 Brian M. Monroe
ii
Table of Contents
List of Figures iii
Abstract v
Chapter 1: Introduction 1
Chapter 2: Preliminary Experiments 5
Experiment #1 5
Table 1 11
Experiment #2 15
Chapter 3: The Last Three Experiments 25
Experiment #3 29
Experiment #4 37
Experiment #5 46
Further Analyses 51
Chapter 4: General Discussion and Future Directions 53
Chapter 5: Conclusion 61
References 63
Appendix A 66
Appendix B 68
Appendix C 70
iii
List of Figures
Figure 1. Attitude change, experiment 1. 10
Figure 2. Time taken, experiment 1. 12
Figure 3. Recall of arguments, experiment 1. 13
Figure 4. Attitude change, experiment 2. 19
Figure 5. Time taken, experiment 2. 20
Figure 6. Cognitive responses, experiment 2. 22
Figure 7. Mediational diagram, experiment 2, substantive arguments. 23
Figure 8. Attitude at time 1, last three experiments combined. 26
Figure 9. Attitude at time 2, last three experiments combined. 27
Figure 10. Need for Cognition, last three experiments combined. 28
Figure 11. Attitude change, experiment 3. 35
Figure 12. Cognitive responses, experiment 3. 36
Figure 13. Cognitive responses, experiment 4. 42
Figure 14. Attitude change, experiment 4. 43
Figure 15. Attitude certainty at time 2, experiment 4. 44
iv
Figure 16. Time taken, experiment 5. 49
Figure 17. Attitude change, experiment 5. 50
v
Abstract
A manipulation of persuasive messages was explored to see what would result in more
attitude change, such that two formats of presenting a series of arguments were used:
massed and spaced. In massed presentation, five arguments were presented at once,
and thought about all at the same time; in spaced presentation, the same five
arguments were presented one-at-a-time, and thought about one-at-a-time. In the first
experiment, control factors were used to account for confounding explanations, and
spaced presentation resulted in more attitude change, regardless of the control
conditions. The second experiment used two different argument sets, and found that
spaced presentation was more effective than massed regardless of the argument set. A
mediational analysis partially supported a cognitive response mediation explanation
for the original set of arguments, but not for the alternate set. A third experiment
showed that a distraction manipulation did not moderate the massed versus spaced
effect, counter to the cognitive response explanation. A fourth experiment failed to
moderate the effect with a set of counterarguing instructions, also inconsistent with the
cognitive response explanation, but the effectiveness of the manipulation was
questionable. A fifth experiment used a set of instructions to ensure serial order
consideration of massed arguments, and the result failed to replicate. Overall, the
results most support the idea that the robust massed versus spaced effect does not
depend on elaboration of message content, but instead is a low-effort heuristic where
people feel like they have considered more information, and thus their attitudes change
more. Future experiments will need to follow up on this possibility to confirm the
explanation for the effect.
1
CHAPTER 1: INTRODUCTION
Attitude change is arguably one of the most studied topics in social
psychology. This prominence reflects both the natural consequence of encountering
new information relevant to one’s beliefs, as well as the pervasiveness of persuasive
appeals encountered in modern life. Of the many ways in which a persuasive message
can vary, one that seems quite important is the degree to which the message is
concentrated in time or alternatively spread out over time. It can be imagined easily
that these variations can happen in the course of a heated conversation, or listening to
a debate, or reading an opinion piece in the newspaper. One (greatly simplified)
focused comparison that might be of practical interest is whether a political campaign
should employ a series of short television ads spread out in time, each focusing on a
single issue, or whether it should rely on a more comprehensive single ad that explains
all the important positions of the candidate laid out at once. Hereafter, these two
strategies will be referred to respectively as spaced, and massed, persuasion. To date,
research has not directly addressed this kind of question: what method of presentation
is more effective and why.
A consideration of these issues suggests two basic lay-theory possibilities: 1.
Seeing all the information together will be more persuasive because the arguments
build on each other and loosely speaking gain persuasive “momentum” or synergy;
and 2. Allowing someone to receive and subsequently think about each argument in
turn will be more persuasive because they can think through each in an uninterrupted
fashion, allowing uncontaminated elaboration of each point. There is published
evidence that could reasonably be used to support both hypotheses. It is the intention
2
of the present studies to both determine which direction this effect might go, if any,
and to identify the processing mechanism that is responsible for the results, whether it
matches one of the intuitive lay-theory explanations, or is something else entirely.
Reception processes have been identified as an important factor in mediating
attitude change due to persuasive messages. Reception is the term that has commonly
come to represent a combination of both attentional and comprehensional processes
(e.g. McGuire, 1972) McGuire’s explicit (and many others’ implicit) models posit that
reception is a critical step in determining the outcome of a persuasive attempt. The
basic postulate is that the greater the attention to and comprehension of the message,
the more likely the message is to effect attitude change. This postulate has generally
been supported by many lines of research (e.g. Eagly, 1974). It is likely that the
difference, if any, between massed and spaced modes of presentation lies in the
domain of message reception. However, isolating the specific antecedents of
reception has proved to be very complex, and there are innumerable ways that the two
steps in reception can be manipulated.
In support of the hypothesis that massed persuasion might be more effective,
there is research that memory, and specifically memory for persuasive
communications decays over time, along with the persuasive effects (N. Miller &
Campbell, 1959; Watts & McGuire, 1964). Thus, if messages are spaced, all
arguments and especially those that are presented earlier have more time to decay, and
at the time of assessment the entire message taken as a whole would have less impact.
However, recall for the persuasive arguments does not always correlate positively with
attitude change in these studies (see also Insko, 1964). Further research has explored
3
alternative relationships between message processing and attitude change, which has
led to the conclusion that direct recall of the message may have less influence than
self-generated cognitive responses to the message(Greenwald, 1968).
In support of the hypothesis that spaced persuasion might be more effective, a
review of the learning literature shows that spaced presentation of items in a learning
phase leads to greater recall of those items later, due either to reduced attention in a
massed presentation, or greater variability in encoding in the spaced condition, leading
to more available recall cues and better retention in the recall phase (Cepeda, Pashler,
Vul, Wixted, & Rohrer, 2006). However, due to the lack of a clear relationship
between recall and persuasion noted above, this evidence is not conclusive regarding
the advantages of spaced over massed. In addition, the subject of presentation in the
massed versus spaced learning experiments was always a single item that was
repeated, as opposed to a number of different persuasive arguments that is more
relevant in an attitude change context.
There is another theoretical approach to attitude dynamics that leads to the
conclusion that a more concentrated message might be more effective. In complex
and/or chaotic systems, there exist certain states of the system that are more stable,
and certain states that are less stable. The more stable a state is, the more likely the
system is to “gravitate” toward that state. These attractors are a natural consequence
of these chaotic systems, where the system evolves “down” an energy curve towards a
local minimum, much like a star or a planet attracts small bodies and pulls them close
to it, where the small body’s gravitational potential energy is minimized. Attitudes are
hypothesized to be an example of attractor systems (Eiser, 1994; Nowak &
4
Lewenstein, 1994; Read, Vanman, & Miller, 1997). The strength of an attractor in an
attitude system would depend on the nature and strength of the interconnections
between concepts and statements in a belief and/or semantic network. It is also
posited that the network prefers to be in a coherent arrangement, where all nodes in
this network fit together in a self-consistent manner. Roughly, when there are more
and stronger connections between relevant concepts, it is harder to dissolve the
existing organization and reorganize the network into a different coherent state, and
the attractor system is stronger. Because of this, a certain amount of impetus is
necessary to overcome this energy well and move the attitude system from its stable
state. Thus, when there is a strong attractor, a more concentrated (massed) message
might be able to give the system sufficient impetus to overcome the stability; in
contrast if the (spaced) persuasion effort is more spread-out, and thus the impetus from
the message is more spread out, the stability of the system might cause the attitude to
revert back to the stable configuration before the persuasive message is able to
overcome it. An alternate way to think of this is that when trying to move a heavy
object, it takes a concentrated effort from several people to successfully move it, while
if each person tried to move it repeatedly but separately these spaced efforts would
have no success.
In the remainder of this paper, a description and analysis of studies
investigating these different possibilities will be laid out. These experiments have
answered several important fundamental questions about this research, and
constrained and guided the theory in explaining the phenomenon, while at the same
time leaving many questions unanswered.
5
CHAPTER 2: PRELIMINARY EXPERIMENTS
Experiment #1
The general outline of the experimental paradigm, used for all experiments in
this proposal, was to first give subjects an initial attitude on a central issue.
Specifically, extensive information was given to induce subjects to be in favor of the
issue. Then the experimental phase exposed subjects to some persuasive arguments
specifically designed to change their opinions, either in the massed or spaced
presentation format, so that their attitude would presumably change in an anti-issue
direction. Finally the attitude was measured again, and the difference in attitudes from
before the persuasion to after the persuasion served as the primary dependent variable
of interest. The first experiment was designed to test for the basic effect, and to rule
out some alternative possibilities.
Method
Design
The experiment used a 2 × 2 × 2 factorial design, with the primary factor being
massed vs. spaced presentation of arguments. The first of the two other control factors
was whether there was an intervening task between the spaced arguments—this was
implemented to ensure that the spaced presentation of arguments was actually spaced,
cognitively speaking, by having intervening mental activity such that the subject
would be thinking about something else in between persuasive arguments. The second
control factor was whether the time given to the subjects to consider the arguments
was fixed, or self-paced (reading time was still self-paced in this condition). This was
6
implemented to explore the possibility that the mode of presentation (massed or
spaced) might result in subjects using a different amount of time to think about the
arguments. If subjects in one condition spent more time considering the arguments, it
might be closely related to or help explain the psychological mechanism involved in
producing the effect. The procedure will be described in more detail below because
the subsequent experiments use most of the elements of this paradigm.
Participants
Participants were 107 undergraduates from the University of Southern
California subject pool, who were given course credit for their participation.
Materials
The issue at the center of the attitude change experiment was a hypothetical
zoning change proposal (see Appendix A). The subjects were given the arguments in
favor of the proposal (pro-arguments) first, in the form of an extended paragraph. The
pro-arguments were given to the subjects to induce an initial position on the central
issue, from which the attitude could be measured and subsequently changed. The
arguments against the proposal (con-arguments) were given after the arguments in
favor, in either the massed or spaced format (see Appendix B). In all conditions there
were a total of five arguments against the proposal. In the massed format, these
arguments were presented as five separate bullet points (each a sentence or two long)
on the same web page. In the spaced format, the arguments were presented on five
successive pages, with one bullet point on each page. The order of presentation of
these arguments was randomized for each subject.
7
Measures
The attitude measure was used twice, once after the pro-arguments phase, and
again after the con-arguments phase. The attitude measure consisted of 3 items (see
Appendix C) on continuous scales that ranged from 0 to 10, with endpoints given
labels (such as not good at all-it’s a great idea) and midpoints labeled neutral. The
first two were designed to measure the main dependent variable with the third being a
more exploratory item. A free recall measure was also used, that assessed retention of
both the pro-arguments and the con-arguments. The intervening task used in the
appropriate conditions was a word jumble task. Subjects were given a set of six
letters. They were asked to come up with as many 3-, 4-, 5-, and 6-letter words as
possible from this combination of letters, and were given 90 seconds to do so. There
were a total of five different word jumbles, which were the same across all subjects.
Word jumble performance was not scored.
Procedure
Subjects started the experiment by viewing the introduction web page. On the
next page, the scenario was introduced, and they were exposed to the pro-arguments.
The reading of the pro-arguments was self-paced, regardless of the condition. At this
point, in order to ensure that they processed the information in detail, they were asked
to write additional thoughts about the proposal and come up with their own reasons
why this proposal was a good idea. They were given five minutes to complete this
writing task. Then, the attitude measure was completed for the first time. After this,
subjects were exposed to the con-arguments. In the self-paced condition subjects were
just asked to think about the arguments for as long as they felt they needed to, and
8
then clicked a button signifying that they were done reading them; in the massed
condition they did this only once after reading all five arguments, and in the spaced
condition they did this five times, once for each argument. Within the fixed-time
condition, subjects in the massed condition were given 100 seconds to think about all
five con-arguments, and in the spaced condition, subjects were given 20 seconds to
think about each of the five con-arguments, at which point they would move on to the
presentation of the next argument (the total time given for them to think was thus also
100 seconds).
In the non-intervening task condition, subjects moved directly from
consideration of one argument to reading the next argument, or if they had finished the
last argument they would move directly to the second attitude assessment. In the
intervening task condition: when subjects in the spaced condition were finished
considering one con-argument, instead of going directly to the presentation of the next
con-argument, they would be presented with a word jumble; in the massed condition,
all five word jumbles were presented consecutively after they finished considering all
five con-arguments.
When all con-arguments were considered and all word jumbles were
completed (where appropriate), the attitude measure was completed for the second
time. Finally, subjects were given five minutes to complete the free-recall measure,
and then they were debriefed and concluded the experiment session.
9
Results
Data screening
Subjects were excluded from the analysis if they failed the induction to
produce the initial attitude: i.e. if either of the items on the time 1 attitude measure
were below 4. This allowed subjects to remain in the analysis even if they were
ambivalent about the target proposal (a score of 5 being a neutral rating), but excluded
those who did not think it was a good idea in the first place, even given only positive
information about it. This led to the exclusion of about 10 subjects. Subjects were
also excluded for taking too little or too much time to read and consider the
arguments. Based on an outlier analysis of the distribution of log-transformed times,
the lower threshold was about 30 seconds, and the upper threshold was about 700
seconds (this is the time taken only to read and consider all the arguments, not the time
taken for the entire experiment.) This led to the exclusion of about 5 subjects. An
outlier analysis was performed on attitude change values, but no outliers were detected
that were not covered by the other exclusion criteria.
10
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
self-paced timed
massed
spaced
Figure 1. Attitude change, experiment 1.
Attitude change
The first two items of the attitude measure had a reliability of alpha=.87 at
time 1, and alpha=.91 at time 2. Thus the items were combined into a composite score
for the main dependent variable. Attitude change scores were calculated as within-
subject difference scores of the attitude composite. The data was then analyzed with
standard ANOVA procedures, with this difference score as the primary dependent
variable (see Figure 1). The results showed a significant main effect for massed
versus spaced (MvS), with spaced persuasion resulting in more attitude change than
massed, F(1,99)=7.069, p=.009, r=.252 (see also Table 1). Main effects for the two
control factors were nonsignificant, for the task factor, F(1,99)=.185, p >.6, and for the
timed/self-paced factor, F(1,99)=.003, p >.9. All interactions were also
11
nonsignificant, ps >.5. Thus the control factors of fixed time vs. self-paced time, and
intervening task vs. no task did not affect how much attitude change there was.
Condition Mean SD
Massed
No task
Self-paced 2.21 1.67
Timed 1.90 1.73
Intervening task
Self-paced 1.52 1.28
Timed 2.13 2.39
Spaced
No task
Self-paced 2.97 2.48
Timed 2.88 1.69
Intervening task
Self-paced 2.92 1.58
Timed 2.79 1.36
_____________________________________________________________________
Table 1. Attitude change, all cells, experiment 1.
Time taken
The total time taken to both read and consider all the arguments was calculated
for each condition, and the patterns were explored (see Figure 2). A factorial
ANOVA revealed a main effect for massed versus spaced, F(1,99)=7.081, p=.009, a
main effect for timed vs. self-paced, F(1,99)=47.8, p<.001, and a marginal interaction
for MvS and timed, F(1,99)=3.820, p=.053. The interpretation of this interaction is
that there is a difference between massed and spaced in the timed conditions (keeping
in mind the amount of time subjects took to read the arguments before they gave them
additional consideration was not timed, so this is where the difference came from), but
not in the self-paced conditions. The simple effect in the timed conditions was
12
F(1,48)=16.7, p<.001, with spaced taking more time than massed. A mediational
analysis of time taken and attitude change (in the timed conditions only since this is
where there were group differences) showed the correlation between time and attitude
change was not significant, p >.5.
0
50
100
150
200
250
300
350
self-paced timed self-paced timed
task no task
read + think time (sec)
massed
spaced
Figure 2. Time taken, experiment 1.
Recall of arguments
For recall, arguments were coded independently by two research assistants, as
either correct recall of an argument (valued at +1), partial recall of an argument
(valued at +0.5), or no recall of an argument (valued at 0). For purposes of this recall,
the pro-arguments were partitioned into six separate arguments, even though they
were presented in continuous fashion during the experiment. The coding of the five
13
con-arguments was straightforward. The recall score for each subject was computed as
the average of the two coders’ scores. The coders’ ratings were reliable at alpha=.86
for the pro-arguments, and alpha=.78 for the con-arguments. Analysis of the recall
scores (see Figure 3) showed that recall was affected by the massed vs. spaced
manipulation, such that con-arguments were recalled more accurately in the spaced
condition, F(1,99)=5.969, p=.016. Differences for total recall were nonsignificant,
F(1,99)=.154, p>.6 (it was not surprising that a total recall measure did not yield
differences, because the pro-arguments were presented before any procedural
differences across conditions.) A mediational analysis showed that this recall was not
significantly correlated with the amount of attitude change, and the indirect effect of
recall as a mediator was nonsignificant as well.
0
1
2
3
4
5
6
7
unfavorable arguments total arguments
Arguments recalled
massed
spaced
Figure 3. Recall of arguments, experiment 1.
14
Attitude certainty
Analyses of attitude certainty showed that certainty at time 1 had nothing to do
with the amount the attitude ultimately changed, correlation p >.3. Analysis of
certainty at time 2 showed a more noticeable pattern. An ANOVA yielded no
significant main effects, massed vs. spaced F(1,99)=1.711, p=.194, massed vs. spaced
and timed interaction F(1,99)=2.742, p = .101. (Theoretically, the effects for the other
factors were not interesting, and all ps > .3.) The simple effect of massed vs. spaced
in the timed conditions showed that those in the spaced condition were marginally less
certain, F(1,48)=3.513, p = .067. This pattern did not mediate the amount of attitude
change, nor did the amount of attitude change mediate the certainty at time 2.
Discussion
The first experiment showed that spaced persuasion was more effective than
massed, regardless of whether there was an intervening task or not, and whether the
time taken to consider the arguments was fixed or self-paced. Thus the results seemed
fairly robust. But an important part of the overall picture needs to include a more
detailed explanation of why this basic effect happens. The results of the time analysis
showed that time taken to read and consider the arguments could differ, but ironically
this difference was not present in the completely self-paced conditions.
As mentioned in the introduction, recall is not always consistently related to
attitude change. Moreover, despite the fact that retention is better after spaced
presentation of the arguments , the lack of mediation in the mediational analysis
suggests that this is not be an adequate explanation. The recall data along with the
15
timing data strongly suggest there is an information processing difference between the
two conditions though, so it suggests looking at theories in attitude change that focus
on information processing stages. The cognitive response framework of attitude
change, which focuses on thoughts generated by the message receiver, rather than the
information explicitly in the message, might be able to account for the attitude results
seen.
Experiment #2
Cognitive Responses in Persuasion
A possible candidate for a cognitive predictor of differential attitude change
across the massed vs. spaced manipulation is cognitive responses to the persuasive
arguments. The cognitive response framework is empirically an extremely well
supported approach for understanding attitude change. This framework postulates that
part of the information processing that happens upon receipt of a persuasive message
involves generating thoughts that relate the new information to the existing set of
beliefs, and it is the nature of these self-generated thoughts that, in large part,
determines the degree of attitude change (Greenwald, 1968; Petty, Ostrom, & Brock,
1981). The support for this framework comes from, among other studies, work that
shows that distraction while receiving a message reduces cognitive responding
(Osterhouse & Brock, 1970; Petty, Wells, & Brock, 1976) and thus attitude change,
and also experiments that manipulate involvement with the attitude object, and thus
the degree of cognitive responding and persuasion(Petty & Cacioppo, 1979b).
The typical measure of cognitive responding in these experiments is a thought-
listing task—this asks the subject to list the thoughts they had in response to the
16
persuasive communication, typically assessed after the attitude measure. These
thoughts would be coded as either pro or con (often by the subject), and could be
summed to form an index that would be used to predict the amount of attitude change.
It is possible that the spaced condition allowed more opportunity for this
elaboration to take place and additional thoughts to be generated. If in the spaced
condition, more thoughts consistent with the persuasive message were in fact
generated, it would be consistent with the cognitive response explanation for the
effect. However it was not yet known if this process would actually be affected by the
massed vs. spaced manipulation, and so the second experiment was conducted to test
this.
Heuristic versus systematic processing
Dual-process models of attitude change, namely the Heuristic-Systematic
Model (Chen & Chaiken, 1999), and the Elaboration Likelihood Model (Petty &
Cacioppo, 1986) suggest that arguments based on fairly simple associative relations,
rather than complex arguments, are less likely to encourage elaborative thought
processes. If this were the case, it might be expected that arguments that don’t
encourage elaboration would exhibit a different effect from substantive arguments as
long as the elaboration process does have a chance to occur. If as theorized the spaced
condition allowed for more opportunity to elaborate, a message receiver would also
need the necessary information in order to elaborate. A simple message like a
heuristic message might not give them enough information to elaborate on, and so
there might not be a difference between the massed and spaced condition for the non-
substantive arguments even if there were for the regular (substantive) arguments. This
17
would be another piece of evidence pointing towards the cognitive response
explanation for the phenomenon.
Method
Design
The second experiment used a 2 × 2 × 2 factorial design. The primary factor
again was massed vs. spaced. The secondary factors were intervening task vs. no task,
and substantive vs. non-substantive con-arguments (see below).
Participants
184 students from the USC subject pool were run through the experiment.
Materials
The scenario materials used in this experiment were the same as in the first
experiment. An additional set of arguments was constructed, though, to test the
possible differences between “substantive” arguments (like the kind used in the first
experiment), and “non-substantive” arguments, that might be less likely to induce
cognitive responding at all, or alternatively less-one-sided cognitive responding. An
example of the non-substantive arguments used is one that refers to the expertise of
the person backing the viewpoint, or the sheer number of points they have as cues for
why the argument should be agreed with, rather than the content of their arguments.
These non-substantive arguments are presented in Appendix B.
Dependent measures
The attitude measures were identical to the ones in the first experiment. A
thought-listing task was added, where subjects were first asked to list thoughts they
had while they were considering the persuasive (con) arguments. After they were
18
finished listing their thoughts, they were asked to code their own responses as positive
(pro-proposal), negative (anti-proposal), neutral, or irrelevant by marking a plus sign,
minus sign, or equal sign next to that comment. This self-coding is typical for the
thought listing measure. The same intervening task (word jumbles) was used as in
experiment 1.
Procedure
The procedure was mostly the same as experiment 1. The important
differences were: all conditions used self-paced con-argument consideration; and the
thought-listing task was inserted after the second attitude measure and the recall task.
The thought-listing task was self-paced.
Results
Data screening
Subjects were excluded from the analysis for the same reasons as in the first
experiment. This led to the exclusion of approximately 25 subjects. In addition,
subjects were excluded from the thought listing analyses only if they completed all
parts of the experiment but failed to follow the thought listing directions. About 20%
of subjects were excluded from the thought listing analyses in this experiment.
Attitude change
At time 1, the reliability of the two-item attitude measure was alpha=.80 at
time 1, and alpha=.81 at time 2. This was unexplainably less than in the first
experiment, but still acceptable. The two items were averaged into a composite score
for analysis and a factorial ANOVA was performed. The main effect of massed
versus spaced replicated the first experiment, F(1,181)=5.508, p=.020. There was also
19
a significant main effect for the argument set, such that the substantive arguments
produced more attitude change than the non-substantive arguments, F(1,181)=6,992,
p=.009. No effect was found for task/no task, F(1,181)=.726, p > .4, replicating the
negative result in the first experiment, and so this factor was collapsed for all further
analyses. Counter to the hypothesis, no interaction was found for the two factors with
significant main effects. Examining simple effects for each argument set, the
substantive arguments produced a significant MvS effect, F(1,99)=5.285, p=.024,
r=.229; the non-substantive arguments did as well, F(1,78)=4.183, p=.044, r=.225.
These results are in Figure 4.
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
substantive arguments non-substantive arguments
Attitude change
massed
spaced
Figure 4. Attitude change, experiment 2.
20
Time taken
There was a very strong MvS effect on time taken, F(1,181)=32.61, p<.001,
which did not show up in the first experiment. An unsurprising weakly marginal
effect was present such that the substantive arguments took longer to read and
consider than the non-substantive arguments, F(1,181)=2.492, p = .12. Interactions
were nonsignificant, ps< .4. These results are in Figure 5.
0
50
100
150
200
250
300
350
self-paced timed self-paced timed
task no task
read + think time (sec)
massed
spaced
Figure 5. Time taken, experiment 2.
Attitude certainty
The MvS effect on certainty at time 2 that was present in the first experiment
did not replicate in this experiment, main effect and simple effects (for each argument
set) ps> .7.
21
Thought listing / cognitive responses
Analysis of the thought listing data with a factorial ANOVA shows no
significant main effects (ps > .4) or interactions (p = .14) on negative (unfavorable to
the proposal) thoughts generated, total thoughts generated, and negative minus
positive thoughts. More important though were simple effects analyses because the
aim was to look at the MvS effects for each argument set individually. The simple
effect on negative thoughts for the substantive argument set was marginally
significant, F(1,99)=3.717, p=.057. A negativity index was computed by taking the
difference in negative and positive thoughts, and dividing by the total number of
thoughts. For the substantive arguments, this effect was also significant,
F(1,99)=5.058, p=.027. For the non-substantive argument set, group differences in
negative thoughts were nonsignificant, F(1,62)=.670, p >.4. The negativity index
group difference was also nonsignificant, F(1,62)=.053, p >.8. These results are
presented in Figure 6.
22
0
0.2
0.4
0.6
0.8
1
1.2
1.4
substantive non-substantive
Normalized thought negativity index
massed
spaced
Figure 6. Cognitive responses, experiment 2. Negativity index formed by subtracting
positive from negative thoughts, and dividing by total (positive plus negative)
thoughts.
Mediation Analyses
Cognitive responses. To see if cognitive responses mediated the MvS effect
for the substantive arguments, a bootstrap mediation analysis (Preacher & Hayes,
2004)was run with attitude change as the dependent variable and the negativity index
as the mediator. The path coefficients are in Figure 7. The 95% confidence interval
for the indirect effect was [.003,.252]. Reverse mediation also comes out
significantly, with a confidence interval of the indirect effect of [.003,.265], which
calls into question the direction of causality. The issue of which dependent variable is
the mediator and which is the outcome in cognitive response research has been
23
discussed by others (N. Miller & Colman, 1981), and is left an open question here,
though the causal model initially prescribed here is generally accepted in the literature.
Assuming the thoughts-as-mediator model, the direct effect of MvS on attitude change
is still marginally significant, suggesting this is only a partial mediation. Because
there were virtually no group differences in thoughts for the non-substantive argument
set, it was not reasonable to assume a mediational pathway of cognitive responses for
those arguments.
Figure 7. Mediational diagram, experiment 2, substantive arguments.
Time taken. An additional variable to test a mediation hypothesis is the time
taken to read and consider the arguments. Time did not mediate attitude change at all
for the substantive argument set, with a zero indirect effect well within the 95%
24
confidence interval, [-.085,.296]. Nor did it mediate attitude change for the non-
substantive argument set, confidence interval [-.824,.521].
Discussion
The results from experiment 2 provide suggestive evidence that the cognitive
response mechanism is at least partly responsible for the massed vs. spaced effect. For
the substantive arguments, the manipulation had an effect on the number of anti-
proposal thoughts generated as well as the proportion of total thoughts that were anti-
proposal, and the number of these thoughts partially mediated the degree of attitude
change. However, for the non-substantive arguments, a mediator was not found
among the measures collected. This suggests either that the thought listing measure is
unreliable to use with the non-substantive arguments, or that the MvS effect can occur
in the absence of differential elaboration. If increased elaboration is unnecessary, the
dual process theories mentioned above explicitly leave the option that the mechanism
producing more attitude change under spaced persuasion is a non-thoughtful one.
Attitudes can be influenced by information other than the content of the messages
themselves, and these options will be discussed later. Or it could be that a yet
unknown mediator is responsible. However, the cognitive response hypothesis
remains the most viable explanation at this point, but because the support for it was
not completely convincing, more experiments with different manipulations will be
used to further test and potentially confirm this explanation, and to extend the findings
and delineate the mechanisms of cognitive processing on a more detailed level that are
responsible for the phenomenon. Whatever the explanation for the effect, a better
handle on the cognitive response mechanism is warranted.
25
CHAPTER 3: THE LAST THREE EXPERIMENTS
The following three experiments were proposed and run at the same time. Because
subject collection was slow, in addition to the USC subject pool, subjects were
simultaneously recruited on the Internet. Also because of slow data collection,
additional subjects were collected from the psychology department subject pool at
California State University, Long Beach. Because it is possible that these different
subject populations worked differently with the materials, some descriptive statistics,
collected across the experiments, are presented and compared among the three
samples.
Data screening
As in the first two experiments, subjects were excluded for an initial attitude
that was too low, and for taking too little or too much time. This led to the exclusion
of about 5% of the sample. An outlier analysis on attitude change was also performed,
and subjects were excluded for exceeding 2.5 standard deviations from the sample
mean across all conditions. This led to the exclusion of another 5% of the sample.
Finally, subjects who failed to follow the thought listing directions were excluded
from the cognitive response analyses. This was the largest category of exclusions,
approximately 25% of the sample. Many of these individual exclusions overlapped,
so the total number of exclusions was less than the sum of the individual categorical
exclusions. In the end, about 550 of the 715 subjects were kept for all the analyses.
26
4
5
6
7
8
9
10
USC CSULB Internet
Time 1 Attitude Score
Figure 8. Attitude at time 1, last three experiments combined.
Descriptives
Attitude at time 1 (see Figure 8) shows that the Internet sample was initially
more favorable towards the proposal in the scenario after reading only the positive
information. A one-way ANOVA showed that these differences were indeed
significant between the students and Internet sample, F(1,539)=20.817, p<.001, and
that the difference between the USC and Long Beach students was also significant,
with USC students being the least initially favorable overall. Attitude at time 2 (see
Figure 9) also shows significant differences between students and Internet with the
Internet subjects still being more favorable, F(1,539)=44,28, p<.001. Differences
between USC and Long Beach were again much smaller than between USC and the
Internet sample. Attitude change was also different between students and Internet
27
subjects, F(1,519)=17.71, p <.001. An analysis of the thought listings used in the
experiments revealed the arguments weren’t as convincing to the Internet subjects,
F(1,519)=17.859, p <.001. The mean of the negative minus positive thoughts
generated was M=1.12, SD=2.62 for students, and M=-.01, SD=3.31 for Internet
subjects. All of these descriptives suggest that the materials weren’t nearly as
convincing to the Internet subjects, although because all subjects did receive
information both for and against the proposal, by itself the thought listing mean
presented above doesn’t mean the negative arguments weren’t effective by themselves
on the Internet subjects. Attitude did still change in the expected direction from time 1
to time 2 after all.
4
5
6
7
8
9
10
USC CSULB Internet
Time 2 Attitude Score
Figure 9. Attitude at time 2, last three experiments combined.
28
Figure 10. Need for Cognition, last three experiments combined. Student sample (top
panel) and Internet sample (bottom panel).
A Need for Cognition (NFC) measure was collected, also showing a very large
group difference between students and non-students, with a one-way ANOVA
29
showing large differences between students and Internet, F(2,519)=50.29, p <.001.
The post-hoc analyses showed that it was the difference between both student groups
and the Internet group that was significant, Tukey ps <.001, and not any difference
between the USC and Long Beach samples, p >.9. The average NFC score for the
USC students was 8.2, for Long Beach it was 7.7, and for the Internet sample it was
22.2 (on a scale that runs from –54 to +54). Histograms of NFC scores are presented
in Figure 10. Overall, these differences clearly suggested looking at the student and
Internet samples separately as well as together, and that is done below.
Finally, across the three experiments, the 2-item main dependent measure of
attitude had a reliability of alpha=.89 at time 1, and alpha=.90 at time 2, so the two
items were again averaged to form a composite measure.
Experiment #3
If in fact the differential effectiveness of spaced over massed persuasion is due
to cognitive responses generated while processing the material, it should be possible to
neutralize this effect by disrupting the cognitive processes that occur during reception
of the persuasive arguments. It is well established that by distracting the receiver of
the persuasive message, their thought process can be altered either to increase or
decrease attitude change. When the subject is naturally motivated to respond
unfavorably to the persuasive message and able to muster counterarguments, a
sufficient distraction reduces counterarguing, and thus increases the effectiveness of
the persuasion attempt (Petty et al., 1976) while if the cognitive responses to the
message are pro-persuasion, the distraction interferes with this generation and reduces
persuasion (Petty & Brock, 1981). Showing that the current research paradigm
30
conformed to these effects would be convincing evidence that spacing promoting or
enabling more cognitive elaboration is the mechanism that drives the massed versus
spaced effect.
Since, with the current materials, the predominant cognitive response is to
generate pro-persuasion thoughts, this results in the hypothesis that increased
distraction should reduce persuasion in both conditions, but more so in the spaced
condition (where more thought generation is taking place) than in the massed
condition. If distraction reduces the difference between massed versus spaced in the
amount of persuasion, this would support the cognitive response interpretation of the
massed versus spaced effect.
As an additional test of the cognitive response framework, Need for Cognition
(NFC) (Cacioppo & Petty, 1982) will be assessed. The NFC scale identifies people
who are naturally motivated to think and process information in a more detailed
manner. If people who are high on NFC indeed think more about the persuasive
arguments, then it might yield differential attitude change. The nature of this effect
could conceivably go either way, with massed persuasion being more or less effective,
depending on whether more thought uniquely due to NFC occurs in the massed or the
spaced condition. If the NFC effect impacts the massed condition more (i.e. there is a
ceiling effect in the spaced condition so regardless of NFC, all subjects will think
about the arguments as much as they need to) then the massed versus spaced effect
should be reduced. If the NFC effect impacts the spaced condition more because there
is more opportunity to think more thoroughly, then the massed versus spaced effect
might increase.
31
Method
Design
The experiment was a 2 × 2 factorial design, with the primary factor being
massed versus spaced, and the second factor being distraction versus no distraction.
The NFC measure constitutes a potential third between-subjects factor. All conditions
for this experiment were fixed-time argument reading and consideration, due to the
nature of the distraction implementation described below.
Participants
152 participants were accrued, 104 students from the University of Southern
California, and California State University Long Beach, as well as 48 gathered from
Internet recruitment.
Materials
The persuasion materials were the same as in the first two experiments, using
the original (substantive) arguments from the first experiment.
The distraction consisted of a flashing numbers task. During the phase of the
experiment where subjects were asked to consider the persuasive arguments, they
were simultaneously asked to keep track of numbers that flash randomly in their web
browser window. The normal experiment webpages, without distraction, contained
black text on a white background. The flashing numbers were white text on a red
background, in a square that was roughly one half inch on a side. The size of the
numbers was in 25-point font (significantly larger than the 12-point font size used for
the text for all other materials in the experiment). The number in the box randomly
varied as an integer from 0 to 9. Each time the number flashed, it was visible for one
32
half second and then disappeared until it flashed again. The position of the flashing
number on the screen was randomized each time it was displayed, and the interval
between successive flashes varied randomly and continuously from 2 to 4 seconds.
Subjects were asked to keep track of how many times the number 2 flashed during
each argument consideration phase. This choice of distraction task was intended to
give subjects enough time to still read over and consider the arguments, while
periodically being forced to change their gaze and switch mental tasks to determine if
the number displayed matched the number they were supposed to keep track of. The
distraction implementation resulted in an expected value of approximately 6.7 total
number flashes for an individual spaced argument (for which the consideration time
was 20 seconds), and approximately 33 for all five massed arguments (for which the
consideration time was 100 seconds). Since there was a one-in-ten random chance of
the number 2 being flashed, the number of times the number 2 actually flashed would
have an expected mean of about 0.7 for a single argument, and 3.3 for the massed
arguments.
Dependent measures
Attitudes and thought listing were assessed the same way as in previous
experiments. At the end of the experiment, the Need For Cognition 18-item inventory
(Cacioppo, Petty, & Kao, 1984) was added.
Procedure
The procedure for this experiment was identical to experiment 2, with the
following modifications. In all conditions, there was no word jumble task. In the
distraction conditions only, before the experiment starts in earnest, each subject was
33
taken through a demonstration of the flashing numbers task they were required to
perform. Then, after each flashing numbers episode (there was one only for the
massed condition, and five for the spaced condition since they corresponded to each
time the subject was asked to consider the argument(s)) a new screen came up and
subjects were asked how many times they saw the number 2, before they continued
further with the experiment. In the no-distraction condition, a 2-second wait screen
took the place each time of this distraction assessment screen. In all conditions, after
the second attitude assessment the thought-listing task and recall tasks was performed
as before, and then finally the NFC questions were answered.
Results
Manipulation check / thought listing
A casual inspection of the free responses in the thought listing task revealed
that many subjects mentioned how distracting the flashing numbers task was, and that
it interfered with their ability to concentrate on the arguments. A more quantitative
analysis of the thought listing revealed that the main effect of distraction on total
thoughts was nonsignificant, F(1,148)=.155, p > .6, and for anti-proposal thoughts (the
ones that might be likely to differ since the anti-proposal information is presented
during distraction) the main effect was also nonsignificant, F(1,148)=.261, p >.6. (In
the student subsample, p > .9, and in the Internet sample, p > .8.) There was one
marginal simple effect, in the total sample for massed arguments only due to
distraction, total thoughts F(1,69)=3.060, p=.085 such that distraction reduced the
number of thoughts. This was not seen for the spaced condition. The thought listing
will be discussed in more detail below.
34
Time taken
With the exception of a few short transition screens (for example, the screen
where subjects were asked how many times the number ‘2’ flashed during the flashing
numbers task), the entire experiment was run under enforced time limits, so there were
no significant group differences in the time taken to read and consider the arguments.
Nonetheless, the average time taken in the critical part of this experiment was 159
seconds, with a standard deviation of 8 seconds.
Attitude change
A factorial ANOVA was once again used to analyze the data. The main effect
of massed vs. spaced replicated previous experiments, F(1,148)=4.456, p=.036. There
was a weakly marginal, and unsurprising main effect for distraction such that
distracted subjects showed less attitude change, F(1,148)=2.434, p=.12. The
interaction was however nonsignificant, F(1,148)=.137, p> .7. Broken down into the
student subsample, the MvS main effect remained significant, F(1,100)=5.927,
p=.023, but was nonsignificant for the Internet sample, F(1,44)=.234, p> .6. The
distraction main effect for the student sample was marginally significant,
F(1,100)=3.501, p=.064. The distraction effect was nonsignificant for the Internet
sample, p >.7, as was also true above with a small sample size. There was also
nothing approaching an interaction in either subsample, all ps > .5. Results are
presented in Figure 11.
35
0
1
2
3
4
5
6
no distraction distraction no distraction distraction no distraction distraction
total sample student internet
Attitude Change
massed
spaced
Figure 11. Attitude change, experiment 3.
An ANCOVA was performed with Need for Cognition as the covariate to see
if it altered the pattern of results, since Need for Cognition could be correlated with an
increased amount of processing and thus attitude change. It did not meaningfully
change the pattern of attitude change.
Finally, using Need for Cognition as a continuous independent variable, and
inputting it into the general linear model with the other categorical factors, yielded 2-
and 3-way interactions all with ps > .4. So Need for Cognition did not have any
significant effect on the pattern of results, within the total sample, or within the
student sample (or the Internet sample, but the small sample size likely hampered any
effort to detect positive effects there).
36
Cognitive responding and mediational analyses
To see if cognitive responding mediated the MvS effect on attitude change, a
mediational model was run. Unfortunately, the MvS manipulation did not
significantly affect cognitive responding, either in the number of negative thoughts
generated, or proportion of negative thoughts generated, so mediation was not
possible. The effect of MvS on negative thoughts was nonsignificant, F(1,148)=..995,
p> .3. In the student subsample, it approached significance, F(1,100)=2.160, p=.145,
and the effect did not approach significance in the Internet sample, p > .6. The
interaction of MvS and distraction was F(1,100)=.060, p> .8. Similar results came out
for the proportion of negative thoughts. Results are shown in Figure 12.
0
1
2
3
4
5
6
7
no
distraction
distraction no
distraction
distraction no
distraction
distraction
total sample student internet
Negative thoughts generated
massed
spaced
Figure 12. Cognitive responses, experiment 3.
37
Discussion
Not surprisingly, the MvS main effect pattern replicated for the non-distraction
condition. More surprisingly though, the pattern replicated for the distraction
condition. This could have been a result of the distraction manipulation not working,
since the manipulation didn’t seem to affect the number of thoughts generated. It is
also possible though, that the thought listing measure was unreliable, or that subjects
used the thought listing, which was administered after the attitude measure, to justify
their attitude judgment, leading to an inaccurate assessment of how much elaboration
actually took place. This possibility however does not explain the MvS difference
under distraction. Notably, this experiment is not directly comparable to the previous
two experiments. This was the first experiment to have an enforced time limit for the
entire course of reading and considering the arguments against the proposal. In the
first experiment there were timed conditions, but in these, reading of the argument was
done at the subject’s own pace, and the enforced time limit only started once they
clicked the button signaling that they were done reading. It is possible that elaboration
occurred during the reading phase. So in that way, this experiment confirms the MvS
effect under these circumstances. This is not surprising though, since the time taken in
this experiment, even under the enforced time limit, was comparable with that in the
self-paced and timed conditions of the first two experiments.
Experiment 4
The results from the second experiment suggested that spaced persuasion gives
people more opportunity to generate cognitive responses, given that the arguments are
substantive. Since these cognitive responses are predominantly pro-persuasion, this
38
may have explained why persuasion is more effective under the spaced method of
presentation. The distraction experiment was designed to potentially confirm and
extend these findings by showing that distraction reduced the pro-persuasion cognitive
responding, and reduced the difference in attitude change between the massed and
spaced conditions. (However, distraction did not have the hypothesized effect.)
It is conceivable though that processes other than the hypothesized pro-
persuasion cognitive responding might have impacted the results. For example, it is
possible that upon hearing the persuasive message that was counter to the subject’s
held attitude, this might incite a response to maintain their attitude due to a natural,
general motivation to maintain cognitive consistency(Abelson et al., 1968) or knowing
the persuasive intent, it might cause a reactance response and a motivation to preserve
the existing attitude (Petty & Cacioppo, 1979a). If in the massed condition hearing
more information at the same time created a stronger motivation to maintain their
initial attitude, this would result in less attitude change in the massed condition, which
is exactly what was seen in the two preliminary experiments. This resistance
hypothesis is an alternative explanation for the existing pattern of results. It is also
possible that reading the persuasive message might generate a reactance effect,
meaning people are made uncomfortable or are induced to feel negative affect by
being exposed to information counter to their attitude, and they are motivated to
respond in a way that maintains their initial attitude, partly to restore their mood.
These kinds of mood effects have been given credence, especially when the persuasive
message targets viewpoints related to the self-concept (Fuegen & Brehm, 2004). Both
of these resistance explanations might likely manifest themselves as increased
39
counterarguing of the persuasive message. (However, that particular explanation was
not conclusively supported by the data in the second experiment, as the number of
counterarguing (pro-initial attitude, anti-persuasion) statements generated in the
thought listing did not significantly differ across groups.) It is also possible that
negative mood effects on persuasion might not manifest themselves through extra
counterarguing, and instead through a non-systematic route that does not necessarily
involve counterarguing. In this case, an additional manipulation of cognitive
responding might help rule out this possibility, and isolate the relevant effects.
If the cognitive response explanation is correct, there are additional
manipulations that might rule out these and other alternative explanations. For
instance, if the subjects’ dominant cognitive response, instead of elaborating with pro-
persuasion thoughts was instead to counterargue the persuasive message, then
increased elaboration would result in more counterarguing, and less persuasion. Thus
a massed vs. spaced manipulation under a motivated counterarguing manipulation
should show an MvS effect in the opposite direction as the original experiment.
Method
Design
This experiment was a 2 × 2 factorial design. The primary factor was massed
vs. spaced, and the second factor was normal instructions vs. counterarguing
instructions. Need For Cognition was again added as a potential third factor. All
conditions employed self-paced consideration.
40
Participants
The participants in the experiment consisted of 133 students from USC and
CSULB, and 55 from Internet recruitment.
Procedure
This experiment preserved the basic massed versus spaced procedure with
thought listing, and the NFC measure at the end. There was no intervening task
between spaced arguments. The only difference between the normal vs.
counterarguing conditions was the instructions given prior to reading the initial
proposal. In the counterarguing condition, subjects were told to imagine that they had
been hired as a consultant to the project, and it was their job to ensure the proposal is
approved. They were also told specifically that when encountering any information
that might be critical of the proposal, they should think about how to counterargue
that information. This instruction set, especially the focus on countering information
critical of the proposal, was designed to get subjects to counterargue the persuasive
(anti-proposal) arguments (but not the initial proposal), without tipping them off ahead
of time exactly what kind of information they would be receiving and when it would
be received. In addition, they were told that at the end of the experiment, they would
be asked to argue in their own words why the proposal should be passed. This kind of
instruction set has been successfully used in previous message reception studies to
manipulate counterarguing (e.g. Eagly, 1974). This last instruction was designed to
motivate the subject by way of making them accountable at the end of the experiment,
to try and ensure they would be motivated to actually counterargue. Because this
instruction set might have made the subjects in the counterarguing condition feel more
41
accountable than under the regular instructions, subjects in the no-counterarguing
conditions were given an additional instruction as well to ensure the same sense of
accountability without also biasing their cognitive responding. They were told at the
end of the experiment they would be asked to explain their opinion, whatever it was,
and to appropriately pay attention carefully. There were no new measures introduced
in this experiment.
Results
Manipulation check / thought listing
A factorial ANOVA was performed on the coded thought listing data to see if
the counterarguing manipulation was successful. There were no positive results for
counterarguing factor main effects. The closest result was for negative thoughts,
F(1,144)=1.261, p=.26, with positive thoughts (which might be the direct result of
counterarguing) p>.5 and the proportion of negative thoughts (p>.3) nonsignificant.
Breaking it up into the student and Internet subsamples, the Internet sample tended to
produce fewer negative thoughts than the student sample, but the counterarguing
manipulation had virtually no effect on group differences in either subsample (see
Figure 13). Thus on the face of it, the counterarguing manipulation was unsuccessful.
42
-3
-2
-1
0
1
2
3
4
student internet
Negativity towards proposal
(neg - pos thoughts)
no counterargue
counterargue
Figure 13. Cognitive responses, experiment 4.
Time taken
A factorial ANOVA showed there were no significant group differences in the
time taken to read and consider the arguments, all ps > .5.
Attitude Change
Even thought the manipulation check was unsuccessful, it could be that the
measure was just insensitive to the cognitive processes, so the main dependent
variable was run through a factorial ANOVA as well. For the total sample, the MvS
main effect was weakly marginal, F(1,184)=2.125, p=.147. The main effect for
counterarguing was nonsignificant, F(1,184)=.659, p > .4. The interaction was also
nonsignificant, F(1,184)=.637, p > .4. Looking at the student subsample, the MvS
main effect was just significant, F(1,129)=3.910, p=.050, with the counterarguing
43
main effect and MvS-counterarguing interaction nonsignificant, ps > .3. (All results
for the Internet sample were nonsignificant.) Results are presented in Figure 14.
While the MvS simple effects for counterarguing and no counterarguing conditions
did not reach significance, it was surprising that the MvS pairwise comparisons
showed the effect was smaller in the no-counterargue condition than in the
counterargue condition.
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
no CA CA no CA CA no CA CA
total sample student internet
Attitude Change
massed
spaced
Figure 14. Attitude change, experiment 4.
Using Need for Cognition as a covariate to explore the results more, an
ANCOVA did not change the pattern of results at all. Using Need for Cognition as a
continuous independent variable revealed a main effect of Need for Cognition on
attitude change, beta=-.157, p=.036, meaning higher NFC resulted in less attitude
change, controlling for condition. Other effects and interactions were nonsignificant,
44
counterargue * Need for Cognition interaction p > .2 (the regression slopes were still
in the same direction) and all other 2- and 3-way interactions p > .3.
Attitude certainty
A Time 2 attitude certainty ANOVA revealed a significant MvS *
counterargue interaction, F(1,184)=4.280, p = .040. Neither main effect was
significant, ps > .2. It seemed to be the simple main effect of MvS in the
counterarguing condition driving this interaction, F(1,87)=4.541, p = .036, with those
in the spaced counterarguing condition being the least certain about their attitudes.
These results are presented in Figure 15. Certainty at time 1 again failed to show any
effect on attitude change.
4
5
6
7
8
9
10
no counterargue counterargue
Time 2 certainty
massed
spaced
Figure 15. Attitude certainty at time 2, experiment 4.
45
Discussion
As with experiment 3, the manipulation seemed to be largely unsuccessful
here. In the student sample, the MvS effect replicated significantly again. A likely
cause of the failure of the manipulation was that the arguments were strong enough so
that counterarguing was difficult. It seems unlikely that the thought listing measure
was insensitive to counterarguing thoughts, because there were many thoughts listed
that did seem to in fact be counterarguments to the target stimulus materials. Even the
Internet sample, which seemed to be most favorable towards the proposal and hence
likely to be motivated to counterargue the arguments, in addition to being higher on
Need for Cognition and thus more able to counterargue, did not appear to be affected
by the manipulation. And the pattern of attitude change did not support the conclusion
that sufficient counterarguing was taking place to moderate the MvS effect as hoped.
Another interesting feature of the data was that the biggest MvS effect was in the
student counterarguing condition, where it was hypothesized that the effect might
actually reverse itself. Of course this was not seen, and in addition, the MvS effect
here was not accompanied by a cognitive response pattern consistent with the
elaboration hypothesis. This, along with experiments 2 and 3, strengthen the
likelihood of the possibility that the MvS effect can occur in the absence of differential
elaboration. Finally, it is interesting that the attitude certainty was least in the
counterarguing conditions, especially under the spaced condition. This is suggestive
(although not conclusive) along with the persistence of the MvS effect in those
conditions, that the attention to unsuccessful counterarguing produced more attitude
46
change, and less certainty at the same time. This could be said to be something of a
meta-cognitive effect.
Experiment 5
Many of the previous studies have focused on identifying the cognitive
consequences of receiving persuasive information in the massed vs. spaced format.
This experiment has a slightly different focus. The data from the previous
experiments shows that the time taken to consider the arguments was not always
significantly different across massed and spaced conditions. If there is greater
elaboration, but the time taken is not different, then what is the cause of this greater
elaboration? Alternatively, why is it that subjects elaborate less in the massed
condition, even though they have the same amount of time? Generally, a process that
would seem to reasonably explain this is a competition or interference explanation. In
other words, it is possible that trying to think about too many things at the same time,
or rather switching back and forth between different pieces of information leads to
competition for attention or other cognitive resources, and this interferes with the
elaboration process. This explanation would be in line with the theory of message
reception processes mentioned in the introduction, in that reduced attention (which is
one of the two parts of message reception) would lead to reduced attitude change.
Task Switching
In the learning literature, there is an extensive history of research that has
focused on performing multiple task sets where one has to switch from doing one task
to doing a different one. Measures of response time show that there is a cost involved
in switching the mental resources from focusing on the first task to focusing on the
47
second one (Koch, 2001; Meiran, Chorev, & Sapir, 2000; e.g. Spector & Biederman,
1976). Error rates also increase when the task on the current trial is different than the
one on the last trial, versus when the task on the current trial remains the same. (e.g.
Meiran et al., 2000) If it is the case that under massed presentation of the persuasive
arguments, readers switch their attention from analyzing one argument to analyzing
another before they have fully thought through any particular one (and possibly back
again), this task switching cost could limit the depth of their elaboration and ultimately
the degree of persuasion. Thus it is hypothesized that if this type of interference
process is occurring, instructing the individual to think about the massed arguments in
a serial fashion instead of a manner that allows them to move back and forth between
arguments will increase the effectiveness of the persuasion such that it will look more
like the spaced condition.
Method
Design
This experiment did not use a full factorial design. In addition to the two regular
massed and spaced conditions, there was one condition that used massed presentation
of persuasive arguments, but instead of the regular instruction set a serial
consideration instruction set was used. Thus it was a single-factor, three-level design.
Participants
The participant sample was comprised of 148 overall, consisting of 109
students from USC and CSULB, and 39 Internet participants.
48
Procedure
As with the counterarguing experiment, the only modification to the procedure
was in the instructions given at a certain point. In the massed serial condition, when
the persuasive arguments were presented, subjects were told to think about the first
argument on the screen until they felt they were done thinking about it, and then asked
to move on to the next one, and not go back to the first one, etc., until they had
considered all five arguments. The second attitude assessment was followed by the
thought-listing task, and the NFC measure.
Results
Time taken
A one-way ANOVA on the time taken to read and consider the arguments
showed a positive result with spaced taking more time than massed or massed serial,
F(2,145)=3.304, p=.040. Post-hoc analysis revealed the difference between the
massed serial order condition and the spaced condition was the greatest, Tukey p=.05,
a weakly marginal difference between the regular massed and spaced conditions,
p=.114, and a nonsignificant difference between the massed and the serial conditions;
in all these pairwise comparisons, spaced took more time. Looking at the student
sample only, the one-way ANOVA was significant, F(2,106)=3.910, p=.023, but with
the biggest difference between the massed and spaced conditions (more time taken in
spaced), Tukey p=.025, the serial-spaced difference marginal at p=.09 (more time
taken in spaced), and a nonsignificant difference between massed and serial. The
Internet sample did not show any significant differences, F(2,36)=1.246, p > .3.
49
Results are presented in Figure 16. Controlling for condition, Need for Cognition did
not correlate with the time taken, p>.6.
0
50
100
150
200
250
300
350
total sample student internet
Read + Think time (sec)
massed
massed serial
spaced
Figure 16. Time taken, experiment 5.
Attitude change
For the total sample, a one-way ANOVA showed no significant results,
F(2,145)=.296, p > .7, with all post-hoc differences p > .7 as well. Looking at the
student sample, F(2,106)=.892, p > .4, and in the Internet sample, F(2,36)=.327, p >
.7, differences were nonsignificant. These patterns are presented in Figure 17. This
overall result showed, disregarding the serial condition, a puzzling lack of replication
of the basic MvS effect. Using Need for Cognition as a covariate did not improve the
results, and using it as an independent variable did not show any significant main
50
effects or interactions. Using time taken as a covariate and a continuous independent
variable did not produce any positive results either.
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
total sample student internet
Attitude Change
massed
massed serial
spaced
Figure 17. Attitude change, experiment 5.
Thought listing
A one-way ANOVA on the cognitive responses was nonsignificant,
F(2,124)=.648, p > .5, all post-hocs ps > .4. For the student sample, F(2,93)=2.025, p
> .4, all post-hocs ps > .4, and for the Internet sample, F(2,28)=.157, p > .8, all post-
hocs ps > .8. Using Need for Cognition as a covariate and independent variable did
not improve the results.
Discussion
The serial order manipulation did not produce clear results, even just
comparing the serial and spaced conditions. It did not seem to increase the time spent,
51
and did not seem to increase the amount of thinking, looking at the cognitive response
pattern. If anything, it interfered with the amount of thinking. Possible explanations
why the main effect did not replicate point mainly to the massed cell for the student
sample, which had by far the highest attitude change of any massed cell in all 5
experiments. The spaced attitude change was in line with the previous experiments.
The inconsistency and inconclusiveness of the role of time taken continued through
this experiment. Within the student sample, a time difference appeared once again
between massed and spaced, although this time is was not accompanied by a
difference in attitude change.
Further Analyses
Need for Cognition
In all conditions for the experiments where Need for Cognition was measured,
it was negatively related to attitude change when pooled across the student and
Internet samples. In some of the cells, the relation was significant, but in most
individually it was not. But the consistency of the direction was striking. A one-way
ANCOVA with the condition as the fixed factor revealed a significant effect for NFC
while controlling for condition, F(1,509)=12.013, p=.001, eta
2
=.023 (r ~.15). This is a
smallish effect size, but it is interesting that high NFC is related to less attitude change
for these materials. Unfortunately, a closer look at the regression coefficient in
individual conditions did not reveal a consistent interpretable pattern. One example is
in the counterarguing cells. For the student sample in the massed counterargue
condition, the most significant regression coefficient (beta=-.605, undstandardized b=-
.05) shows that high NFC subjects apparently counterargued better (the coefficient
52
meaning that about 20 points in NFC, on a scale that ranges from –54 to +54,
corresponds to a full point less in attitude change). But in the spaced counterargue
condition, NFC was positively and nonsignificantly related to attitude change, so the
relationship was confusingly inconsistent.
Following up on this, the relation between NFC and the time 1 attitude was
explored. Altogether, there was a positive but small correlation between NFC and time
1 attitude, r=.170, p<.001. In all conditions this relationship was positive, although it
did not reach significance in most conditions individually. Having this individual
difference influence the receptiveness to the initial materials was not intentional, and
probably added some variability to the main dependent measures, and reducing power
for the experiments somewhat. This would explain the negative relationship between
NFC and attitude change, in a way consistent with Haugtvedt & Wegener (1994),
showing that people who elaborate more initially tend to change their attitudes less
upon receiving new persuasive information. Additionally, a Need for Cognition *
MvS analysis was run across all conditions, and did not show any interaction between
the two variables.
53
CHAPTER 4: GENERAL DISCUSSION AND FUTURE DIRECTIONS
General Discussion
The massed versus spaced effect appears fairly robust after this series of five
experiments, but at the same time the explanation for it is not at hand. None of the
possible processes proposed in the introduction have been confirmed. Looking at the
experiments collectively though, some tentative speculations can be made, particularly
about the role of elaboration processes. In the second experiment, the thought listing
measure of cognitive responding did successfully mediate the MvS effect for the
substantive arguments. But for the non-substantive arguments in the second
experiment, and with the substantive arguments in the distraction condition of the third
experiment, as well as the counterarguing condition in the fourth experiment the MvS
effect was present without an accompanying pattern of thoughts. The size of the one
indirect effect mentioned above was also rather small, leaving a lot of it unaccounted
for (i.e. the mediation was barely significant, and did not come close to complete
mediation of the main effect). What emerges is the possibility that most of the massed
versus spaced effect does not have to do with elaboration after all. Unfortunately this
leaves me back at the beginning in terms of searching for an explanation.
The elaboration hypothesis has not completely been ruled out though, and
there are some things that suggest it might still be viable. In the first experiment, the
recall measure showed that subjects in the spaced condition recalled more of the target
arguments than in the massed condition. This effect did not mediate the attitude
change. But along with the somewhat unreliable difference in time taken to read and
consider the arguments such that subjects under spaced consideration took more time,
54
it suggests there is some information processing difference present. The distraction
experiment showed the MvS effect even with equal times being strictly enforced
though, but this does not rule out the possibility of decreased attention in the massed
condition as the time increased, relative to the spaced condition where the subjects
were forced to renew their attention focus for each argument presented. The smaller
MvS effect in the no-counterarguing condition of that experiment suggests that maybe
the accountability increased scrutiny to the arguments enough to eliminate the effect,
possibly mediated by the extent of elaboration as well. Future experiments that would
be important to follow up on, considering the apparent unsuccessful manipulations in
the distraction and counterarguing experiments include: 1) An even stronger
distraction (which could be implemented with the same task or a different one)
possibly combined with a short time limit to restrict elaboration should be carried out
to ensure that the failure of the present experiment to moderate the MvS effect was not
just a result of a weak manipulation. (This is said, despite the unsolicited open-ended
thought listing responses from many subjects saying the distraction task made
considering the arguments very difficult.) 2) Rather than use a counterarguing
manipulation to try and reverse the dominant cognitive responses of the subjects,
implement a manipulation using weak (vs. strong) arguments instead. If the pattern of
attitude change showed that those in the spaced condition changed their attitude less
than in the massed condition, it would still support the idea that those in the spaced
condition had more opportunity to elaborate.
It was somewhat surprising that Need for Cognition correlated with less
attitude change overall, even though the arguments appeared to effectively induce
55
more negative thoughts than positive ones. This might be interpreted as counter to the
elaboration hypothesis, but it is hard to tell since the negative arguments were
themselves somewhat of counterarguments to the initial proposal. Thus the arguments
could be seen as inducing doubt in the initial attitude rather than causing persuasion
per se. The time 2 mean attitudes for the student sample were approximately at the
neutral point of the scale, meaning the subjects didn’t think that the negative
arguments were so strong as to definitely be against the proposal. The initial purpose
of picking this set of materials was to give subjects the ability to resist the persuasion
in the first place, in the case that the massed argument format might give them enough
information to overcome that resistance. This possibility has not been confirmed with
this set of materials though.
Another interesting possible explanation for the massed versus spaced effect
that was not explored in the present set of experiments is that the effect is not caused
by careful processing of the content of the arguments, but instead it is a somewhat
heuristic process caused by a ‘feeling’ of being more persuaded, using other cues to
inform this disposition. This is somewhat supported by the result with the non-
substantive arguments in the second experiment and the results under distraction (even
though the effectiveness of that manipulation was questionable). The lack of any
convincing effect for Need for Cognition does not shed light on this either way,
because it is conceivable that it could have eliminated the effect or enhanced it under
the elaboration hypothesis. However, presuming that high NFC subjects would be
thinking about the arguments more thoroughly, any cue-based effect should have been
weakened relative to the impact careful consideration of argument content had, and
56
thus it might be expected that the massed versus spaced effect might have been
attenuated by high NFC if it were enabled through a peripheral attitude change route.
The data did not support this hypothesis. The heuristic hypothesis is consistent with
the attenuated MvS effect under the accountability manipulation in the counterarguing
experiment though, but this is also inconclusive.
Future experiments that could be run to address the heuristic hypothesis would
need to identify what exactly the heuristic is that people are using to inform their
opinions. A direct accountability/involvement manipulation, possibly using a different
set of materials for which the subject population could be highly involved (for
example the well-tested senior comprehensive exams issue paradigm, (e.g. Petty,
Harkins, & Williams, 1980)) might be useful for this purpose, for which involvement
would eliminate the MvS effect if it were heuristic and the arguments were effective.
This set of materials also includes weak arguments that could be used, possibly under
distraction or low involvement, that might shed light on the heuristic hypothesis as
well; if weak arguments still yielded the effect, it would support the hypothesis.
Finally, in order to identify the heuristic people are using, additional measures
would need to be included. It is possible people feel they have received more
information, when their attention is focused five separate times rather than just once
on the persuasive information, and this mediates their attitude change. This type of
meta-observation as a heuristic has been shown recently to be utilized by people more
than previously realized (e.g. Petty & Briñol, 2008). A measure of perceived
knowledgeablilty has been used successfully to predict the amount of attitude change,
in the context of manipulating the previous amount of knowledge given, before the
57
target persuasive information (Tormala & Petty, 2007). This suggests that mechanism
could be present here too. It could be perceived knowledgeablilty, or instead the
number of arguments (which has also been shown to be a utilized heuristic (Petty &
Cacioppo, 1984)) if the presentation format makes it seem like the number of
arguments is greater. It is also possible that an effect due to number of arguments
could be mediated by knowledgeability.
Special notice should be paid to the possibility that time taken could mediate
the effect or be used as a heuristic. This mediating path was not confirmed in any of
the experiments. Nor does it seem likely, since the MvS effect was seen in several
cases where time limits were enforced and/or no additional time was taken. This
would have supported an explanation that somehow allowing time to “absorb” each
argument was necessary for it to have maximum effect. The current set of
experiments fell short of a sure test of this mechanism, only measuring time taken to
both read and consider each argument. A more direct test of this idea would need to
control and enforce the time subjects were exposed to each argument, possibly by
changing the modality so that the arguments were heard and not read. Then the time
allowed for cognitive responding in between each argument could be self-paced, or
controlled and manipulated, to see how time really might affect the degree of
persuasion if at all.
Related to this idea, it is possible that the massed presentation did not allow
enough time for subjects to absorb the information in each argument, even if it were
only at a superficial level (that might still produce the effect even if it is caused by the
heuristic mechanism mentioned above). This might be a cognitive overload
58
explanation, which is slightly different than the distraction manipulation used in the
third experiment. Whereas a distraction could be said to require performing a
completely different task, cognitive overload might require nothing more than the
target task, but require too much processing in the allotted time, such that the impact
of each individual piece of information was reduced. In the attitude change literature,
this distinction between distraction and cognitive overload has rarely been made.
Because the MvS effect is seen in self-paced conditions though, it is hard to imagine
the source of the effect is that people physically aren’t able to process the required
amount of information to their own satisfaction. So the current data seems more to
rule this out than encourage this exploration.
Further speculation leaves different possibilities to explain the effect. It is
possible that knowing the intent is to persuade, subjects lose interest more quickly
because they know what to expect, and so information processing drops off with
successive arguments in the massed condition. This could be tested by controlling for
argument order (which was completely randomized in all the experiments reported
here), and coding cognitive responses to see which arguments were processed most
thoroughly, or counterbalancing the orders and doing a serial position weight analysis
to find out which argument positions are most influential. Also, manipulating
knowledge of intent to persuade might moderate the effect, such that people
encountering evaluative-laden information in a context without the explicit intent to
influence their attitude might in fact yield no effect or even a flipped effect.
Finally, attention mechanisms might be a likely place to look as the source of
the MvS effect. In the attitude change literature, attention has not been studied much
59
beyond the time taken to attend to messages, and subsequent memory for them
(Wegener & Carlston, 2005). These quantities did not seem to be especially related to
attitude change in this context. One way to more precisely examine attention is by
measuring gaze fixation, but this has not been used yet in persuasion research, and
would not apply to messages delivered in audio format. It is clear that decreased
attention leads to decreased persuasion for these arguments as shown in the third
experiment, but based on the current data this does not seem to drive the MvS effect.
Limitations
It is important to note that while the current experiments represent one set of
circumstances with a robust effect that spaced persuasive arguments were more
effective than massed, this is only one possible instantiation of the MvS
operationalization. This set of materials used fairly rational arguments, in a very
cognitively-oriented task, under attention clearly focused on the task. It is not yet
known what would happen in circumstances where subjects weren’t explicitly told
what they should pay attention to. In this case, a massed set of arguments or attributes
might capture an individual’s attention more, leading to a stronger evaluation,
especially if attitude assessment were delayed. Under delayed assessment, one is
more likely to base their evaluative judgment on recalled information, so anything that
enhances attention and recall might yield more attitude change. There are also
attitudes that are based not on substantive attributes, but affective reactions, and these
attitudes might respond differently to the manipulation. If as speculated above, people
are using a knowledgeability heuristic to inform their attitudes and this drives the
effect seen in these experiments, then it might be that attitudes based on affective
60
responses rather than calculation based on cognitive arguments would not show the
same effect. Due to the nature of the MvS manipulation, it seems more applicable to
environments where people are gathering information, which is more easily
encountered in a massed or spaced presentation format; attitudes based on things like
source attractiveness or expertise are more often singular pieces of information and
less likely to form a series of data points on which to base an attitude.
61
CHAPTER 5: CONCLUSION
The results of several of these experiments have been disappointing and
yielded little information, but the fact remains there is a robust massed versus spaced
effect. According to McGuire’s (1985) information processing model, the critical
relevant steps at which this effect could take place are in attention, comprehension,
cognitive responding, and yielding(agreement). Within cognitive responding, the
effect could utilize message-based thoughts, or peripheral thoughts (possibly
metacognitions). The former has not received strong support, and so future work will
focus more on the latter. Yielding is most closely tied to the message content and
other motivations to agree or disagree that were not manipulated here, and so should
not be of primary interest. This leaves possibilities for attentional processes and
comprehension. Measures for these processes are underdeveloped in the context of
attitude change, and might need to be developed to account for the effect if the
heuristic explanation proves insufficient.
The focus of these experiments has been more applied than basic in that the
phenomenon is a practical one; what is being manipulated is not a known fundamental
process, but an ostensibly surface feature of a persuasion attempt. Thus the
explanation might be more complicated; there might in fact be effects from multiple
fundamental processes (the various information processing steps from above) that
could either complement each other or compete with each other. The current set of
experiments might favor only a subset of these processes. It is very possible that with
different kinds of persuasive information, and under different circumstances, including
62
over different time spreads, different processes will be more influential and massed
presentation might still be more effective.
63
References
Abelson, R. P., Aronson, E., McGuire, W. J., Newcomb, T. M., Rosenberg, M. J., &
Tannenbaum, P. H. (Eds.). (1968). Theories of cognitive consistency: A
sourcebook. Chicago: Rand McNally.
Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of Personality
and Social Psychology, 42(1), 116-131.
Cacioppo, J. T., Petty, R. E., & Kao, C. F. (1984). The efficient assessment of need for
cognition. Journal of Personality Assessment, 48(3), 306-307.
Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed
practice in verbal recall tasks: A review and quantitative synthesis. Psychological
Bulletin, 132(3), 354-380.
Chen, S., & Chaiken, S. (1999). The heuristic-systematic model in its broader context.
In S. Chaiken, & Y. Trope (Eds.), Dual-process theories in social psychology.
(pp. 73-96). New York, NY, US: Guilford Press.
Eagly, A. H. (1974). Comprehensibility of persuasive arguments as a determinant of
opinion change. Journal of Personality and Social Psychology, 29(6), 758-773.
Eiser, J. R. (1994). Toward a dynamic conception of attitude consistency and change.
In R. R. Vallacher, & A. Nowak (Eds.), Dynamical systems in social psychology.
(pp. 197-218). San Diego, CA, US: Academic Press.
Fuegen, K., & Brehm, J. W. (. (2004). The intensity of affect and resistance to social
influence. Mahwah, NJ, US: Lawrence Erlbaum Associates Publishers.
Greenwald, A. G. (1968). Cognitive learning, cognitive response to persuasion, and
attitude change. In A. G. Greenwald, T. C. Brock & T. M. Ostrom (Eds.),
Psychological foundations of attitudes (pp. 147-170). San Diego, CA: Academic
Press.
Haugtvedt, C. P., & Wegener, D. T. (1994). Message order effects in persuasion: An
attitude strength perspective. Journal of Consumer Research, 21(1), 205-218.
Insko, C. A. (1964). Primacy versus recency in persuasion as a function of the timing
of arguments and measures. Journal of Abnormal & Social Psychology, 69(4),
381-391.
64
Koch, I. (2001). Automatic and intentional activation of task sets. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 27(6), 1474-1486.
McGuire, W. J. (1972). Attitude change: The information-processing paradigm. In C.
G. McClintock (Ed.), Experimental social psychology (pp. 108-141). New York:
Holt, Rinehart, & Winston.
McGuire, W. J. (1985). Attitudes and attitude change. In G. Lindzey, & E. Aronson
(Eds.), Handbook of social psychology (3rd ed., pp. 233-346). New York:
Random House.
Meiran, N., Chorev, Z., & Sapir, A. (2000). Component processes in task switching.
Cognitive Psychology, 41(3), 211-253.
Miller, N., & Colman, D. E. (1981). Methodological issues in analyzing the cognitive
mediation of persuasion. In R. E. Petty, T. M. Ostrom & T. C. Brock (Eds.),
Cognitive responses in persuasion (pp. 105-126). Hillsdale, NJ: Earlbaum.
Miller, N., & Campbell, D. T. (1959). Recency and primacy in persuasion as a
function of the timing of speeches and measurements. Journal of Abnormal &
Social Psychology, 59(1), 1-9.
Nowak, A., & Lewenstein, M. (1994). Dynamical systems: A tool for social
psychology? In R. R. Vallacher, & A. Nowak (Eds.), Dynamical systems in social
psychology. (pp. 17-53). San Diego, CA, US: Academic Press.
Osterhouse, R. A., & Brock, T. C. (1970). Distraction increases yielding to
propaganda by inhibiting counterarguing. Journal of Personality and Social
Psychology, 15(4), 344-358.
Petty, R. E., & Briñol, P. (2008). Persuasion: From single to multiple to metacognitive
processes. Perspectives on Psychological Science, 3(2), 137-147.
Petty, R. E., & Brock, T. C. (1981). Thought disruption and persuasion: Assessing the
validity of attitude change experiments. In R. E. Petty, T. M. Ostrom & T. C.
Brock (Eds.), Cognitive responses in persuasion (pp. 55-80). Hillsdale, NJ:
Earlbaum.
Petty, R. E., & Cacioppo, J. T. (1979a). Effects of forewarning of persuasive intent
and involvement on cognitive responses and persuasion. Personality and Social
Psychology Bulletin, 5(2), 173-176.
Petty, R. E., & Cacioppo, J. T. (1979b). Issue involvement can increase or decrease
persuasion by enhancing message-relevant cognitive responses. Journal of
Personality and Social Psychology, 37(10), 1915-1926.
65
Petty, R. E., & Cacioppo, J. T. (1984). The effects of involvement on responses to
argument quantity and quality: Central and peripheral routes to persuasion.
Journal of Personality and Social Psychology, 46(1), 69-81.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of
persuasion. Advances in experimental social psychology (pp. 123-205). San
Diego, CA: Academic Press.
Petty, R. E., Harkins, S. G., & Williams, K. D. (1980). The effects of group diffusion
of cognitive effort on attitudes: An information-processing view. Journal of
Personality and Social Psychology, 38(1), 81-92.
Petty, R. E., Ostrom, T. M., & Brock, T. C. (Eds.). (1981). Cognitive responses in
persuasion. Hillsdale, NJ: Earlbaum.
Petty, R. E., Wells, G. L., & Brock, T. C. (1976). Distraction can enhance or reduce
yielding to propaganda: Thought disruption versus effort justification. Journal of
Personality and Social Psychology, 34(5), 874-884.
Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating
indirect effects in simple mediation models. Behavior Research Methods,
Instruments & Computers. Special Issue: Web-Based Archive of Norms, Stimuli,
and Data: Part 2, 36(4), 717-731.
Read, S. J., Vanman, E. J., & Miller, L. C. (1997). Connectionism, parallel constraint
satisfaction processes, and gestalt principles: (re)introducing cognitive dynamics
to social psychology. Personality and Social Psychology Review, 1(1), 26-53.
Spector, A., & Biederman, I. (1976). Mental set and mental shift revisited. American
Journal of Psychology, 89(4), 669-679.
Tormala, Z. L., & Petty, R. E. (2007). Contextual contrast and perceived knowledge:
Exploring the implications for persuasion. Journal of Experimental Social
Psychology, 43(1), 17-30.
Watts, W. A., & McGuire, W. J. (1964). Persistence of induced opinion change and
retention of the inducing message contents. Journal of Abnormal & Social
Psychology, 68(3), 233-241.
Wegener, D. T. (., & Carlston, D. E. (. (2005). Cognitive processes in attitude
formation and change. Mahwah, NJ, US: Lawrence Erlbaum Associates
Publishers
66
Appendix A
Main Experimental Scenario
In this experiment, you are being asked to consider an episode involving a city council
in a small suburb of a large metropolitan city.
Recently, a proposal has been floated to take some lots on the corner of the main
commercial street in town, and change the zoning on those lots so they can be used for
multi-story apartments and condominiums (instead of just houses or small stores).
There are currently none of these multi-family dwellings in the city limits, nor have
there ever been any in the past.
This is the message introducing and advocating the proposal, which has been looked
over in a committee already and approved to be read by the whole council. It was
drafted by the council member who heads that committee and believes the proposal is
a positive thing.
Allowing multi-family dwellings in our town is an important opportunity. The project
on Grant Avenue will allow us to grow our tax base without eating up more land or
open space. Because more people are living on less land, it will increase property tax
revenues, and it will drive commerce in our town center with its beloved "Main Street"
feel, and create a significant destination for people from a wide area in the surrounding
towns to visit and shop and dine. This will in turn increase the land value for the
commercial properties, which increases property taxes more, and it will also drive
sales tax revenue. The result is all these taxes will help fund services we need like
police, emergency services, and schools, all without raising local taxes. Having this
compact development also creates affordable housing options for people who can't
afford the increasing price of homes in the town (which many people have complained
about recently.) Allowing condominiums allows these people to buy homes and build
equity in the homes through their mortgage payments so they can use the money later
and improve their economic situation in general. Because we don't have to take away
open space, the infrastructure we need to support the development (like roads, sewers,
and communications lines) is already there, and doesn't require additional incremental
maintenance costs. In other words, the more miles of roads and sewers we have, the
more it costs to maintain them, and so by doing this we avoid these costs. Having the
new housing next to the town center, with its train station that goes into the city, will
allow those who choose it to take the public transportation and avoid traffic and
wasted time, which will attract young professionals (as studies have shown convenient
commuting options do) which makes our community stronger. And because less
people will be driving, it literally reduces the air pollution levels on our local streets,
which for example has been shown to have an effect on childhood asthma prevalence
and overall lung development. Because the housing is close to the shopping and
dining, people will walk more, instead of driving, which promotes a healthier lifestyle
(for example, reducing obesity.) In summary, we need to focus on creating livable,
sustainable communities rather than growing out, we believe we are situated to take
67
advantage of the development and the benefits that we see coming from it, and for
these reasons the zoning changes are encouraged.
68
Appendix B
Persuasive Arguments
Substantive Arguments:
(In arbitrary order)
Multi-family housing, according to FBI statistics, is universally associated with higher
crime rates because poorer people live there. We cannot deny the effect that lack of
economic success has on crime. This is something that all communities have had to
prepare for when they have introduced this kind of housing. Many residents believe
this will require a large increase in police funding.
If we go forward with the development, it might very well cause "gentrification",
which means that commercial property values will go up so much that high-end chain
merchants will move in and businesses that have been here for years (and are owned
by members of the community) will get squeezed out because of high rents. The same
thing can happen with the residential property values.
The builders of the aforementioned housing developments (the apartments and
condos) are asking for generous financial incentives in the form of handouts and tax
deferments that are a ridiculous way to spend our town's budget, especially when we
already don't have enough money to cover the services we need. The incentives are in
excess of five million dollars in handouts to already rich companies. Several different
developers have all said that they won't build there without the incentives, and if no
one builds, we'll have ugly vacant lots sitting there for years, which will be an eyesore
and detract from the community, possibly hurting commerce as well.
The increased traffic from all the new residents, not to mention the shoppers and
diners the Grant Avenue project will attract will spread onto the residential streets.
This will turn quiet safe streets where people can sit outside on their porches and
where kids can play, things that we all love and are accustomed to, into noisy
dangerous ones.
We would have to upgrade the sewer system because our town pipes were put in a
long time ago and were not designed for such a high volume of water in a confined
geographical area. This would involve tearing up Parker Street, one of our main
streets, which also crosses Grant Avenue. It would take at least nine months, and
possibly up to 18 months, while the work is done and the system is upgraded. Not to
mention the expense involved, which is estimated at several hundred thousand dollars
because the pipes have to be replaced all the way to the county main trunk line (that
runs to the treatment plant), which is over a mile away.
69
Appendix B
(continued)
Non-substantive Arguments:
(In arbitrary order)
The concerned citizens have enlisted a university professor who is an expert in urban
planning and demographic changes. He has come out against the planned changes,
saying they are a bad idea.
Despite the sponsor's (another city council member's) enthusiasm for the proposal, all
the local newspapers have come out with editorials opposing the plan. Not one has
come out supporting it.
The proposal sponsor is not well-liked by the general public; his 'popularity' rating is
the lowest of any current city council member, at 31%. This is backed up by other
council members' opinions of him, saying he is extremely hard to get to know, much
less work with.
The concerned citizens committee has a list of 37 points that lay out all their
objections to the proposal, which they say far exceed any benefits that might come
from it. This document would be provided upon request later during the full hearing.
One of the council members has noted that projects of this scope rarely go smoothly,
and even when trying to anticipate some 'bumps in the road', it's likely that unforeseen
problems will arise, and these projects take much longer than expected to finish, and
cost a lot more money than anticipated.
70
Appendix C
Attitude Measure Items
1. Based purely on its merits, how much would you be in favor of going forward with
this proposal if you were on the council?
(continuous from 0 to 10;
low point: not at all;
midpoint: neutral;
high point: extremely in favor)
2. Based on the proposal's merits, how good of an idea is this for the town?
(continuous from 0 to 10;
low point: not good at all;
midpoint: neutral;
high point: it’s a great idea)
3. How certain do you feel about your opinion?
(continuous from 0 to 10;
low point: not at all;
midpoint: somewhat;
high point: extremely certain)
Abstract (if available)
Abstract
A manipulation of persuasive messages was explored to see what would result in more attitude change, such that two formats of presenting a series of arguments were used: massed and spaced. In massed presentation, five arguments were presented at once, and thought about all at the same time
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The angry brain: neural correlates of interpersonal provocation, directed rumination, trait direct aggression, and trait displaced aggression
PDF
Cueing compliance and conferring context: testing the group-as-cue and group-as-conveyor models of social influence.
PDF
Neural correlates of learning and reversal of approach versus withdraw responses to in- and out-group individuals
PDF
Neurobiological correlates of skilled and disabled reading
PDF
The role of framing in health social comparisons by multiple sclerosis patients
PDF
Temporal dynamics of perceptual decision
PDF
Social identification and the treatment of in-group deviants: the black sheep effect or the devil protection effect
PDF
“Under the radar”: the relative effectiveness of gay-vague, gay-explicit, and control anti-smoking messages for targeted and nontargeted audiences
PDF
Investigating the effects of personality factors on goal shielding
PDF
The impact of statistical method choice: evaluation of the SANO randomized clinical trial using two non-traditional statistical methods
PDF
Exploring the effects of media on the self-concepts and achievement of African American high school students
PDF
The neural representation of faces
PDF
The effects of mortality salience and perceived terrorism threat on attitudes toward Muslim and Muslim Americans
PDF
Effects of nicotine abstinence on orienting, executive function, arousal and vigilance
PDF
Cognitive-linguistic factors and brain morphology predict individual differences in form-sound association learning: two samples from English-speaking and Chinese-speaking university students
PDF
Implicit theory ideology about human attributes and aggression
PDF
Projection on outgroups: complementary projection in aversive intergroup contexts
PDF
Object adaptation in the lateral occipital complex: shape or semantics?
PDF
Characterizing the perceptual performance of a human obseerver as a function of external noise and task difficulty
PDF
Adaptation in fusiform face area (FFA): image or person?
Asset Metadata
Creator
Monroe, Brian M.
(author)
Core Title
The effect of massed versus spaced arguments on persuasion
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Psychology
Publication Date
07/22/2008
Defense Date
06/09/2008
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
attitudes,OAI-PMH Harvest,persuasion,spacing effects
Language
English
Advisor
Read, Stephen J. (
committee chair
), Lu, Zhong-Lin (
committee member
), Miller, Norman (
committee member
), Priester, Joseph (
committee member
)
Creator Email
monroe@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m1381
Unique identifier
UC177999
Identifier
etd-Monroe-20080722 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-193792 (legacy record id),usctheses-m1381 (legacy record id)
Legacy Identifier
etd-Monroe-20080722.pdf
Dmrecord
193792
Document Type
Dissertation
Rights
Monroe, Brian M.
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
attitudes
persuasion
spacing effects