SURVEYS AS COMMUNICATION:
IT’S NOT JUST WHAT YOU ASK, BUT HOW YOU ASK IT
By
Geneviève Risner

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Communication
2011

ABSTRACT
SURVEYS AS COMMUNICATION:
IT’S NOT JUST WHAT YOU ASK, BUT HOW YOU ASK IT
By
Geneviève Risner
As survey rates decline, survey researchers must consider the factors leading to
reductions in response rates, especially the respondent’s survey-taking experience. This research
sought to study the effects of question type and involvement on a number of outcomes of survey
participation. Respondents participated in an online survey experiment about one of four issues
(universal health care, same-sex marriage, copyrighted music file-sharing, or space exploration)
in one of two question formats (open- or closed-ended). Regression analysis was used to test the
effect of these variables on time to complete the survey, feelings of self-expression, fatigue,
survey enjoyment, willingness to endorse a related campaign, and willingness to participate in a
future survey. Results demonstrate that using open-ended survey questions may increase the selfexpression of the respondent. This increase in self-expression may lead to more survey
enjoyment and a greater willingness to participate in future surveys. This study also found that
asking open-ended survey questions may interact with self-expression to increase the willingness
of the respondent to endorse a campaign related to the topic covered in the survey. While neither
outcome- nor value-relevant involvement appeared to have strong effects on variables of interest,
topic appeared to have an important role. Implications of these findings and considerations for
future research are discussed.

This dissertation is dedicated to the two most important men in my life - my husband, Shane, and
my dad. To Shane, thank you for encouraging and allowing me to fulfill my dreams, bringing
laughter into my life when I most needed it, giving me the gift of time to work on this research,
and being my very best friend. To my dad, thank you for teaching me three important life lessons
that allowed me to write this dissertation: 1) anything is possible with a good plan, 2) people
need to be heard, and 3) when you fall down, get up, and walk it off. I love you both.

iii

ACKNOWLEDGMENTS

During my graduate school experience, I have been blessed to be surrounded by talented,
bright, caring people who have shaped how I see the world. My sincere gratitude goes to Dr. Tim
Levine for sharing countless hours not only advising but mentoring. You have taught me
important lessons about research, teaching, and life for which I am eternally grateful. I am in awe
of your ability to balance brilliance with friendliness, science with creativity, and your talent for
seeing any given issue from a number of perspectives. It has been my honor to be your advisee.
Thank you to my guidance committee members for sharing your time, support, positivity,
and expertise. To Dr. Dan Bergan, thank you for teaching me about field experiments, supporting
my ideas, showing me how to turn them into scientific studies, and providing me with countless
learning opportunities. To Dr. Hee Sun Park, thank you for helping me learn statistics, being
supportive along the way, and always finding new ways to provide teachable moments. To Dr.
Stan Kaplowitz, thank you for designing a wonderful course on social attitudes with valuable
assignments, articles, books, and discussions; the ideas we discussed in your class led to the
creation of this study.

iv

TABLE OF CONTENTS

LIST OF TABLES

vii

LIST OF FIGURES

viii

INTRODUCTION
CHAPTER 1
BACKGROUND

1
2

CHAPTER 2
TOWARD A THEORY OF SURVEY TAKING OUTCOMES
Open- Versus Closed-ended Questions
Time Spent Answering Questions
Self-expression and Value Expressive Attitudes
Involvement
Fatigue
Survey Enjoyment
Intent to Participate in Future Surveys

4
4
5
6
11
19
21
22

CHAPTER 3
MODEL

31

CHAPTER 4
METHOD
Sample
Procedure
Measures

24
25
25
27

CHAPTER 5
RESULTS
Pilot Test
Main Study

33
33
34

CHAPTER 6
DISCUSSION

39

CHAPTER 7
CONCLUSION

44

APPENDICES
Appendix A Manipulation Questions for Closed-ended Surveys by Issue

59
59

v

Appendix B
Appendix C
Appendix D
Appendix E
Appendix F
Appendix G
Appendix H

Outcome-relevant Involvement Scale
Value-relevant Involvement Scale
Piper Fatigue Scale
Self-expression Scale
Enjoyment Scale
Knowledge Scale
Writing Apprehension Scale

REFERENCES

69
70
71
72
73
74
75
77

vi

LIST OF TABLES

Table 1 Descriptive Statistics for Continuous Variables by Condition

46

Table 2 Descriptive Statistics for Categorical Variables by Condition

47

Table 3 Correlations Among Variables

48

Table 4 Means, Standard Deviations, and Mean Differences for Topics used in Pilot Test

49

Table 5 OLS Regression Results for Effect of Survey Type on Time Spent Answering
Survey

50

Table 6 OLS Regression Results for Effects of Survey Type, Time, and Value- and Outcomerelevant Involvement on Self-expression
51
Table 7 Logistic Regression Results for Effect Survey Type and Self-expression on Willingness
to Sign-on to a Campaign
52
Table 8 OLS Regression Results for Effects of Time and Self-expression on Fatigue

53

Table 9 OLS Regression Results for Effects of Fatigue and Self-expression on Enjoyment

54

Table 10 OLS Regression Results for Effect of Enjoyment on Intent to Participate in Future
Surveys

55

vii

LIST OF FIGURES

Figure 1 Path Model Predictions for Effects of Question Type on Survey-taking Outcomes

56

Figure 2 Mean Scores for Outcome- and Value-relevant Involvement by Pilot Test Topic

57

viii

Introduction
“For more than 75 years, sample surveys have remained a remarkably useful and efficient
tool for learning about people’s opinions and behaviors” (Dillman, Smyth & Christian, 2009, pg.
1). Organizations of all types use surveys to create programs, policies, and communication
campaigns, as well as develop business, fundraising, or customer service strategies, to name only
a few purposes. Some researchers, such as George Gallup, have viewed surveys as a forum to
facilitate “scientific democracy,” while others have proposed that survey results have the
potential to shape politician’s policy stances on social issues (Hogan, 1997). While surveys serve
a critical role in generating policy decisions and new knowledge, today’s survey researcher faces
a problematic era of declining survey rates. Potential respondents are more willing than ever to
decline to participate, and social norms reinforce that it is perfectly acceptable to say no
(Dillman, Smyth & Christian, 2009). As a result, collecting data to solve critical issues becomes
increasingly more difficult.
One approach to solving this problem is to gain a better understanding of respondents’
experience during the survey process. Schwarz (1999) explains that survey participation is a
communication process that relies on norms of conversational conduct. If we look at surveys in
this way, as a two-way process of communication between the survey sponsor and the
respondent, the understanding of surveys extends beyond seeing a survey as a mere data
collection method. In fact, it is likely, under some conditions, the very act of participating in a
survey may transform the respondent’s attitudes and, in some cases, actual behaviors (Morwitz,
2005). If researchers ask questions in ways and about topics that elicit an unfavorable response,
the respondent may come to associate unfavorable feelings with survey participation. Therefore,
1

the objective of this study is to improve the understanding of the respondent’s survey experience
and determine the attitudinal and other related outcomes that arise during the survey process. The
premise is that improving respondents’ survey-taking experience will lead to desirable outcomes
including increased participation rates. Yet, in order to improve the survey-taking experience, we
must first understand the relationships among various factors that may lead to survey enjoyment.
Background
Surveys of the public have been the source of extensive examination and scrutiny. One
line of literature, primarily in political science, suggests the lack of stable attitudes among
members of the electorate (Converse, 1964). In addition, some scholars have concluded that
during surveys, respondents rely on the most salient considerations available to them at the time,
and use these considerations to express an attitude (Zaller & Feldman, 1992). As a result, surveys
of the public may not measure true, well-formulated, stable attitudes, but rather the changing
considerations accessible to people at the time of the survey (Zaller & Feldman, 1992).
The marketing literature on the mere measurement effect builds on the notion that
attitudes may not exist at the time of the survey, finding that the act of participating in a survey
(or participation in repeated surveys over time) can alter the thoughts and behavior of the
respondent (Morwitz, 2005). According to the mere measurement effect, the act of survey
participation may cause respondents to form judgments that may not have evolved in the absence
of a survey (Dholakia & Morwitz, 2002; Feldman & Lynch, 1988). The mere measurement
effect has been shown to increase purchase intentions and brand loyalty (Borle, Dholakia, Singh
& Westbrook, 2007; Dholakia & Morwitz, 2002; Fitzsimons & Morwitz, 1996), increase voting
behavior (Greenwald, Carnot, Beach & Young, 1987), and increase health behaviors such as
2

blood donations, registration in blood drives (Godin, Sheeran, Conner & Germain, 2008), dental
flossing, and consumption of non-fatty foods (Levav & Fitzsimons, 2006).
From the social psychology and survey methodology literature, the effects of the
questions themselves are studied. Scholars have focused on question wording, question order,
and context effects (Tourangeau, Rips & Rasinski, 2000; Schwarz, 1996, 1999; Schuman &
Presser, 1981). Additionally, the value of open- compared to closed-ended questions has
received considerable attention. However, when scholars compare open- and closed-ended
questions, they typically assess the quality and validity of the responses acquired (Ivis, Bondy &
Adlaf, 1997; Kealy & Turner, 1993; Schuman, 1966), length (Dillman, Sinclair & Clark, 1993;
Herzog & Bachman, 1981), or expense (Dillman, 1978).
Given the potential for surveys to shape respondent’s attitudes and subsequent behaviors,
it seems critical to understand the role of question types used in surveys on other attitudes. Of
particular interest is the use of open- versus closed-ended question types. Typically, researchers
consider the use of open- and closed-ended questions based on their research question, and the
cost and time required to collect, code, and analyze the data. When we think of using question
types in this way, based on our resources and research needs, the respondent becomes a means to
an end - a mere data conduit. However, as Schwarz (1999) describes, respondents rely on norms
of conversational conduct when participating in a survey, making the survey process akin to a
two-way form of communication. If we think of surveys as a channel through which a survey
sponsor communicates with the respondent, and vice versa, it becomes important to ask, what are
the attitudinal and other related outcomes that arise from using open- versus closed-ended
questions? This research attempts to answer this question.
3

Specifically, this research attempts to demonstrate how question type is related to time
spent answering survey questions, feelings of fatigue, feelings of self-expression, enjoyment of
survey participation, subsequent behavior, and intent to participate in future surveys. It is also
argued that involvement with the topic plays an important role in the time spent participating and
self-expression. A new model with familiar variables is proposed to explain outcomes of
participating in open- compared to closed-ended surveys.
Toward a Theory of Survey-taking Outcomes
Open- Versus Closed-ended Questions
Open- and closed-ended questions are different in many regards. Closed-ended questions
provide a set of potential options from which the respondent may choose. These question types
elicit the potential for a host of negative effects including satisficing, question order, and context
effects (Tourangeau, Rips & Raskinski, 2000). However, survey researchers have treated closedended questions as the preferred choice (Greer, 1991). Data from closed-ended questions are
easier to collect, code, and analyze compared to open-ended questions, making this question type
the favored choice among survey researchers (Greer, 1991 citing Schuman & Presser, 1981).
Although more cumbersome than closed-ended questions, open-ended questions provide
numerous benefits. Most importantly, open-ended questions provide more information compared
to their closed-ended counterparts. In answering an open-ended question, the respondent is not
restricted to a response by the options provided. Rather, the respondent can answer the question
according to his/her interpretation of the question. In this way, open-ended questions do not
suffer restriction in range problems that may be present in closed-ended questions (Tourangeau,
Rips & Rasinski, 2000). Also, open-ended questions do not present opportunities for the
4

respondent to be biased by the order of response options, as these options are not provided (see
Schwarz, Hippler, and Noelle-Neuman 1992, and Schuman and Presser, 1981 for additional
discussion of response order effects).
Also, open-ended questions allow the respondent to answer questions in his/her own
words – providing him/her a “voice” in the survey process (Schuman, 1966). As a result, openended questions have been used when the respondent and surveyor are especially different in
“education, cultural characteristics, or life chances” (Schuman, 1966, pg. 218). Thus, open-ended
questions allow the respondent to express his/her thoughts, feelings, or attitudes in his/her own
way – unlimited by predetermined language and categories created by the investigator.
Time Spent Answering Questions
When answering open-ended questions, the respondent must read and consider the
question and its meaning, then formulate his/her own response, and write this response onto the
survey. Conversely, with a closed-ended question, the respondent is provided with the response
options. In this case, the respondent still needs to read and consider the question and its meaning,
but rather than generate a response, he/she chooses from a set of pre-selected options. Usually,
the respondent need only check a box or circle a response – not write down the entire answer in
his/her own words.
Responding to closed-ended questions is easier for the respondent than responding to
open-ended questions (Mitchell & Carson, 1989). Responding to open-ended questions requires
more effort on the part of the respondent than responding to closed-ended questions. Open-ended
questions require the respondent to generate a response, unassisted by pre-determined categories
provided by the researcher, and transfer this response onto the survey in his/her own words.
5

Thus, responding to open-ended questions should take more time than responding to
closed-ended questions (H1).
Self-expression and Value Expressive Attitudes
Even though open-ended questions may take more time to complete, the respondent may
find value in sharing his/her thoughts in his/her own words. Sjoberg (1954) studied public
opinion about participating in surveys among people from high and low socioeconomic areas and
found that not only did respondents believe people should be asked their opinions but, on
average, people like to be interviewed. Eighty-three percent of high socioeconomic participants,
and 70% of low socioeconomic participants reported a belief that “people should be asked their
opinions about social and economic problems” (Sjoberg, 1954, pg. 424). He also found that 52%
of high socioeconomic participants liked to be interviewed, while only 28% did not, and 38% of
low socioeconomic participants liked to be interviewed, while 26% did not (Sjoberg, 1954).
Heberlein and Baumgartner (1978) conducted a meta-analysis on factors affecting mail
survey response rates. They found that salience of the survey topic (defined as those topics
covering important and current behaviors or interests of the respondent) was significantly and
positively correlated with response rates. The studies deemed highly salient (N = 112) had a 77%
return rate, while the low salience studies (N = 43) had only a 44% return rate. A subsequent
study by Goyder (1982) replicated these findings, again, concluding that issue salience had a
significant effect on mail survey return rates. These results suggest that the desire to express an
attitude about a topic might serve an important purpose in the survey process.
Specifically, the question type employed in a survey and the resulting time spent
answering the question may affect the respondent’s feelings of self-expression. Previous research
6

has found that expressing one’s personal attitude on a topic may serve an important purpose.
According to Katz’s (1960) functional view of attitudes, attitudes serve four primary functions:
“the adjustive function of satisfying utilitarian needs, the ego-defensive function of handling
internal conflicts, the value-expressive function of maintaining self-identity and of enhancing the
self-image, and the knowledge function of giving understanding and meaning to the ambiguities
of the world around us” (Katz, 1960, pg. 204). A respondent’s self-expression, an ability to share
an attitude in one’s own words, is linked to the value-expressive function.
Values are “abstract principles that people consider to be guiding principles in their life”
(Maio & Olson, 2000, pg. 249 citing Rokeach, 1973 and Schwarz, 1996). Examples of values
include helpfulness, opportunity, and freedom. According to Katz (1960), when a person shares
an attitude linked to his/her personal values, the individual experiences satisfaction. “[The valueexpressive] function is central to doctrines of ego psychology which stress the importance of
self-expression, self-development, and self-realization” (Katz, 1960, pg. 170).
Katz (1960) specified two primary conditions in which the value expressive function of
attitudes can be aroused. First, an environmental cue or stimulus that is linked to the attitude
must be made salient. For example, a student is strongly opposed to an increase in school tuition
and opposes paying a public university to receive a college education. The student is a member
of a local group on campus that is discussing the recent news from the university to increase
tuition in the upcoming school year. According to Katz (1960), the environmental cue
(discussion of the tuition increase) will arouse the student’s attitude toward opposition of the new
tuition rate, regardless of the university’s rationale.

7

Second, the value-expressive function of the attitude will be aroused when the
individual’s ability to express his/her attitude in the recent past has been precluded (Katz, 1960).
If the student from the earlier example was denied the opportunity to express his/her views about
the tuition increase that occurred in the previous school year, the value-expressive function of
his/her attitude about the university’s tuition increase would become aroused. As a result, this
student would be likely to seek opportunities to express his/her attitude about the issue. When
respondents have an opportunity to express their attitude on a topic, which they care about and
have previously felt unheard, they fulfill a need for self-expression.
Open-ended questions may provide exactly this type of opportunity. They provide the
respondent with an outlet for sharing his/her attitude in his/her own words, while closed-ended
questions restrict attitude expression to the options provided by the researcher. Plus, being able
to share one’s thoughts in his/her own words, without the fear of social sanctions, may allow the
individual an opportunity to express his/her true self.
Psychologists have noted that individuals have multiple senses of self (Goffman, 1959;
Jung, 1953). These senses of self include public and inner selves, and the inner self may reside in
one’s unconscious ego (Bargh, McKenna & Fitzsimons, 2002 citing Goffman, 1959 and Jung,
2002). Literature on the self-concept suggests these selves may be categorized as ideal (desired),
ought (required), or actual (existing) (Bargh, McKenna & Fitzsimons, 2002 citing Higgins,
1987). Another type of self, the true self, suggests a difference between the public, inner, and
actual selves (Bargh, McKenna & Fitzsimons, 2002 citing Rogers, 1951). The true self is the
inner self a person possesses, but does not show in public life. It may exist at a conscious or
unconscious level. In either case, the individual has a strong need to express his/her true self
8

(Bargh, McKenna & Fitzsimons, 2002). The ability to express one’s true self is important in
establishing and reaffirming one’s individual identity (Bargh, McKenna & Fitzsimons, 2002).
Open-ended surveys provide a unique opportunity for the individual to express his/her
true self. Because surveys are anonymous, the individual does not need to fear social sanctions
for his/her response. In particular, online surveys should reduce demand effects and fear of social
sanctions for sharing one’s inner self. In an online survey, the respondent does not receive
feedback for his/her response. Because of this anonymity, the respondent should feel more
comfortable expressing his/her true self when answering open-ended survey questions. If this is
the case, we predict two relationships. As the respondent spends more time answering survey
questions, he/she should experience greater feelings of self-expression (H2). Also,
respondents should indicate greater self-expression after responding to open-ended survey
questions compared to closed-ended survey questions (H3).
In addition to providing the respondent with an outlet for expressing one’s true self,
studies in psychology, education, political science, and survey methodology have found
numerous benefits arising from the act of self-expression. In particular, these studies point to the
importance of writing as a means to facilitate self-expression. This relationship between selfexpression and writing is particularly interesting because it suggests benefits that are
transferrable to open-ended surveys.
Writing may allow people to “understand their experiences and themselves” (Pennebaker
& Seagal, 1999, pg. 1243). By writing, people cognitively process and organize their thoughts,
feelings, and experiences using language (Pennebaker & Seagal, 1999; Shumacher & Nash,

9

1991). The writing process forces people to integrate, differentiate, and organize ideas, values,
and information (Pennebaker & Seagal, 1999; Schumacher & Nash, 1991).
As a result of writing, individuals reap positive benefits. Some of these benefits may
include improvements in physical health (Pennebaker & Seagal, 1999), mental health
(Pennebaker & Segal, 1999), emotional intelligence (Wing, Schutte & Bryne, 2006), life
satisfaction (Wing, Schutte & Byrne, 2006), and learning (Shumacher & Nash, 1991). Positive
effects are found for writing about a variety of topics including personal extreme traumas
(Pennebaker & Seagal, 1999), imaginary traumas (Greenberg, Wortman & Stone, 1996), positive
events with emotion regulation cues (Wing, Schutte & Bryne, 2006), non-extreme events such as
attending college (Spera, Buhrfiend, & Pennebaker, 1994), and through more general topical
essays in college courses (Shumacher & Nash, 1991). Plus, in at least one study, writing about
personal events has been found to increase participant’s intent to participate in an additional
study (Pennebaker, 1997).
In addition to positive effects from writing, the mere measurement effect also posits that
survey participation can have desirable outcomes. As mentioned earlier, the mere measurement
effect suggests the act of survey participation may shape or alter respondent’s attitudes and
behaviors (Morwitz, 2005). By participating in a survey, the respondent cognitively processes
questions asked and shares his/her thoughts, behaviors, or attitudes though his/her response
(Fitzsimons & Shiv, 2001). As a result of this cognitive processing and response generation, the
respondent may think about a topic in a new way, consider a perspective he/she had not
considered, and/or form an opinion on a topic he/she had not considered previously (Dholakia &
Morwitz, 2002; Feldman & Lynch, 1988). Additionally, the survey process can increase the
10

salience of the survey topic for the respondent (Levav & Fitzsimons, 2006). For example, in a
survey about dental flossing, a respondent may be asked to consider the benefits of flossing and
his/her current behaviors. After processing information provided, questions asked, and his/her
responses, he/she may decide to begin flossing more frequently (Levav & Fitzsimons, 2006). In
fact, the mere measurement effect has found such outcomes for flossing (Levav & Fitzsimons,
2006) and a host of other behaviors mentioned previously.
Plus, at their very heart, surveys are about generating self-expression from the
respondent. Sure, researchers want to know about habits and behaviors (e.g., How often do you
exercise?). However, surveys also want to know about respondents’ values, attitudes, and
opinions (e.g., How important is exercising to you?). When researchers ask respondents to share
their thoughts in an open-ended format, they may enhance the opportunity for self-expression.
No matter how much time a researcher spends crafting the comprehensive list of closed-ended
response options or how accurate the linguistic features of the options may be, the provided
responses can never be a respondent’s “own words.” As a result, the ability for closed-ended
surveys to achieve true self-expression are limited. Therefore, when open-ended surveys provide
a means for a respondent to share his/her thoughts in his/her own words, and especially through
writing, we would expect these respondents will be more likely to engage in a related behavior
compared to respondents who complete a closed-ended survey. For example, if respondents
participate in an open-ended survey about a social issue, we would expect as self-expression
increases, respondents should be more likely to sign-on to a related campaign (H4).
Involvement

11

Involvement should also affect the time spent responding to a survey and feelings of selfexpression. Involvement is broadly defined as “the extent to which an individual cares about a
particular topic” (Oskamp & Schultz, 2005, pg. 229). While the involvement construct has been
widely studied in the area of persuasion research, various researchers have defined and
operationalized involvement in different ways. In particular, Johnson and Eagly (1989) specify
distinctions between value-relevant and outcome-relevant involvement.

1

Value-relevant involvement refers to “the motivational state induced by an association
between an activated attitude…” and an individual’s enduring values (Johnson & Eagly, 1989,
pg. 293; Johnson & Eagly, 1990). Johnson and Eagly (1989) assert that value-relevant
involvement, typically, is measured by a subject’s self-reports and has been studied, primarily,
using social policy issues (Johnson & Eagly, 1989). Through meta-analysis, Johnson and Eagly
(1989) found that high value-relevant involvement prevented persuasion from occurring, which
aligns with earlier findings related to the effect of ego involvement on persuasion.
In fact, the value-relevant involvement construct evolved from ego involvement, a key
component of Sherif and Hovland’s (1961) Social Judgment Involvement Theory (SJIT). Ego
involvement refers to an attitude that is a defining component of one’s sense of self or values
(Sherif & Hovland, 1961). According to SJIT, attitudes fall along a continuum, and the

1

Impression relevant involvement, a third type of involvement, links a subject’s attitude to a
desire to make “a favorable impression on others” (Johnson & Eagly, 1989, pg. 293). Impression
relevant involvement is not directly applicable to this study for a number of reasons: respondents
to a survey are typically anonymous, may not experience a need to defend their positions in an
identifiable setting, and are not communicating with an interviewer because the study is online.
Therefore, impression-relevant involvement is not likely to occur in the current study, and I do
not elaborate on its role in the persuasion literature.
12

persuasiveness of incoming messages are dependent on where the message falls relative to
reference points along one’s attitude continuum. For any given issue, an individual has a
reference point, or anchor, which is one’s own attitude. Along the attitude continuum, the
individual also has latitudes of acceptance, which encompass all positions the individual finds
acceptable, latitudes of rejection, which encompass all positions the individual finds
unacceptable, and latitudes of noncommittal, which include all positions the individual finds
neither acceptable nor unacceptable (Sherif & Hovland, 1961).
When a person receives a persuasive message, it is placed on his/her attitude continuum
and evaluated relative to the anchor (his/her own attitude) and the latitudes of acceptance,
rejection, and noncommittal. Messages within one’s latitude of acceptance tend to increase
persuasiveness, while those within one’s latitude of rejection tend to decrease persuasiveness
(Sherif & Hovland, 1961). SJIT states that people with low ego involvement will have larger
latitudes of acceptance and non-committal and smaller latitudes of rejection, while people with
high involvement will have larger latitudes of rejection and smaller latitudes of noncommittal
and acceptance (Sherif & Hovland, 1961). Thus, SJIT predicts that as ego involvement increases,
attitude change becomes less likely to occur, making persuasion more difficult. While SJIT,
itself, has resulted in many conflicting findings in the literature (Oskamp & Schultz, 2005), the
ego involvement concept has received ample attention in studies of persuasion and has been
closely linked to value-relevant involvement (Johnson & Eagly, 1989).
Another type of involvement includes outcome-relevant, which links a subject’s attitude
to his/her “purposive aspects of the self” (Johnson & Eagly, 1989). The more relevant an
outcome is to the subject’s current goals, the greater his/her outcome-relevant involvement
13

should be, while the less relevant an outcome to the subject’s current goals, the less his/her
outcome-relevant involvement should be (Johnson & Eagly, 1989). Typically, outcome-relevant
involvement is manipulated by altering the likelihood that subjects will be affected or unaffected
by some outcome (e.g., in the present or distant future, at the subject’s school or a distant
school). Primarily, outcome-relevant involvement has been studied using college student issues
(Johnson & Eagly, 1989). For example, researchers have created outcome-relevant involvement
by manipulating whether the university proposes to institute comprehensive exams as a
graduation requirement (Johnson & Eagly, 1989).
Outcome-relevant involvement is similar to the issue involvement construct, which is a
key variable in Petty and Cacioppo’s (1981) Elaboration Likelihood Model (ELM) (Johnson &
Eagly, 1989). According to the ELM, when a person is highly involved with a topic or issue,
he/she experiences greater motivation to process an incoming message or engage in high levels
of elaboration. As the amount of elaboration increases, the individual employs a central
processing approach. Central processing is an effortful, careful, and cognitive approach to
evaluating content and features of messages, communicators, and other elements of a persuasive
attempt. Conversely, when involvement is low, the individual experiences less motivation to
process the message and engages in low levels of elaboration. As the amount of elaboration
decreases, the individual employs a peripheral processing approach. Peripheral processing is less
effortful, deliberate, and cognitive, using heuristics and cues to make judgments (Petty,
Cacioppo, & Goldman, 1981).
In a test of the ELM, Petty, Cacioppo, and Schumann (1983) conducted a 2 x 2 x 2
factorial experiment with 160 undergraduate students in which they manipulated involvement
14

(high vs. low), argument strength (strong vs. weak), and source (celebrity vs. non-celebrity) to
test these variables’ effects on attitude toward a product and cognitive considerations. They
found that high involvement subjects relied on central processing and were affected by argument
quality in forming an attitude toward the product. Thus, high involvement increases persuasion
when the argument is strong (Johnson & Eagly, 1989). Conversely, low involvement subjects
relied on peripheral processing and were affected by source cues to form an attitude toward the
product (Petty, Cacioppo & Schumann, 1983).
While much work has been devoted to understanding the role of involvement in
persuasion, there remains an important potentially confounding variable in studies; that is,
outcome- and value-relevant involvement appear to be confounded with knowledge of the topic
used in the study (Park, Levine, Kingsley-Westerman, Orfgen & Foregger, 2007; Johnson &
Eagly, 1989). Studies measuring value-relevant involvement tend to use real social issues, for
which subjects are likely to possess at least some degree of knowledge, while studies
manipulating outcome-relevant involvement tend to use fictitious or hypothetical scenarios, for
which subjects are unlikely to possess knowledge (Park et al., 2007; Johnson & Eagly, 1989). As
a result, typically, value-relevant involvement has been studied as attitude change, while
outcome-relevant involvement has been studied as attitude formation (Park et al., 2007; Johnson
& Eagly, 1989).
Cho and Boster (2005)’s work to develop measurement scales for the various
involvement constructs led to the conclusion that outcome relevant-involvement was highly
correlated with information seeking. They also found that value-relevant and outcome-relevant
involvement were highly correlated with attitude extremity (Cho & Boster, 2005). Taken
15

together, these results suggest that the effect of involvement on persuasion is not as
straightforward as one might hope, and studies should seek to parse out the effects of knowledge
from the effects of involvement; the current study included.
An important distinction also exists between previous work and the current study.
Typically, researchers employ designs that test the effect of involvement on persuasive outcomes
while also manipulating messages and/or argument strength. In these cases, the predictions posed
by SJIT and ELM have not always held (e.g., Park et al., 2007). However, in the current study,
neither the persuasive power of specific messages nor the strength of arguments is of interest.
Objective surveys should present respondents with the opportunity to express their views on
either side of an issue in an anonymous context. Surveys should not attempt to persuade
respondents to adopt any particular position.
Also, it is also argued that many social issues, which surveys commonly address, are
likely to activate an individual’s values and concern about outcomes. Take for example the issue
of health care. It is likely an individual may have high value-relevant involvement as the issue
links his/her attitude with values of equality for access to care, opportunity for people to live free
from pain or disease, or freedom to choose a physician. At the same time, the issue of health care
may activate outcome-relevance as the same survey may seek to determine whether an individual
is willing to pay some additional amount in taxes to fund a new health care proposal, or whether
the individual believes a health care proposal will improve (or harm) his/her current levels of
service.
Public transportation is another example of an issue likely to activate value- and
outcome-relevant involvement among people. It is likely that value-relevant involvement may
16

cause people to support a policy about public transit based on values of equality of employment
opportunity for people with disabilities or stewardship of the environment, while others may
oppose a policy based on values of limited government and free-market enterprise. For some
people, public transportation may be outcome-relevant if they personally rely on public
transportation, provide rides to people who could use public transit if it was available, or are
unable to afford an additional amount in taxes to fund expanded routes or service hours.
Given the potential that many social issues (and especially surveys about these issues) are
likely to activate more than one type of involvement at a time, it becomes important to
understand the potential for differential effects of each type of involvement on attitudes. Of
particular interest to the current study are the effects of value- and outcome-relevant involvement
on the relationship between time spent answering a survey and feelings of self-expression.
There are a number of reasons to believe that as value-relevance increases, the
relationship between time spent answering a survey and feelings of self-expression will
strengthen. First, as mentioned earlier, Katz (1960) suggests that individuals experience feelings
of self-expression, particularly when they have an opportunity to express their attitude about an
issue closely linked to their values, and when they have previously felt unheard. For reasons
stated previously, a number of social issues are likely to activate a person’s values. Also, it is
expected that, often, people do not have a forum to express their views on many social issues on
which surveys collect data. Furthermore, SJIT suggests that when people’s attitudes are closely
linked to their values, value-relevant involvement increases, and the individual should be less
persuaded by incoming messages. This reduction in persuasion suggests that individuals with
strong values about an issue may be better able to develop counterarguments to messages (i.e.,
17

they are able to refute messages that are discrepant with their own attitude or values). Also,
because value-relevant involvement studies have been thought of as attitude change studies, this
suggests that a number of people already possess attitudes on topics addressed; they have given
the issue at least some thought, which should provide the survey respondent with more
arguments or content to include in his/her response. As a result, as value-relevant involvement
with a survey topic increases, the respondent should spend more time answering the survey
questions, resulting in greater feelings of self-expression (H5a).
Additionally, ELM research suggests that high outcome-relevant involvement may cause
individuals to spend more time processing incoming messages, or in the current study – spend
more time thinking about and carefully answering survey questions. The results of Cho and
Boster’s (2005) meta-analysis also suggest that outcome-relevant involvement significantly
correlates with attitude extremity and information seeking. Thus, it is likely that as outcomerelevant involvement increases, and individuals experience a stronger association between their
attitude and a personal goal, they should also spend more time expressing their views on an
issue. Therefore, as outcome-relevant involvement with a survey topic increases, the
respondent should spend more time answering the survey questions, resulting in greater
feelings of self-expression (H5b).
While predictions are put forth regarding the effects of outcome- and value-relevant
involvement, a research question remains: Which type of involvement, or particular
combination of outcome- and value-relevant involvement, will create the strongest
relationship between time spent answering the survey and feelings of self-expression (RQ)?

18

Because previous research does not offer an explicit answer, this study will seek to provide an
initial look at the proposed question.
In summary, when involvement with the topic addressed in a survey is high, “…the cost
of responding may be reduced, and personal input to the study may be judged by the respondent
as more important” (Heberlein & Baumgartner, 1978, pg. 458). Many surveys (and issue topics)
are not limited to activating only value- or outcome-relevant involvement. Therefore, the current
investigation assesses the effect of both types of involvement, as well as combinations of the
two, on the relationship between time spent answering a survey and feelings of self-expression.
Also, in this study, outcome- and value-relevant involvement are treated as continuous variables
to retain degree information and allow the possibility that involvement may exist in different
forms and amounts given the issue covered in a survey (using approaches similar to Park et al.,
2007, and Cho and Boster, 2005).

2

Finally, to ascertain the unique effects of involvement -

separate from knowledge about a particular issue - we control for knowledge about the particular
topic addressed in the survey.
Fatigue
One concern with open-ended questions is the potential for respondent fatigue. Fatigue is
“[t]he awareness of a decreased capacity for physical and/or mental activity due to an imbalance
in the availability, utilization, and/or restoration of resources needed to perform activity”

2

Antil (1984) argues for involvement to be treated as a continuous variable based on the
variations of the construct among different individuals/topics and the potential of increased
consistency among scholars who use involvement in their studies.
19

(Aaronson et al., 1999, pg. 46). Often, fatigue is defined by the subjective experience of the
individual (Aaronson et al., 1999).
Fatigue has been thought of as one of the factors leading to survey non-response (de Heer
& Israels, 1992). As the burden of the survey or fatigue increases, the respondent is less likely to
participate (Bogen, 1996). Open-ended surveys have been presumed to be especially prone to
generating respondent fatigue (Bogen, 1996).
Bogen (1996) conducted a literature review on the effect of questionnaire length on
response rates and concluded the literature is mixed and somewhat inconclusive. While some
studies provide evidence that increased survey length leads to greater fatigue and lower response
rates, other studies seem to suggest the opposite. Bogen (1996) finds the literature to be
disproportionately represented by mail surveys, and many studies to be wrought with poor
measurement and confounding of variables.
Galesic and Bosnjak (2009) sought to improve upon previous studies by manipulating the
length of a web survey (10, 20, or 30 minutes) and studying its effect on participation and quality
of responses to open-ended questions. Their results demonstrated a negative relationship between
survey length and willingness to participate in a survey. When respondents were told the survey
would be 10 minutes long 75% continued with the survey, but only 63% continued when the
survey length was stated to be 30 minutes (Galesic & Bosnjak, 2009). Galesic and Bosnjak
(2009) also found a negative relationship between quality of responses to open-ended questions
and survey length. As survey length increased, they found that the quality (length) of the
responses to open-ended questions decreased. These results suggest the following: In general,

20

the more time a respondent spends answering a question, the more fatigue he/she should
feel (H6).
However, the experience of fatigue could be mitigated by the opportunity to express an
attitude in the respondent’s own words and on a topic that is relevant to the respondent. To the
author’s knowledge, no studies, to date, have specifically looked at the effect of feelings of selfexpression on respondent fatigue in the context of a survey. Yet, the literature on valueexpressive attitudes suggests that some respondents may find value in communicating their
views. Some people may be seeking an outlet to share their attitudes and express their feelings
on a topic. When this is the case, the respondent should not experience burden from increased
length of a survey due to open-ended questions. Rather, the respondent should be energized by
sharing his/her attitude. Thus, the greater the respondent’s feeling of self-expression the less
fatigue he/she should report (H7).
Survey Enjoyment
Fatigue and self-expression should also affect the extent to which the respondent enjoys
participating in the survey. Fatigue is a negative emotion or an aversive state (Charlton, n.d.). It
is perceived by the brain as a signal of exhaustion and a need to cease activity and rest (Charlton,
n.d.). Previous research has shown that fatigue results in more negative attitudes toward the task
at hand. Kumashiro and Nagae (1984) found that as workers’ self-reported feelings of fatigue
increased, their attitude toward their jobs became less favorable.
Similar effects should be realized in a survey context. Fatigue may cause respondents to
tire of the survey, to wish to cease participation, and to enjoy the survey less. Therefore, the

21

more fatigue a respondent experiences, the less he/she should enjoy participating in the
survey (H8).
However, as noted earlier, fatigue could be combated by a chance to express an attitude.
Swan, Trawick, and Carroll (1981) collected descriptive data on respondents’ ratings of their
survey experience after completing a questionnaire in a restaurant setting. They found that
among 340 respondents, 87% stated they liked the opportunity to express an opinion. Also, a
large majority (92%) of respondents indicated a belief the survey was conducted so that the
restaurant could find out how to please its customers (Swan, Trawick & Carroll, 1981). When
respondents experience value in sharing their thoughts in a survey, they should feel energized,
rather than fatigued. In this case, greater feelings of self-expression should result in more
enjoyment participating in the survey (H9).
Intent to Participate in Future Surveys
Respondent enjoyment or displeasure with the survey should affect the resulting attitude
toward participating in future surveys. In fact, Rogelberg et al. (2001) studied the effect of a
general attitude toward surveys on intent to participate in future studies. The authors found that
the more positive one’s attitude toward surveys, the greater one’s willingness to participate in
3

future surveys. This finding should apply in a more specific sense. More survey enjoyment
should result in an increase in one’s intent to participate in future studies by the same
survey sponsor (H10).
3

It is important to note, the Rogelberg et al.’s (2001) study asked questions about attitudes
toward surveys after conducting a specific survey, making it difficult to ascertain whether they
were measuring a general attitude toward surveys or a specific, salient attitude toward the
recently completed survey.
22

Model
When a respondent participates in an open-ended survey in which he/she is highly
involved and has previously felt unheard, he/she experiences greater feelings of self-expression.
These feelings of self-expression off-set the fatigue he/she feels from the increased time to
complete the survey and may increase the likelihood of engaging in a related behavior, such as
signing on to a campaign. These feelings of self-expression make the survey enjoyable. Feelings
of survey enjoyment increase the respondent’s intent to participate in future surveys by the same
sponsor.
Closed-ended questions should also create a similar effect as involvement increases.
However, because the time spent answering the survey will be less in a closed-ended format, so
too should the feelings of self-expression and the likelihood of signing on to a campaign as a
result of self-expression. In turn, while the respondent will also report feelings of survey
enjoyment and an intent to participate in future surveys, the open-ended survey should elicit
stronger positive attitudes than the closed ended survey.
However, we would expect different outcomes among subjects with low involvement
when different questions types are employed. In the closed-ended condition compared to the
open-ended condition, respondents will spend less time answering a survey, and they will report
lower levels of fatigue. Additionally, because they have less involvement with the issue in the
survey than their high involvement counterparts, they will also report lower levels of selfexpression. As a result, their attitude toward the survey will be neutral to slightly unfavorable.
They are also likely to report neutral to slightly unlikely intent to participate in future surveys
and sign on to a campaign.
23

In the open-ended condition, as respondents’ level of involvement decreases, we would
expect the same relationships among the variables. Yet, we would expect the relationships to be
stronger. Respondents with low levels of involvement will not spend as much time answering
questions as respondents with high involvement, but they should spend slightly more time in the
open-ended compared to closed-ended condition due to the nature of the responses requested. As
a result, as involvement decreases, respondents to the open-ended survey compared to closedended survey will experience greater fatigue. Although open-ended respondents compared to
closed-ended respondents may experience more self-expression, because the topic is not one they
care about, it is likely this effect of self-expression will not be enough to off-set their fatigue. As
a result, as involvement decreases, respondents to the open-ended survey compared to
respondents to the closed-ended survey are likely to like the survey less, report a lower intent to
participate in future surveys, and be unwilling to sign on to a related campaign. Figure 1 provides
the predicted path model.
Method
This study employed a 2 x 4 experimental design. The independent variables were
question type (open- and closed-ended) and topic (universal health care, same-sex marriage,
copyrighted music file-sharing, and space exploration). Within each topic, outcome- and valuerelevant involvement were treated as measured, continuous variables. To ensure variation on
involvement, a number of issue topics were pilot tested. The four topics that exhibited the
greatest variation were used as topics for the open- and closed-ended surveys, creating eight
conditions. The dependent variables were time spent answering the survey, fatigue, feelings of
self-expression, enjoyment of the survey, willingness to sign on to a campaign, and intent to
24

participate in a future survey. In addition, a number of variables were controlled including
attitude toward the issue, knowledge of the survey topic, willingness to communicate,
conservativism, age, ethnicity, and gender.
Sample
Participants (N = 400) were Michigan State University undergraduate students enrolled in
4

communication courses (50 participants per cell). Respondents were between the ages of 18 and
40 (M = 20.16, SD = 1.98), and were primarily female (67%). The majority of respondents were
White (78%), while 8.50% were Asian, 8.25% were Black, 2.5% were multi-racial, 1% were
Hispanic, 1% were Arab, 0.50% were a non-listed race, and 0.25% were Pacific Islander.

5

Tables 1 and 2 display descriptive statistics of the sample for all variables of interest.
Procedure
Participants were randomly assigned to one of eight experimental conditions through an
online survey. This format was selected for three reasons. First, Dillman, Smyth, and Christian
(2009) suggested that respondents tend to provide more detailed responses to open-ended survey
questions when they are provided through a web survey compared to a paper and pencil format.

4

Completion rates varied by condition. The following were the total number of incomplete
responses for open-ended surveys by topic: universal health care (12), same-sex marriage (5),
space exploration (6), music file-sharing (6). Incompletes for the closed-ended conditions
included: universal health care (0), same-sex-marriage (6), space exploration (2), and music filesharing (2). Inspection of these cases demonstrated that most people, whether in an open- or
closed-ended condition, did not complete any questions or only completed the first question, then
ended the survey. Only a couple people completed half the survey before stopping.

5

Although this study had a high number of Asian respondents, they were evenly dispersed
throughout conditions.
25

Therefore, a web-based survey was selected to strengthen the open-ended question inductions.

6

Second, the online format provided greater consistency in timing survey completion. The survey
host automatically times participation from the time the survey link is accessed until the time the
survey is closed. Third, given the sample is college undergraduates, which is a technologically
savvy population that lives online (Howe & Strauss, 2000), we can assume they are used to
expressing themselves in an online format.
After consenting to participate, each respondent was asked ten questions on one issue
topic in either a closed- or open-ended format; these questions served as the experimental
induction. Ten questions were thought to be ample to induce induction effects without agitating
respondents and causing a reduction in survey completion rates. Questions in both the open- and
closed-ended conditions were identical. The only differences were that the closed-ended
condition had response options provided and included the following statement: “Please select the
answer that best represents your views.” In the open-ended condition, a response box was
provided for the respondent to type his/her answer and included the following statement: “Please
provide a response that best represents your views.” At the end of each open-ended question,
respondents were asked to “Please explain.” The questions used in the induction are provided in
Appendix A. Following the experimental induction, all respondents were asked a series of
closed-ended questions to measure dependent and control variables of interest. Variables were
asked in the following order after the manipulation: outcome-relevant involvement, valuerelevant involvement, attitude toward survey topic, willingness to sign on to a campaign, feelings
6

It is assumed that any individual differences due to typing proficiency among participants will
be randomly dispersed throughout the conditions.
26

of self-expression, fatigue, survey enjoyment, willingness to participate in future surveys,
knowledge about survey topic, writing apprehension, conservativism, demographics; time was
automatically recorded. Correlations among all continuous variables are provided in Table 3.
7

Additionally, the topics of the surveys varied (i.e., topic was a random effect). Four
different survey topics were used, which were pilot tested for variation on value- and outcomerelevant involvement among respondents. The four topics used in the survey were universal
health care, same-sex marriage, copyrighted music file-sharing, and space exploration. Questions
for each topic were taken from existing surveys (e.g., The Roper Center, The Pew Forum on
Religious and Public Life, and Gallup).
At the conclusion of each survey, respondents were thanked for their time. Once all
responses were collected, each respondent was sent a debriefing message via e-mail to let
him/her know the true purpose of the survey.

8

Measures
A dummy variable for survey type was created. Closed-ended surveys were treated as the
baseline. Open-ended surveys were dummy coded as 1, with closed-ended surveys coded as 0.
Outcome-relevant involvement was measured as a continuous level variable using the
question and scale created by Park et al. (2007). Outcome-relevant involvement was
7

The disadvantage of varying topics is inconsistency in survey wording between studies of
different topics. However, the advantage to varying topics is that the study presents a stronger
test of the effects of open-ended and closed-ended inductions, as well as the ability to enhance
external validity by looking at whether hypothesized predictions hold across a variety of topics
which public opinion surveys seek data.

8

The survey process took approximately two months to collect all responses.
27

operationalized as the average response to a four-item Likert-type scale. Respondents were asked
to consider an issue topic and how it might affect them personally. Then, respondents were asked
to indicate their agreement with four items using a scale ranging from 1 (strongly disagree) to 7
(strongly agree). An example item is “Universal health care would affect me personally.”
Appendix B provides this scale. This scale had a Cronbach’s alpha coefficient of .89. Higher
scores reflect greater outcome-relevant involvement (M = 3.76, SD = 1.50).
Value-relevant involvement was measured as a continuous level variable. Value-relevant
involvement was operationalized as the average of subject’s answers to a four-item scale,
combining features of the question used by Park et al. (2007) and scales used by Cho & Boster
(2005). Respondents were asked to consider an issue topic and how it fits with their values.
Then, respondents were asked to indicate their agreement with four items using a scale ranging
from 1 (strongly disagree) to 7 (strongly agree). An example item is “My attitude toward
universal health care is based on my core values.” Appendix C provides this scale. This scale had
a Cronbach’s alpha of .88. Higher scores reflect greater value-relevant involvement (M = 3.70,
SD = 1.40).
9

Time spent answering the survey was measured as a continuous level variable. Time
was operationalized as the number of seconds from when the respondent opened the first page of
the survey until the time when the respondent exited the survey, signaling completion. Higher
scores reflected more time spent answering the survey (M = 1134.29, SD = 853.87).

9

10

Subjects were not told they were timed during the study to prevent a demand effect.

10

Seven times were excluded. These times indicated 12 to 36 hours to complete the survey. This
possibility seemed highly unlikely. If respondents failed to select the “done” button on the last
28

Fatigue was measured with a modified version of the Piper Fatigue Scale (Piper et al.,
1998). The original scale consists of 11 semantic differential pairs each measured on a 10-point
scale. Higher scores indicate greater subjective feelings of fatigue. Subjects rate the degree to
which they are now feeling: strong/weak, refreshed/tired, awake/sleepy, able to
concentrate/unable to concentrate. Appendix D provides all items in this scale. As recommended
by Dillman, Smyth, and Christian (2009), to decrease respondent burden, the ranges of scales
used throughout a survey should be consistent. Therefore, the scale range was altered from a 10to a 7-point scale to maintain consistency in the survey’s design. This scale had a Cronbach’s
coefficient α of .91 (M = 3.72, SD = 0.98).
Feelings of self-expression were measured with three items asking subjects to report their
feelings of self-expression during the survey. Respondents were asked three questions based on
their subjective ratings of their survey-taking experience using a scale ranging from 1 (not at all)
to 7 (a great deal). An example item includes, “How much did you feel like you shared your
thoughts?” Appendix E provides all items used. This scale had a Cronbach’s coefficient α of .90.
Higher scores reflect greater feelings of self-expression (M = 4.97, SD = 1.26).
Enjoyment of the survey was measured with three questions regarding the respondent’s
subjective ratings of his/her experience. Respondents rated their enjoyment on a scale from 1
(not at all) to 7 (a great deal). An example question includes, “How much did you enjoy
completing the survey?” Appendix F provides all items used. One item proved to be a poor
measure of enjoyment; this item was “How much did you dislike participating in the survey?”
page of the survey and either opened another window to surf the web or left their computer
running, the timer on the survey would continue to run. Because these scenarios seemed highly
likely, these outliers were removed from the dataset.
29

After removing this item, the enjoyment scale had a Cronbach’s coefficient α of .85. Higher
scores reflect more enjoyment (M = 3.99, SD = 1.40).
Willingness to sign on to a campaign was treated as a dichotomous variable. Based on
their response to an attitude toward the topic question, respondents were asked whether they
would be willing to sign on to a campaign as a supporter or opponent of the issue in the survey.
For example, if the respondent indicated he/she supported universal health care, he/she was
provided with the following statement and question. We are creating a list of people who support
universal health care to share with decision-makers. May we add your name to this list? “Yes”
responses were coded as 1, and “No” responses were coded as 0 (Yes = 45.25%).
Intent to participate in a future survey was measured with a modified version of
MacKenzie, Lutz, and Belch’s (1986) intention to purchase scale consisting of three items
measured using a seven-point semantic differential scale. Respondents were asked: What is the
probability that you will take another survey conducted by (The Gillens Group) if they are to
contact you again: likely/unlikely, probable/improbable, possible/impossible. This scale had
Cronbach’s alpha coefficient of .95. Higher scores indicate a greater intent to participate in future
research (M = 4.45, SD = 1.48).
Attitude toward the topic was controlled. Attitude was measured with two items.
Respondents were asked to indicate to what extent they oppose/support and do not favor/favor
the issue topic in the survey using a 7-point semantic differential scale. This scale had a
Cronbach’s alpha coefficient of .95. Higher score reflect a more favorable attitude toward the
topic (M = 4.75, SD = 1.65).

30

Respondents indicated knowledge about the topic of the survey subjectively. While there
are differences between subjective knowledge (what the respondent thinks he/she knows) and
objective knowledge (what the respondent actually knows), in some conditions objective
knowledge may be significantly correlated with subjective knowledge (Carlson, Vincent,
Hardesty & Bearden, 2009). In a meta-analysis, Carlson et al. (2009) found an overall significant
correlation between objective and subjective knowledge (r = .37) as well as for societal issues (r
= .36) and public goods (r = .65). Strategies to enhance the accuracy of subjective ratings
include, using multiple indicators and framing the question in terms of a comparison between the
respondent’s knowledge and an expert’s knowledge on the topic (Carlson et al., 2009).
Therefore, subjective knowledge was measured with three items using a 7-point scale ranging
from 1 (none) to 7 (a great deal). An example item includes, “Compared to an expert on the
topic, please rate your knowledge on universal health care. Appendix G provides this scale. This
scale had Cronbach’s alpha coefficient of .92. Higher scores indicated higher ratings of
subjective knowledge (M = 2.73, SD = 1.26).
Writing apprehension, an additional non-demographic variable, was controlled. Writing
apprehension is a general fear of communicating through writing (Daly & Miller, 1975). Akin to
communication apprehension, it is a general trait anxiety about writing either within or outside of
a classroom. The fear of writing causes people to avoid writing whenever possible, and,
typically, those with writing apprehension select employment that requires as little writing as
possible (Daly & Miller, 1975). In part, this study seeks to study the effect of writing responses
to open-ended questions on attitudes; therefore, it is possible that when an individual has writing

31

apprehension, he/she may report unfavorable attitudes toward the survey due to trait anxiety, not
the manipulations in the study.
Writing apprehension was measured using a modified version of Daly and Miller’s
(1975) 5-point scale, ranging from strongly agree (1) to strongly disagree (5) and consisting of
20 items for use outside of the classroom. To decrease respondent burden and maintain
consistency with the current survey, respondents were asked to indicate the degree to which each
statement describes them using a 7-point scale ranging from 1 (does not describe me at all) to 7
(describes me exactly). Examples of items include: “I avoid writing;” “I look forward to writing
down my ideas;” “I don’t seem to write as well as most other people.” Appendix H lists all items
in this scale. This scale had a Cronbach’s alpha coefficient of .94. Higher scores reflect greater
writing apprehension (M = 3.57, SD = 1.10).
Conservativism was also controlled in the analysis. Previous research suggests people
with conservative ideologies may engage in structured decision-making, while people with
liberal ideologies have an easier time engaging in cognitive tasks that involve ambiguity,
complexity, and novelty (Amodio, Jost, Master, & Yee, 2007). Open-ended questions are less
structured than closed-ended questions. Plus, open-ended questions require the respondent to
deal with ambiguity when they must form their own answer, complexity when they must
integrate ideas, and novelty in determining how to respond in their own words. Closed-ended
responses may be easier for conservatives than liberals to answer. Using the same conservative
measure as used by Amodio et al. (2007), respondents indicated their conservativism on a
semantic differential scale ranging from -5 (extremely liberal) to +5 (extremely conservative).
Higher scores reflect more conservativism (M = -0.39, SD = 2.45).
32

Demographics were also measured. Age was treated as a continuous variable (M = 20.16,
SD = 1.98). Sex was dummy coded so that males served as the baseline; females were coded as 1
(Females = 67%). Race was dummy coded so that White respondents served as the baseline.
Black was coded as 1, with all other values coded as 0. Hispanic was coded as 1, with all other
values coded as 0. Asian was coded as 1, with all other values coded as 0. Pacific Islander was
coded as 1, with all other values coded as 0. Native American was coded as 1, with all other
values coded as 0. Arab was coded as 1, with all other values coded as 0. Multi-racial was coded
as 1, with all other values coded as 0. Other was coded as 1, with all other values coded as 0.
Results
Pilot Test
A pilot test was conducted with (N = 54) Michigan State University students enrolled in
communication courses to identify issue topics likely to vary on outcome- and value-relevant
involvement. The average pilot test respondent was a 22 year old (SD = 1.17) female (63%).
Fourteen topics were pilot tested (See Table 4). To select topics, the websites of polling firms
(e.g., The Roper Center, The Pew Forum on Religious and Public Life, and Gallup) were
reviewed. Popular topics listed on these sites were selected, such as same-sex marriage, the
economy, and immigration policy. Many of the topics listed on these polling firms’ websites are
likely to be high on one or both types of involvement for many respondents. Therefore, topics
likely to be low on both types of involvement were added to the list, such as space exploration
and congressional term limits.
As a result of the pilot test, four topics were selected for the main study. Universal health
care was deemed high on both outcome-relevant involvement, (M = 5.31, SD = 1.29), t(53) =
33

7.46, p < .001, and value-relevant involvement, (M = 5.06, SD = 1.49), t(53) = 5.27, p < .001.
Space exploration was deemed low on both outcome-relevant involvement, (M = 2.85, SD =
1.62), t(53) = -5.20, p < .001, and value-relevant involvement, (M = 2.81, SD = 1.69), t(53) = 5.20, p < .001. Same-sex marriage was deemed low on outcome-relevant involvement, (M =
2.21, SD = 1.33), t(53) = -9.85, p < .001, and high on value-relevant involvement, (M = 5.03, SD
= 1.88), t(53) = 4.02, p < .001. Copyrighted music file-sharing was deemed high on outcomerelevant involvement, (M = 4.86, SD = 1.47), t(53) = 4.31, p < .001, and low on value-relevant
involvement, (M = 3.39, SD = 1.70), t(53) = -2.64, p < .05. Figure 3 displays scores on valueand outcome-relevant involvement for each issue topic.
Main Study
A manipulation check was conducted on respondent’s involvement with the four issues
selected for the main study. Universal health care was high on outcome-relevant involvement,
(M = 4.75, SD = 1.21), t(99) = 6.19, p < .001, however, not significantly high on value-relevant
involvement, (M = 4.19, SD = 1.11), t(99) = 1.68, p = .10. Space exploration was deemed low on
both outcome-relevant involvement, (M = 3.56, SD = 1.30), t(99) = -3.39, p < .01, and valuerelevant involvement, (M = 3.08, SD = 1.36), t(99) = -6.77, p < .001. Same-sex marriage was
deemed low on outcome-relevant involvement, (M = 2.59, SD = 1.23), t(99) = -11.51, p < .001,
however, not significantly high on value-relevant involvement, (M = 4.27, SD = 1.39), t(99) =
1.91, p < .06. Copyrighted music file-sharing was not significantly high on outcome-relevant
involvement, (M = 4.14, SD = 1.33), t(99) = 1.03, p = .31, and low on value-relevant
involvement, (M = 3.23, SD = 1.31), t(99) = -5.89, p < .001. The results suggest regression
toward the mean (Campbell & Stanley, 1963), as some issues did not produce levels of
34

involvement that significantly differed from neutral. However, the overall pattern of mean
involvement scores aligned with the pilot test results.
Hypothesis 1 predicted that open-ended questions should take more time to answer than
closed-ended questions. OLS regression was used to analyze the relationship among the
variables. Table 5 provides the results. As predicted, open-ended questions took more time to
answer than closed-ended questions (β = .44, t = 9.26, p < .001, two-tailed). Thus, the data were
consistent with hypothesis 1.
Hypothesis 2 predicted as the respondent spends more time answering survey questions,
he/she should experience greater feelings of self-expression. OLS regression was used to analyze
the relationship among the variable. Table 6 provides the results. Contrary to this prediction, as
time spent answering survey questions increased, the respondent’s feelings of self-expression did
not change significantly (β = -.22, t = -0.71, p = .48, two-tailed). Although predictions were not
put forth, attitude toward the topic in the survey (β = .14, t = 2.63, p < .01, two-tailed) appeared
to increase feelings of self-expression. Therefore, the data were not consistent with hypothesis 2.
Hypothesis 3 predicted respondents would indicate greater self-expression after
responding to open-ended survey questions compared to closed-ended survey questions. OLS
regression was used to analyze the relationship among the variables. Table 6 provides the results.
As predicted, respondents indicated significantly greater feelings of self-expression after
responding to open- compared to closed-ended survey questions (β = .11, t = 1.99, p < .05, twotailed), controlling for other covariates. Thus, the data were consistent with hypothesis 3.
Hypothesis 4 predicted that open-ended surveys would increase self-expression, thus,
increasing the likelihood of signing on to a campaign. Logistic regression was used to analyze
35

the relationship among the variables. Table 7 provides the results. A significant interaction was
found between survey type and self-expression (b = .45, z = 2.33, p = .02, two-tailed),
controlling for other covariates. In open-ended surveys, as self-expression increased, a
respondent’s probability of signing on to a campaign increased by 13%. This effect was not
present in closed-ended surveys. Thus, the data were consistent with hypothesis 4.
Two hypotheses were posed for the effect of involvement. Hypothesis 5a predicted as
value-relevant involvement with a survey topic increases, the respondent should spend more time
answering survey questions, resulting in greater feelings of self-expression. OLS regression was
used to analyze the relationship among the variables. Table 6 provides the results for valuerelevant involvement. Value-relevant involvement and time did not interact to increase feelings
of self-expression (β = .09, t = 0.24, p = .81, two-tailed), controlling for other covariates.
Hypothesis 5b predicted as outcome-relevant involvement with a survey topic increases, the
respondent should spend more time answering survey questions, resulting in greater feelings of
self-expression. Also, outcome-relevant involvement and time did not interact to increase
feelings of self-expression (β = .40, t = 0.96, p = .34, two-tailed), controlling for other covariates.
Table 6 provides the results for outcome-relevant involvement. Thus, the data were not
consistent with either hypothesis 5a or 5b. That is, the effect of involvement on self-expression
did not vary with time spent answering a survey.
A research question was also posed: Which type of involvement, or particular
combination of outcome- and value-relevant involvement, will create the strongest relationship
between time spent answering the survey and feelings of self-expression? Interactions were
tested using OLS regression. The effect of value-relevant involvement on self-expression was
36

found to be significant using a one- but not two-tailed test (β = .35, t = 1.77, p = .078, twotailed), providing support by conventional standards. Therefore, as value-relevant involvement
increases, one’s feelings of self-expression in a survey may also increase. However, outcomerelevant involvement does not seem to effect self-expression, nor does any particular
combination of outcome- and value-relevant involvement or the effect of time spent responding
to a survey. These analyses are presented in Table 6.
Hypothesis 6 predicted the more time a respondent spends answering a survey, the more
fatigue he/she should feel. OLS regression was used to analyze the relationship among the
variables. Table 8 provides the results. Contrary to this prediction, as time spent answering a
survey increased, the respondent’s feelings of fatigue did not change significantly (β = .07, t =
1.56, p = .12, two-tailed), controlling for other covariates. Therefore, the data were not consistent
with the sixth hypothesis; time spent responding to a survey did not affect feelings of fatigue.
However, increases in attitude (β = -.15, t = -3.00, p < .01, two-tailed) and knowledge (β = -.14, t
= -2.91, p < .01, two-tailed) about the topic in the survey appeared to decrease fatigue, while
increases in writing apprehension (β = .12, t = 2.46, p < .05, two-tailed) increased fatigue
significantly.
Hypothesis 7 predicted the greater the respondent’s feeling of self-expression the less
fatigue he/she should report. OLS regression was used to analyze the relationship among the
variables. Table 8 provides the results. As self-expression increased, fatigue significantly
decreased (β = -.21, t = -4.27, p < .001, two-tailed), controlling for other covariates. The data
were consistent with hypothesis 7.

37

Hypothesis 8 predicted the more fatigue a respondent experiences, the less he/she should
enjoy participating in the survey. OLS regression was used to analyze the relationship among the
variables. Table 9 provides the results. As fatigue increased, enjoyment significantly decreased
(β = -.29, t = -6.62, p < .001, two-tailed), controlling for other covariates. The data were
consistent with hypothesis 8.
Hypothesis 9 predicted greater feelings of self-expression should result in more
enjoyment participating in the survey. OLS regression was used to analyze the relationship
among the variables. Table 9 provides the results. As self-expression increased, enjoyment
significantly increased (β = .36, t = 8.38, p < .001, two-tailed), controlling for other covariates.
The data were consistent with hypothesis 9.
Hypothesis 10 predicted more survey enjoyment should result in an increase in one’s
intent to participate in future studies by the same survey sponsor. OLS regression was used to
analyze the relationship among the variables. Table 10 provides the results. As enjoyment
increased, willingness to participate in a future survey significantly increased (β = .47, t = 9.72, p
< .001, two-tailed), controlling for other covariates. The data were consistent with hypothesis 10.
Although specific hypotheses were not put forth, the effect of open- compared to closedended survey questions was tested on the variables in the model, controlling for other covariates.
Open-ended questions significantly increased self-expression (β = .11, t = 2.26, p < .05, twotailed) and had a marginally significant effect on fatigue (β = .09, t = 1.86, p = .06, two-tailed).
Open-ended questions did not have a statistically significant effect on enjoyment (β = -.03, t =
-.61, p = .54, two-tailed), but significantly decreased one’s intent to participate in future surveys

38

by the same sponsor (β = -.44, t = -2.22, p < .05, two-tailed).

11

Other tests included the effect of

self-expression (β = .14, t = 2.17, p < .05, two-tailed) and a potential interaction between survey
type and self-expression (β = .33, t = 1.56, p = .12, two-tailed) on intent to participate in a future
survey.
Because the data did not display statistically significant predictions for time and
involvement, path modeling was not conducted. The purpose of the previously mentioned
models was to create a visual of how the variables in this study might fit together. Because the
data were not consistent with some of these predictions, the models are not consistent with the
data.
Discussion
The purpose of this study was to determine the relationships among various factors that
may lead to survey enjoyment. By understanding these relationships, researchers may better
understand the respondent’s survey experience and the outcomes that arise from survey
participation. Of particular interest to this study was to learn whether researchers can ask
questions in ways and about topics that elicit a (un)favorable response and, if so, to understand
the effects of doing so.
This study had several interesting findings that conformed to predictions and others that
were contrary. The findings suggest that open-ended surveys significantly increase a
respondent’s perceived self-expression. Although self-expression may increase enjoyment of
11

After testing the effect of question type on intent to participate in future surveys by topic, only
the topic of space exploration generated a statistically significant effect; the effect was negative.
This finding provides evidence that asking open-ended questions about topics people don’t care
about can decrease people’s intent to participate in future surveys.
39

survey participation and intent to participate in a future survey, open-ended surveys do not
appear to interact with self-expression to create significant levels of survey enjoyment. In fact,
open-ended surveys appear to have a negative main effect on intent to participate in a future
survey. This finding suggests the type of questions employed have important implications for
survey researchers. Also, the role of self expression in the survey process appears to have
important implications for future participation. In particular, the fact that self-expression in
surveys can increase enjoyment, which can increase intent to participate is incredibly important
given the trend in declining survey rates.
One way to increase felt self-expression is to ask about topics related to the respondent’s
values. This effect is not surprising given Katz’s (1960) functional of attitudes and Roger’s
(1951) conceptualization of the true self, both of which posit one’s need to express his/her core
values. A somewhat surprising finding was the lack of a statistically significant effect of
outcome-relevant involvement on self-expression, coupled with the negative signs for Beta
coefficients in regression models and the negative correlation with intent to participate in future
surveys. Taken together, it appears as though people may not find value in discussing social
issues that may impact them. Rather, respondents enjoy expressing their values.
Because the statistical effect of value-relevant involvement was marginal (p = .07), and
the topics used in the main study were not as extreme on involvement as in the pilot test, issue
topics were assessed. Inspection of mean scores by issue topic demonstrates same-sex marriage
(M = 5.48, SD = 1.19) generated more self-expression than universal health care (M = 4.79, SD =

40

1.30), space exploration (M = 4.62, SD = 1.22), or music file-sharing (M = 4.97, SD = 1.20).

12

Therefore, the particular effect of the issue topic used in the survey should be considered.
The issue of same-sex marriage is unique to the other topics used in this study in many
ways. First, many people feel the issue of same-sex marriage is related to their values and may
not necessarily have an effect on them. In fact, same-sex marriage created the greatest
discrepancy between value- (M = 4.27, SD = 1.39) and outcome-relevant involvement (M = 2.59,
SD = 1.23) compared to the other issue topics. Plus, same-sex marriage may not be an issue
many people would feel comfortable expressing their true views about to others in a social
setting. People may conceal parts of their true selves when they fear social sanctions for stating
their innermost thoughts and feelings about a topic (Bargh, McKenna & Fitzsimons, 2002). Plus,
when people are given the opportunity to express their thoughts on an issue related to their
values, and on which they have previously felt unheard, they may fulfill a need for selfexpression (Katz, 1960). Thus, a survey about a socially sensitive, value-related topic such as
same-sex marriage may provide respondents an outlet for sharing their innermost thoughts on a
topic that is not frequently or openly discussed using an anonymous tool, for which people can
avoid the fear of social judgment. When the topic of a survey is a sensitive, social issue, surveys
may provide an outlet for respondents to fulfill a need for self-expression. Further testing of this
possibility is needed. Future studies may consider studying topics related to race, disability, and
gender. Additionally, studies may consider including questions related to whether/how often

12

OLS regression was also conducted to confirm differences between issue topic on selfexpression controlling for other variables; the results confirmed same-sex marriage created more
self-expression than the other issue topics.
41

people have expressed their true thoughts on this issue prior to the current survey. Asking such
questions may provide additional evidence of the opportunity for surveys to provide a unique
opportunity for self-expression that may not exist in normal discourse.
Also, this survey found that self-expression in an open-ended survey increases the
likelihood of signing on to a campaign. A caveat to this finding should be noted. Respondents
were asked to sign on to a campaign that advocated his/her “side” of the issue. That is, the
campaign’s purpose aligned with the respondent’s support or opposition of an issue. However,
when alignment occurred, people who experienced greater self-expression in an open-ended
survey were more likely than those in the closed-ended survey condition to add their name to a
campaign.
This finding has important implications for political and social campaigns. This
approach, asking for campaign endorsements within the context of a survey, may suggest an
efficient way to stimulate public support. Some people might argue this is simply a foot-in-thedoor or cognitive dissonance reduction approach. By asking people to take a survey first, then
sign on to a campaign, or asking people whether they support or oppose an issue, then asking
that person to provide evidence by adding his/her name, the survey provides a means to employ
one of these traditional campaign approaches. However, the current findings suggest that
engaging people in the survey process may increase willingness to endorse a campaign due to
self-expression that occurs within the survey. By allowing people to share their voice in their
own words, surveys may be able to send an implicit message that this campaign cares about
“you” and what you really think. When people find value in being able to express their true

42

thoughts, the campaign may become another outlet to continue to express their attitude through
another means – adding their name as an endorser.
Although the data supported several of the predicted effects, they failed to detect support
for the effect of time. Time did not affect self-expression or interact with either type of
involvement to affect enjoyment. These findings suggest variables, other than time, affect selfexpression. In fact, in regression analysis, knowledge and attitude toward the survey topic had a
significantly positive relationship with self-expression. Therefore, self-expression appears to
occur as a result of being able to talk about issues one supports, not as a result of spending more
time sharing thoughts and opinions.
This study had a number of limitations. First, this study did not assess the effect of a
potentially important personality variable, self-monitoring. Self-monitoring is one’s control over
his/her verbal or nonverbal expressive acts (Snyder, 1974). Individuals who are high selfmonitors intentionally control their verbal and nonverbal expressions and behaviors according to
situational cues; whereas, low self-monitors more freely express their thoughts and emotions
across situations (Snyder, 1974). The survey context may provide high self-monitors with the
opportunity to express their true self without concern for social appropriateness, especially in the
context of an online, open-ended, anonymous study. However, self-expression in a survey may
not be sufficient to encourage high self-monitors to engage in a behavior such as adding their
name to an endorsement list, especially if they believe doing so would not be socially acceptable.
Future studies may consider the role of self-monitoring on variables such as self-expression,
enjoyment, willingness to engage in behaviors (e.g., signing on to a related campaign), and intent
to participate in future surveys.
43

Second, the current study did not address the issue of impression-relevant involvement
and only focused on value- and outcome-relevant involvement. Not including impressionrelevant involvement is a limitation because many surveys are conducted over the phone or faceto-face. The presence of an interviewer may create pressure for the respondent to consider social
norms and the impression his/her responses may make on the interviewer, thereby limiting the
self-expression that occurs in the context of the survey. Additional studies may seek to parse out
the effect of the presence or absence of an interviewer on self-expression, especially when the
nature of the survey topic is value-relevant and socially sensitive.
Finally, this study failed to fully create the expected manipulations for involvement.
While the assumption is that regression toward the mean occurred in this study as opposed to an
unintended effect of the manipulation, future studies should consider including an off-set control
group to measure involvement, attitude, and knowledge toward the issue. This additional element
would provide evidence that differences in involvement scores from pre- to post-tests are a result
of respondents’ scores moving closer to the population mean and not from a the manipulation
that occurred in the context of the survey.
Conclusion
As researchers, we need to consider the impact our surveys have on our respondents. If
we ask questions in ways and about topics that elicit an unfavorable response, we may be
contributing to the problem of declining survey rates. However, as demonstrated in this study,
when we can generate increases in a respondent’s self-expression, we create opportunities for
beneficial survey outcomes. Some of these positive effects of self-expression include increasing
behavioral outcomes of endorsing a campaign, increasing enjoyment of the survey experience,
44

and increasing willingness to participate in a future survey by the same sponsor. Thinking about
the respondent as merely a unit of data can be easy to do in the midst of busy schedules and work
overload. However, when we step back and critically analyze how the respondent might feel
about his/her participation and the opportunities we are providing for that individual to express
his/her true thoughts, we may consider the survey process differently. Surveys communicate
information; what and how we ask matter.

45

Table 1
Descriptive Statistics for Continuous Variables by Condition (N=50 participants per cell)
Health care
Same-sex marriage Space exploration
Music
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Variable
M
M
M
M
M
M
M
M
(SD)
(SD)
(SD)
(SD)
(SD)
(SD)
(SD)
(SD)
Time
756.40
1623.54
820.22 1542.22 796.10 1440.00 713.63
1406.82
(400.52) (1025.18) (482.57) (993.98) (555.39) (950.50) (361.75) (1021.08)
Age
20.36
20.16
20.56
20.12
20.12
19.82
20.25
19.86
(3.37)
(1.42)
(2.20)
(1.80)
(1.72)
(1.38)
(1.82)
(1.31)
OR
4.64
4.87
2.47
2.71
3.45
3.67
4.20
4.04
Involve
(1.20)
(1.22)
(1.27)
(1.18)
(1.19)
(1.40)
(1.31)
(1.40)
VR
4.09
4.29
4.16
4.38
3.07
3.09
3.27
3.21
Involve
(1.14)
(1.10)
(1.40)
(1.37)
(1.16)
(1.55)
(1.31)
(1.34)
Attitude
4.17
3.97
5.50
5.19
4.67
4.76
4.53
5.21
(1.55)
(1.88)
(1.60)
(2.07)
(1.42)
(1.33)
(1.30)
(1.38)
Express
4.44
5.14
5.31
5.65
4.65
4.59
4.93
5.01
(1.41)
(1.09)
(1.19)
(1.17)
(1.26)
(1.18)
(1.27)
(1.15)
Enjoy
3.77
3.77
4.54
4.41
4.04
3.55
4.01
3.87
(1.41)
(1.13)
(1.34)
(1.46)
(1.19)
(1.54)
(1.52)
(1.32)
Fatigue
3.91
3.86
3.35
3.53
3.61
4.03
3.65
3.93
(0.96)
(0.97)
(0.75)
(0.96)
(0.98)
(1.16)
(0.89)
(0.97)
Intent
4.31
4.29
5.03
4.62
4.63
3.73
4.60
4.41
(1.32)
(1.51)
(1.32)
(1.35)
(1.48)
(1.51)
(1.54)
(1.46)
Willing
.50
.48
.68
.57
.44
.42
.28
.26
(.51)
(.50)
(.47)
(.49)
(.50)
(.50)
(.45)
(.44)
Know
2.87
2.53
3.23
3.08
2.64
2.29
2.60
2.60
(1.10)
(1.27)
(1.24)
(1.16)
(1.16)
(1.35)
(1.40)
(1.15)
Writing
3.08
3.45
3.44
3.72
3.64
3.89
3.64
3.71
App.
(1.12)
(0.93)
(1.06)
(1.13)
(1.06)
(0.96)
(1.10)
(1.11)
Conser.
-0.76
0.42
-0.64
-0.82
-0.18
-0.48
-0.32
-0.32
(2.75)
(2.24)
(2.41)
(2.43)
(2.71)
(2.31)
(2.50)
(2.12)

46

Table 2
Descriptive Statistics for Categorical Variables by Condition (N=50 participants per cell)
Health care
Variable

Sex
Female
Male
Race
White
Black

Space
exploration
Closed Open
N
N
(%)
(%)
38
30
(76%) (60%)
12
20
(24%) (40%)
41
(82%)
3
(6%)
1
(2%)
3
(6%)

Closed
N
(%)
33
(66%)
17
(34%)

Open
N
(%)
29
(58%)
21
(42%)

38
(76%)
3
(6%)

39
(78%)
3
(6%)

5
(10%)

6
(12%)

37
(74%)
6
(12%)
1
(2%)
4
(8%)

1
(2%)
1
(2%)

1
(2%)
1
(2%)

Hispanic
Asian

Same-sex
marriage
Closed Open
N
N
(%)
(%)
41
31
(82%) (62%)
9
19
(18%) (38%)
45
(90%)
1
(2%)

3
(6%)

36
(72%)
6
(12%)

7
(14%)

Pacific Islander
Arab
Multiracial
Other Race

3
(6%)
1
(2%)

2
(4%)
1
(2%)

47

1
(2%)

Music
Closed
N
(%)
32
(64%)
18
(36%)

Open
N
(%)
35
(70%)
15
(30%)

36
(72%)
6
(12%)
1
(2%)
4
(8%)
1
(2%)
1
(2%)
1
(2%)

40
(80%)
5
(10%)
1
(2%)
2
(4%)

1
(2%)
1
(2%)

Table 3
Correlations Among Variables
Valuerelevant
Involve

Outcomerelevant
Involve

Selfexpress

Valuerelevant
Involve

.17***

.25***

-.04

.27***

.02

.46***

Fatigue

-.18***

.07

-.25***

Intent

Willing

1.00

Enjoy

Time

1.00

Selfexpress

Fatigue

1.00

Outcomerelevant
Involve

Enjoy

.04

1.00
-.45***
-.02

1.00

Time

.05

-.04

Intent

.12*

-.12*

.22***

.47***

-.33***

-.02

1.00

Willing

.27***

-.04

.18***

.23***

-.10*

-.04

.20***

48

.08

1.00

1.00

Table 4
Means, Standard Deviations, and Mean Differences for Topics used in Pilot Test (N=54)

Topic

Outcome
Relevant
Involvement
M
(S.D.)

Public transportation
Tuition
Abortion
Copyrighted music file-sharing
Marijuana
Genetically modified food
Same-sex marriage
Space exploration
Global warming
War in Afghanistan
Universal health care
Economy
Congressional term limits
Immigration

M∆
(value = 4)

3.68
(1.65)
5.94
(1.42)
3.10
(1.54)
4.86
(1.47)
3.01
(1.75)
4.49
(1.62)
2.21
(1.33)
2.85
(1.62)
4.93
(1.63)
4.01
(1.48)
5.31
(1.29)
6.02
(1.40)
3.43
(1.49)
3.17
(1.81)

-0.32
1.94***
-0.90***
0.86***
-0.99***
0.49*
-1.79***
-1.15***
0.93***
0.01
1.31***
2.02***
-0.57**
-0.83**

*p < .05, **p < .01, ***p < .001

49

Value
Relevant
Involvement
M
(S.D.)
2.61
(1.54)
3.86
(1.78)
5.29
(1.71)
3.39
(1.70)
3.73
(1.92)
3.90
(1.77)
5.03
(1.88)
2.81
(1.69)
4.36
(1.83)
4.82
(1.44)
5.06
(1.49)
4.01
(1.83)
2.84
(1.58)
4.15
(1.70)

M∆
(value = 4)
-1.39***
-0.14
1.29***
-0.61*
-0.27
-0.10
1.03***
-1.19***
0.36
0.82***
1.06***
0.01
-1.16***
0.15

Table 5
OLS Regression Results for Effect of Survey Type on Time Spent Answering Survey
β

742.88***

80.26

.44

19.29
8.74
-5.66
5.30
-4.80

25.98
32.54
38.06
17.11
20.59

.04
.01
-.01
.02
-.01

43.87

86.18

.02

-2.54
-193.36
140.86
-31.16
-98.08
634.07*
46.47
713.98
6.86***
.17

Variable
Survey Type (baseline = closed-ended)
Open-ended

Time (in seconds)
s.e.

150.38
394.00
143.29
786.28
394.71
251.14

-.08E-2
-.02
.05
-.002
-.01
.12

556.12

.004

b

Controls
Attitude
Knowledge
Writing Apprehension
Conservativism
Age
Sex (baseline = male)
Female
Race (baseline = White)
Black
Hispanic
Asian
Pacific Islander
Arab
Multi-racial
Other race
Constant
F
2

Adj. R
N
*p<.05; **p<.01;***p<.001, two-tailed

393

50

Table 6
OLS Regression Results for Effects of Survey Type, Time, and Value- and Outcome-relevant
Involvement on Self-expression
β

.14

.11

.04E-2
.18

-.22
.35

.18
.01E-2
.01E-2
.05
.04E-3

-.04
.40
.09
-.17
-.30

.11**
.10
-.04
-.01
-.04

.04

.14

.05
.06
.03
.03

.10
-.04
-.02
-.06

-.11

.13

-.04

.26
-.53
-.38
1.36
-.24
-.19
.76
4.04***

Variable
Survey Type (baseline = closed-ended)
Open-ended

Self-expression
s.e.

.23
.61
.23
1.23
.61
.40
.87
1.03

.05
-.04
-.08
.05
-.02
-.02
.04

b

Time (in seconds)
Value-relevant Involvement
Outcome-relevant Involvement
Outcome-relevant Involvement x Time
Value-relevant Involvement x Time
Outcome- x Value-relevant Involvement
Outcome x Value-relevant Involvement x Time
Controls
Attitude
Knowledge
Writing Apprehension
Conservativism
Age
Sex (baseline = male)
Female
Race (baseline = White)
Black
Hispanic
Asian
Pacific Islander
Arab
Multi-racial
Other race
Constant
F
2

Adj. R
N
*p<.05; **p<.01;***p<.001; a =.078, two-tailed

51

.28*
-.03E-2
a
.32
-.03
.01E-2
.03E-3
-.02
-.02E-3

2.84***
.09
393

Table 7
Logistic Regression Results for Effect Survey Type and Self-expression on Willingness to Sign-on
to a Campaign

Variables
Survey Type (baseline=closed-ended)
Open-ended
Self-expression
Survey Type x Self-expression
Controls
Attitude
Knowledge
Writing Apprehension
Conservativism
Age
Sex (baseline = male)
Female
Race (baseline = White)
Black
Hispanic
Asian
Pacific Islander
Arab
Multi-racial
Other race
Constant
Log Likelihood
N

Willingness to Sign-on to a Campaign
Model 1
Model 2
Model 3
Open-ended
Closed-ended
Logit
s.e.
Logit
s.e.
Logit
s.e.
-2.43*
.05

1.00

.45*

.19

.26***
.32***
-.04

.53**

.16

.04

.13

.07

.16

.10

.39**

.12

.09

.25

.13

.14

-.10*
.05

.11
.05

.02
-.11

.16
.08

.49**
-.11
-.10

.16
.06

.06

.00

.11

.08

.07

.03

.24

-.06

.34

.04

.37

-.99*
-.31
.01

.44

-.88

.66

-1.29

.66

-.04

.58

-.25
.14

1.56
.59

1.37

1.30

-.80

.83

-4.51

2.53

-.63
-.06
-.61

.12

1.18
.40
1.04
.68
1.48
1.54

-3.24*
-240.63
399

*p<.05; **p<.01; ***p<.001
Omitted scores had perfect predictions and were dropped.
52

-115.37
194

-4.68* 2.08
-114.24
195

Table 8
OLS Regression Results for Effects of Time and Self-expression on Fatigue
b
.08E-3

Asian
Pacific Islander
Arab
Multi-racial
Other race

-.14

.04

.12

.02
.02

-.04
-.01

.10

.07

.17

-.12

-1.48**
-.15
1.11
-.57
.31
1.31*

Hispanic

.04

-.43*

Conservativism
Age
Sex (baseline = male)
Female
Race (baseline = White)
Black

-.15

.14

Writing Apprehension

.03

.11*
-.01
-.01

Knowledge

-.21

-.11**

Controls
Attitude

.04

-.09**

Self-expression

β
.07

-.16***

Variable
Time (in seconds)

Fatigue
s.e.
.05E-3

.46

-.15

.17
.91
.46
.29
.64

-.04
.06
-.06
.05
.10

Constant

4.86***

F

6.20***
.17

2

Adj. R
N
*p<.05; **p<.01;***p<.001, two-tailed

393

53

Table 9
OLS Regression Results for Effects of Fatigue and Self-expression on Enjoyment
b
-.42***

Constant
F
Adj. R
N

.06
.16

.05
.02
.03

-.05
-.02E-1
.05

.12

.09

.21
.57
.20
1.12
.56
.36
.80

-.02
.02
.08
.03
.05
.02
-.08

2.17**

Race (baseline = White)
Black
Hispanic
Asian
Pacific Islander
Arab
Multi-racial
Other race

.04
.05

-.10
.24
.38
.76
.72
.15
-1.67*

Writing Apprehension
Conservativism
Age
Sex (baseline = male)
Female

.36

.27*

Controls
Attitude
Knowledge

.05

.05
.18***
-.07
-.01E-1
.04

Self-expression

β
-.29

.39

Variable
Fatigue

Enjoyment
s.e.
.06

.79

16.85***
.37

2

400

*p<.05; **p<.01;***p<.001, two-tailed

54

Table 10
OLS Regression Results for Effect of Enjoyment on Intent to Participate in Future Surveys
Intent to Participate in Future Surveys
b
s.e.
β
.05
.47
.49***

Variable
Enjoyment
Controls
Attitude
Knowledge
Writing Apprehension
Conservativism
Age
Sex (baseline = male)
Female
Race (baseline = White)
Black
Hispanic
Asian

.05
.04
.02
.01
.01

.02

.24
.65
.23

-.05
.03
-.11

-4.41**
-.40
.16
-.56
1.89*

F

.14

-.29
.45
-.59*

Arab
Multi-racial
Other race
Constant

.05
.03
.02
.01
.02

.06

Pacific Islander

.04
.05
.06
.03
.03

1.30

-.15

.65
.41
.92
.80

-.03
.02
-.03

10.00***
.24

2

Adj. R
N
*p<.05; **p<.01;***p<.001, two-tailed

400

55

Figure 1
Path Model Predictions for Effects of Question Type on Survey-taking Outcomes
Time spent
answering
survey

+
Survey
Type*

+
Fatigue
-

+
+
Involvement

Enjoyment

+

+

Selfexpression
+

+

Willingness to
sign on to a
campaign

*With open-ended dummy coded as 1 and closed-ended coded as 0.

56

Intent to
participate
in future
surveys

Figure 2
Mean Scores for Outcome- and Value-relevant Involvement by Pilot Test Topic

57

APPENDICES

58

Appendix A
Manipulation Questions for Closed-ended Surveys by Issue
Universal Health Care
First, we’d like you to think about the current health care system in the United States.
For each of the following questions, please select the answer that best represents your views.
1. How well does our current health care system work for most Americans?
• Not at all well
• Not very well
• Just somewhat well
• Fairly well
• Very well
• Don’t know
2. How well does our current health care system work for you personally?
• Not at all well
• Not very well
• Just somewhat well
• Fairly well
• Very well
• Don’t know
3. What issue should be the most important focus for reform to the U.S. health care system, if
any?
• Expanding coverage for uninsured Americans
• Reducing out-of-pocket costs for individuals
• Keeping the U.S. health care system out of the hands of government
• Controlling escalating costs for new drugs and medical technologies
• Maintaining excellent-quality care
• Affording the cost of reforming the U.S. health care system
• Maintaining individual choice of doctors, hospitals and treatments
• I don’t think we need any changes/reform to the U.S. health care system.
• Don’t know
4. How satisfied or unsatisfied are you with the cost & affordability of health care in the U.S.?
• Extremely unsatisfied
• Unsatisfied
• Somewhat unsatisfied
59

•
•
•
•
•

Neither unsatisfied nor satisfied
Somewhat satisfied
Satisfied
Extremely satisfied
Don’t know

5. How satisfied or unsatisfied are you with the quality of health care in the U.S.?
• Extremely unsatisfied
• Unsatisfied
• Somewhat unsatisfied
• Neither unsatisfied nor satisfied
• Somewhat satisfied
• Satisfied
• Extremely satisfied
• Don’t know
Now, we’d like to ask your thoughts about changes to health care in the United States.
6. Do you think it is the government’s responsibility to make sure that everyone in the United
States has adequate health care, or don’t you think so?
• Yes
• No
• Don’t know
7. How would you feel about, “Having a national health plan in which all Americans would get
their insurance through an expanded, universal form of Medicare-for-all”?
• Strongly oppose
• Oppose
• Somewhat oppose
• Neither oppose nor support
• Somewhat support
• Support
• Strongly support
• Don’t know
8. How would you feel about universal health coverage if it meant longer waits for
nonemergency treatment and a limited choice of doctors?
• Strongly oppose
• Oppose
• Somewhat oppose
• Neither oppose nor support
• Somewhat support
60

•
•
•

Support
Strongly support
Don’t know

9. How much would you be willing to pay per year in additional federal taxes to assure every
American citizen received health care coverage?
• $0
• Less than $100
• $100-$199
• $200-$299
• $300-$399
• $400-$499
• $500-$599
• $600-$699
• $700-$799
• $800-$899
• $900-$999
• $1,000 or more
• Don’t know
10. As you may know, a new health reform bill was signed into law earler this year. Given what
you know about the new health reform law, what is your opinion of it?
• Strongly unfavorable
• Unfavorable
• Somewhat unfavorable
• Neither unfavorable nor favorable
• Somewhat favorable
• Favorable
• Strongly favorable
• Don’t know
Same-sex Marriage
First, we’d like you to think about movies, television, and books you see these days.
For each of the following questions, please select the answer that best represents your views.
1. Do you think the entertainment media are including too many gay themes and characters these
days, not enough gay themes and characters, or about the right amount?
• Too many
• Not enough
• The right amount
61

•

Don’t know

Thinking about homosexuality in general…
2. In your opinion, when a person is homosexual, is it something that people are born with, or is
it something that develops because of the way people are brought up, or is it just the way that
some people prefer to live?
• Something people are born with
• Something that develops because of the way people are brought up
• Just the way some people prefer to live
• Don’t know
3. Do you think a gay or lesbian person’s sexual orientation can be changed or cannot be
changed? Please explain.
• Can be changed
• Cannot be changed
• Don’t know
Now, thinking about homosexuality and your religious beliefs…
4. Do you think it is a sin, or not, to engage in homosexual behavior?
• Yes, it is a sin.
• No, it is not a sin.
• Don’t know
Next, please think about homosexuality in the United States.
5. Do you think more acceptance of gays and lesbians would be a good thing or a bad thing for
the country – or
that it would not make much difference either way?
• Good for country
• Bad for country
• Wouldn’t make much difference
• Don’t know
6. How would you describe your overall opinion of gay men?
• Very favorable
• Mostly favorable
• Neither favorable nor unfavorable
• Mostly unfavorable
• Very unfavorable
• Don’t know
62

7. How would you describe your overall opinion of lesbian women?
• Very favorable
• Mostly favorable
• Neither favorable nor unfavorable
• Mostly unfavorable
• Very unfavorable
• Don’t know
Now, please think about your views on policies regarding same-sex couples.
8. How do you feel about allowing gay and lesbian couples to enter into legal agreements with
each other that would give them many of the same rights as married couples?
• Strongly favor
• Favor
• Somewhat favor
• Neither favor nor oppose
• Somewhat oppose
• Oppose
• Strongly oppose
• Don’t know
9. How do you feel about allowing gays and lesbians to marry legally?
• Strongly favor
• Favor
• Somewhat favor
• Neither favor nor oppose
• Somewhat oppose
• Oppose
• Strongly oppose
• Don’t know
10. How do you feel about allowing same-sex couples to legally adopt a child?
• Strongly favor
• Favor
• Somewhat favor
• Neither favor nor oppose
• Somewhat oppose
• Oppose
• Strongly oppose
• Don’t know
63

Space Exploration
First, we’d like you to think about space exploration in general.
For each of the following questions, please select the answer that best represents your views.
1. To what degree should the U.S. explore space?
• There should not be a space program.
• We should reduce our efforts to explore space.
• We should continue our efforts at current levels.
• We should have a much more comprehensive space program.
• Don’t know
2. What is the main reason driving American space exploration?
• It is human nature to explore
• To maintain the United States' status as the international leader in space
• To provide benefits on Earth
• To keep the nation safe
• It inspires people
• It motivates children
• Don’t know
3. What, in your opinion, should be the ultimate purpose of the U.S. planetary exploration
program?
• Determine the suitability of other planets for human colonization
• Establish permanent robotic outposts on other planets
• Scientific Exploration—learn everything we can about the planets
• Learn lessons about the Earth by studying other planets
• Look for life on other planets
• Search for resources in space for use in space
• Search for resources in space for export to Earth
• Search for any potential danger to Earth from space
• Understand the origins of the Solar System
• Don’t know
Now, please think more specifically about future space exploration missions.
4. What is the mission that you think is the most important for space agencies to accomplish in
the next decade?
• Mercury Mission
• Venus Mission
• Moon Mission
• Mars Mission
64

•
•
•
•
•
•
•
•
•
•
•
•

Phobos Mission
Comet Mission
Asteriod Mission
Jupiter Mission
Io Mission
Europa Mission
Saturn Mission
Titan Mission
Uranus Mission
Neptune Mission
Pluto/Kuiper Belt Mission
Don’t know

5. There has been much discussion about attempting to land an astronaut on the planet Mars.
How would you feel about United States setting aside money for such a project?
• Strongly favor
• Favor
• Somewhat favor
• Neither favor nor oppose
• Somewhat oppose
• Oppose
• Strongly oppose
• Don’t know
Now, thinking about the U.S. space program.
6. How would you rate the job being done by NASA – the U.S. space agency?
• Excellent
• Good
• Fair
• Poor
• Don’t know
7. It is now 40 years since the United States first landed men on the moon. Do you think the
space program has brought enough benefits to this country to justify its costs, or don’t you think
so?
• The space program has brought enough benefits to justify its costs.
• The space program has not brought enough benefits to justify its costs.
• Don’t know
8. Do you think spending on the U.S. space program should be increased, kept at the present
level, reduced, or ended altogether?
65

•
•
•
•
•

Increased
Kept at the present level
Reduced
Ended altogether
Don’t know

9. How should space exploration be funded?
• All funding should come from the private sector.
• Governments should continue to provide the bulk of the funds for space exploration.
• We should have increases in our taxes to fund space exploration.
• We should have decreases in other public spending areas to fund space exploration.
• Don’t know
10. If you could advise the next president about the U.S. space program, what would you say?
• Maintain current activities, but do not commit to new projects
• Seek international cooperation for lunar landings
• Seek international cooperation for Mars exploration
• Build a lunar base as a stepping-stone to Mars
• End the space program altogether
• Don’t know
Copyrighted Music File-sharing
First, we’d like you to think about your online music downloading habits.
For each of the following questions, please select the answer that best represents your views.
1. How often do you download music from the Internet?
• Never
• Almost never
• Sometimes
• Almost every day
• Every day
• Don’t know
2. How much do you care whether the music you download onto your computer is copyrighted?
• Do not care at all
• Slightly care
• Somewhat care
• Moderately care
• Care a great deal
• Don’t know
66

3. How much do you care whether or not the files you allow others to download from your
computer are copyrighted?
• Do not care at all
• Slightly care
• Somewhat care
• Moderately care
• Care a great deal
• Don’t know
4. In your opinion, is there anything wrong with downloading copyrighted music from the
Internet?
• Yes
• No
• Don’t know
5. How would you feel about allowing people to legally download music from up-and-coming
artists for free?
• Strongly oppose
• Oppose
• Somewhat oppose
• Neither oppose nor support
• Somewhat support
• Support
• Strongly support
• Don’t know
6. How would you feel about allowing people to legally download music from established artists
for free?
• Strongly oppose
• Oppose
• Somewhat oppose
• Neither oppose nor support
• Somewhat support
• Support
• Strongly support
• Don’t know
Now, we’d like to know your thoughts about statements people sometimes make about
downloading copyrighted music.
7. To what extent do you agree or disagree with the following statement, “Music downloading
and file-sharing is so easy to do, it’s unrealistic to expect people not to do it”?
67

•
•
•
•
•
•
•
•

Strongly disagree
Disagree
Somewhat disagree
Neither agree nor disagree
Somewhat agree
Agree
Strongly agree
Don’t know

8. To what extent do you agree or disagree with the following statement, “As long as people are
still buying music, it’s okay if they download or share some copyrighted music for free”?
• Strongly disagree
• Disagree
• Somewhat disagree
• Neither agree nor disagree
• Somewhat agree
• Agree
• Strongly agree
• Don’t know
9. To what extent do you agree or disagree with the following statement, “It’s never really okay
to download or share copyrighted music files online without paying for them”?
• Strongly disagree
• Disagree
• Somewhat disagree
• Neither agree nor disagree
• Somewhat agree
• Agree
• Strongly agree
• Don’t know
Now, thinking about most people…
10. What do you think prevents most people from downloading copyrighted music?
• Fear of downloading a virus
• Fear of legal trouble
• Spyware
• The person’s morals
• Don’t know

68

Appendix B
Outcome-relevant Involvement Scale
Consider the topic of _____ and how it might affect you. Please answer each of the following
questions.
1. _____ would affect me personally.
2. _____would have a big impact on my life.
3. _____ would have important consequences for me.
4. _____ would have little effect on me personally.*

*This item was reverse scored.

69

Appendix C
Value-relevant Involvement Scale
Please consider the topic of ______ and how it fits with your own values. Then, please answer
each of the following questions.
1. My attitude toward _____ is based on my core values.
2. My position on _____ reflects who I am.
3. Arguments for or against _____ are relevant to the core principles that guide my life.
4. Knowing my position on _____ is central to understating the kind of person I am.

70

Appendix D
Piper Fatigue Scale
To what degree are you now feeling:
1. Strong/Weak
2. Awake/Sleepy
3. Lively/Listless
4. Refreshed/Tired
5. Energetic/Unenergetic
6. Patient/Impatient
7. Relaxed/Tense
8. Exhilarated/Depressed
9. Able to concentrate/Unable to concentrate
10. Able to remember/Unable to remember
11. Able to think clearly/Unable to think clearly

71

Appendix E
Self-expression Scale
1. How much did you feel like you shared your thoughts?
2. How much do you feel like you communicated your views in a meaningful way?
3. How much do you feel like you accurately expressed your views?

72

Appendix F
Enjoyment Scale
4. How much did you enjoy completing this survey?
5. How much did you like sharing your thoughts in this study?
6. How much did you dislike participating in the survey?*

*This item was reverse scored, then dropped from the scale due to reduced reliability.

73

Appendix G
Knowledge Scale
Compared to an expert on the topic, please rate:
1. Your knowledge on _____.
2. How great of an understanding you have about _____.
3. How much data you can recite about _____.

74

Appendix H
Writing Apprehension Scale
1. I avoid writing.
2. I have no fear of my writing being evaluated.*
3. I look forward to writing down my ideas.*
4. My mind seems to go blank when I start to work on a composition.
5. Expressing ideas through writing seems to be a waste of time.
6. I would enjoy submitting my writing to magazines for evaluation and publication.*
7. I like to write my ideas down.*
8. I feel confident in my ability to clearly express my ideas in writing.*
9. I like to have my friends read what I have written.*
10. I am nervous about writing.
11. People seem to enjoy what I write.*
12. I enjoy writing.*
13. I never seem to be able to clearly write down my ideas.
14. Writing is a lot of fun.*
15. I like seeing my thoughts on paper.*
16. Discussing my writing with others is an enjoyable experience.*
17. It is easy for me to write good compositions.*
18. I don't think I write as well as most other people do.
19. I don't like my compositions to be evaluated.
20. I am no good at writing.

* These items were reverse scored.
75

REFERENCES

76

REFERENCES

Aaronson, L.S., Teel, C. S., Cassmeyer, V., Neuberger, G. B., Pallikkathayil, L., Pierce, J., Press,
A. N., Williams, P. D., & Wingate, A. (1999). Defining and measuring fatigue. Journal
of Nursing Scholarship, 31(1), 45-50.
Amodio, D. M., Jost, J. T., Master, S. I., & Yee, C. M. (2007). Neurocognitive correlates of
liberalism and conservativism. Nature, 10(10), 1246-1247.
Antil, J. H. (1984). Conceptualization and operationalization of involvement. In T. C. Kinnear
(Ed.), Advances in consumer research (vol. 11). Provo, UT: Association for Consumer
Research.
Bargh, J. A., McKenna, Y. A., and Fitzsimons, G. M. (2002). Can you see the real me?
Activation and expression of the “true self” on the Internet. Journal of Social Issues,
58(1), 33-48.
Bogen, K. (1996). The effects of questionnaire length on response rates: A review of the
literature. Retrieved from U.S. Bureau of the Census, November 28, 2009
http://www.census.gov/srd/papers/pdf/kb9601.pdf
Borle, S., Dholakia, U. M., Singh, S., & Westbrook, R. (2007). The impact of survey
participation on subsequent customer behavior: An empirical investigation. Marketing
Science, 26(5), 711-726.
Campbell, D. T., & Stanley, J. C. (1963). Experimental and quasi-experimental designs for
research. Chicago: Rand McNally College Publishing Company.
Carlson, J. P., Vincent, L. H., Hardesty, D. M., & Bearden, W. O. (2009). Objective and
subjective knowledge relationship: A quantitative analysis of consumer research findings.
Journal of Consumer Research, 35, 864-876.
Champion, D. J., & Sear, A. M. (1969). Questionnaire response rate: A methodological analysis.
Social Forces, 47(3), 335-339.
Charlton, B. G., M.D. (n.d.). Treatment of chronic fatigue with analgesics. Retrieved November
28, 2009 from http://www.hedweb.com/bgcharlton/fatigue.html.
Cho, H., & Boster, F. J. (2005). Development and validation of value-, outcome-, and
impression-relevant involvement scales. Communication Research, 32, 235-264.
Converse, P. E. (1964). The nature of belief systems in mass publics. Critical Review, 18, 1-74.
77

de Heer, W. F., & Israels, A. Z. (1992). Response trends in Europe. ASA Proceedings of the
Section on Survey Research Methods, 92-101.
Daly, J. A., & Miller, M. D. (1975). The empirical development of a scale to measure writing
apprehension. Research in the Teaching of English, 9(3), 242-249.
Dillman, D. A. (1978). Mail and telephone surveys: The total design method. New York: John
Wiley & Sons.
Dillman, D. A., Sinclair, M. D., & Clark, J. R. (1993). Effects of questionnaire length,
respondent-friendly design, and a difficult question on response rates for occupantaddressed census mail surveys. Public Opinion Quarterly, 57(3), 289-304.
Dillman, D. A., Smyth, J., & Christian, L. M. (2009). Internet, mail and mixed-mode surveys:
rd
The total design method (3 ed.). Hoboken, NJ: John Wiley Co.
Dholakia, U. M., & Morwitz, V. G. (2002). The scope and persistence of mere-measurement
effects: Evidence from a field study of customer satisfaction measurement. Journal of
Consumer Research, 29, 159-167.
Feldman, J. M., & Lynch, J. G., Jr. (1988). Self-generated validity and other effects of
measurement on belief, attitude, intention and behavior. Journal of Applied Psychology,
73, 421-435.
Fitzsimons, G. J., & Morwitz, V. M. (1996). The effect of measuring intent on brand level
purchase behavior. Journal of Consumer Research, 23, 1-11.
Fitzsimons, G. J., & Shiv, B. (2001). Nonconscious and contaminative effects of hypothetical
questions on subsequent decision making. The Journal of Consumer Research, 28(2),
224-238.
Galesic, M., & Bosnjak, M. (2009). Effects of questionnaire length on participation and
indicators of response quality in a web survey. Public Opinion Quarterly, 73(2), 349-360.
Godin, G., Sheeran, P., Conner, M., & Germain, M. (2008). Asking questions changes behavior:
Mere measurement effects on frequency of blood donation. Health Psychology, 27(2),
179-184.
Goffman, E. (1959). The presentation of self in everyday life. New York: Doubleday.
Goyder, J. C. (1982). Further evidence on factors affecting response rates to mailed
questionnaires. American Sociological Review, 47, 550-553.
Greer, J. G. (1991). Do open-ended questions measure “salient” issues? Public Opinion
Quarterly, 55, 360-370.
78

Greenberg, M.A.,Wortman, C.B., & Stone, A.A. (1996). Emotional expression and physical
health: Revising traumatic memories or fostering self-regulation? Journal of Personality
and Social Psychology, 71(3), 588– 602.
Greenwald, A. G., Carnot, C. G., Beach, R., & Young, B. (1987). Increased voting behavior by
asking people if they expect to vote. Journal of Applied Psychology, 72, 315-318.
Heberlein, T. A., & Baumgartner, R. (1978). Factors affecting response rates to mailed
questionnaires: A quantitative analysis of the published literature. American Sociological
Review, 43, 447-462.
Herzog, A. R., & Bachman, J. G. (1981). Effects of questionnaire length on response quality.
Public Opinion Quarterly, 45(4), 549-559.
Higgins, E. T. (1987). Self-discovery theory. Psychological Review, 94, 1120-1134.
Hogan, J. M. (1997). George Gallup and the rhetoric of scientific discovery. Communication
Monographs, 64(2), 161-179.
Howe, N., & Strauss, W. (2000). Millennials rising: The next great generation. New York:
Vintage Books.
Ivis, F. J., Bondy, S. J., & Adlaf, E. M. (1997). The effect of question structure on self-reports of
heavy drinking: Closed-ended versus open-ended questions. Journal of Studies on
Alcohol, 58, 622-624.
Johnson, B. T., & Eagly, A. H. (1989). Effects of involvement on persuasion: A meta-analysis.
Psychological Bulletin, 106(2), 290-314.
Johnson, B. T., & Eagly, A. H. (1990). Involvement and persuasion: Types, traditions, and the
evidence. Psychological Bulletin, 107(3), 375-384.
Jung, C. G. (1953). Two essays in analytical psychology (trans. R. F. Hull). New York:
Pantheon.
Katz, D. (1960). The functional approach to the study of attitudes. Public Opinion Quarterly,
24(2), 163-204.
Kealy, M. J., & Turner, R. W. (1993). A test of the equality of closed ended and open ended
contingent valuations. American Journal of Agricultural Economics, 75(2), 321-331.

79

Kumashiro, M., & Nagae, S. (1984). Workers’ subjective feeling of fatigue and attitudes towards
work: effects of age and job difference. Journal of University Occupational and
Employment Health, 6(3), 273-81.
Levav, J., & Fitzsimons, G. J. (2006). When questions change behavior: The role of ease of
representation. Psychological Science, 17(3), 207-213.
MacKenzie, S. B., Lutz, R. J., & Belch, G. E. (1986). The role of attitude toward the ad as a
mediator of advertising effectiveness: A test of competing explanations. Journal of
Marketing Research, XXIII, 130-143.
Maio, G. R., & Olson, J. M. (2000). What is a value-expressive attitude? In G. R. Maio & J. M.
Olson (Eds.), Why we evaluate: Functions of attitudes (pp. 249-269). Mahwah, NJ:
Erlbaum.
Mitchell, R. C., & Carson, R. T. (1989). Using surveys to value public goods: The contingent
valuation method. Washington, D.C.: Resources for the Future.
Morwitz, V. G. (2005). The effect of survey measurement on respondent behavior. Applied
Stochastic Models in Business and Industry, 21, 451-455.
Oskamp S., & Schultz, P. W. (2005). Attitudes and opinions. (ed. 3). New York: Lawrence
Erlbaum Associates.
Park, H.S., Levine, T. R., Kingsley-Westerman, C. Y., Orfgen, T., & Foregger, S. (2007). The
effects of argument quality and involvement type on attitude formation and attitude
change: A test of dual-process and social judgment predictions. Human Communication
Research, 33, 81-102.
Pennebaker, J. W., & Seagal, J. D. (1999). Forming a story: The health benefits of narrative.
Journal of Clinical Psychology, 55(10), 1243-1254.
Pennebaker, J.W. (1997). Opening up: The healing power of emotional expression. New York:
Guilford.
Petty, R. E., & Cacioppo, J. T. (1981). Attitudes and persuasion: Classic and contemporary
approaches. Dubuque, IA: William C. Brown.
Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement as a determinant of
argument-based persuasion. Journal of Personality and Social Psychology, 41, 847-55.
Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to
advertising effectiveness: The moderating role of involvement. Journal of Consumer
Research, 10, 135-146.
80

Piper, B. F., Dibble, S. L., Dodd, M. J., Weiss, M. C., Slaughter, R. E., & Paul, S. M. (1998).
The revised Piper Fatigue Scale: Psychometric evaluation in women with breast cancer.
Oncology Nursing Forum, 25(4), 677-684.
Rogers, C. (1951). Client-centered therapy. Boston: Houghton-Mifflin.
Rogelberg, S. G., Fisher, G. G., Maynard, D. C., Hakel, M. D., & Horvath, M. (2001). Attitudes
toward surveys: Development of a measure and its relationship to respondent behavior.
Organizational Research Methods, 4(3),3-25.
Rokeach, M. (1973). The nature of human values. New York: Free Press.
Schuman, H. (1966). The random probe: A technique for evaluating the validity of closed
questions. American Sociological Review, 31(2), 218-222.
Schuman, H., & Presser, S. (1981). Questions and answers in attitude surveys: Experiments in
question form, wording, and context. New York: Academic Press.
Schwarz, N. (1996). Cognition and communication: Judgmental biases, research methods and
the logic of conversation. Hillsdale, NJ: Erlbaum.
Schwarz, N. (1999). Self-reports: How the questions shape the answers. American Psychologist,
54, 93-105.
Schwarz, N., Hippler, H., & Noelle-Neuman, E. (1992). A cognitive model of response-order
effects in survey measurement. In N. Schwarz & S. Sudman (Eds.), Context effects in
social and psychological measurement (pp. 187-201). New York: Springer-Verlag.
Sherif, M. & Hovland, C. I. (1961). Social judgment: Assimilation and contrast effects in
communication and attitude change. New Haven: Yale University Press.
Shumacher, G. M., & Nash, J. G. (1991). Conceptualizing and measuring knowledge change due
to writing. Research in the Teaching of English, 25(1), 67-96.
Sjoberg, G. (1954). A questionnaire on questionnaires. Public Opinion Quarterly, 18(4), 423427.
Snyder, M. (1974). Self-monitoring of expressive behavior. Journal of Personality and Social
Psychology, 30(4), 526-537.
Spera, S.P., Buhrfeind, E.D., & Pennebaker, J.W. (1994). Expressive writing and coping with job
loss. Academy of Management Journal 37(3), 722–733.

81

Swan, J. E., Trawick, I. F., & Carroll, M. G. (1981). Effect of participation in marketing research
on customer attitudes toward research and satisfaction with a service. Journal of
Marketing Research, 18(3), 356-363.
Tourangeau, R., Rips, L. J., & Rasinski, K. (2000). The psychology of survey response.
Cambridge University Press.
Wing, J. F., Schutte, N. S., & Byrne, B. (2006). The effect of positive writing on emotional
intelligence and life satisfaction. Journal of Clinical Psychology, 62(10), 1291-1302.
Zaller, J., & Feldman, S. (1992). A simple theory of the survey response: Answering questions
versus revealing preferences. American Journal of Political Science 36(3), 579-616.

82