Empirical Methods in Communication
Application Assignment #2
Experimental Research
Overview: You will briefly answer five essay questions that require you to dissect a published
experiment by McGlone et al (attached). Your answers must be typed and double spaced.
Provide a 1½ inch margin on all sides of the page for written comments. Your paper must
absolutely not be longer than 1000 words; that is about 200 words per question. All word
processors have a word count feature. Please provide the word count of your essay (excluding
your name and other identifying information).
Due Date: February 15. Turn your paper in at the beginning of the class session. Electronic
(email attachment) submissions will not be accepted.
Essay Questions
1. McGlone and his colleagues used a 2×2 factorial design. Explain why this design was
necessitated by the research hypotheses.
2. Experimental control is a critical feature of any experiment. If we wish to know the effects
of independent variables on dependent variables we must “hold constant” other variables
that would make causal inference challenging. What efforts were taken in this experiment
to achieve a high level of experimental control?
3. What efforts were made to determine if the manipulations of the two independent variables
were successful? What other strategies might have been employed to check on the quality
of the manipulations?
4. Verbal assignment of agency had a much more substantial effect than visual (image-
based) agency assignment. Carefully examine the two images used to manipulate agency
visually (Figures 1 and 2) and speculate on why the agentic imagery did not work in this
experiment.
5. Speculate on how the information gained from this study about linguistic agency could be
used in advertising.
Evaluation
Your answers to these five questions will be equally weighted. Evaluation is based on the
thoughtfulness of your response and the quality of your writing.
Think of this essay as a take home exam. The work you submit must be your own. You may
not discuss this assignment or your ideas with other people. Doing so will be considered
cheating.
Address correspondence to Matthew S. McGlone, Department of Communication Studies,
University of Texas at Austin, 1 University Station A1105, Austin, TX 78712, USA. E-mail:
matthew_mcglone@mail.utexas.edu
1
Journal of Health Communication, 0:1–17, 2012
Copyright © Taylor & Francis Group, LLC
ISSN: 1081-0730 print/1087-0415 online
DOI: 10.1080/10810730.2012.727950
Don’t Let the Flu Catch You: Agency Assignment
in Printed Educational Materials About the
H1N1 Influenza Virus
MATTHEW S. M C GLONE
Department of Communication Studies,
University of Texas at Austin, Austin, Texas, USA
ROBERT A. BELL
Department of Communication, University of California
at Davis, Davis, California, USA
SARAH T. ZAITCHIK AND JOSEPH M C GLYNN III
Department of Communication Studies,
University of Texas at Austin, Austin, Texas, USA
In English and in other languages, the agency for viral transmission can be
grammatically assigned to people (e.g., Thousands may contract H1N1) or to the
virus itself (e.g., H1N1 may infect thousands). These assignment options shape
different conceptions of transmission as attributable either to social contact within
one’s control or to pursuit of an active predator. The authors tested the effect
of agency assignment and agentic images on young adults’ (N = 246) reactions
to educational materials about H1N1 influenza. The authors hypothesized that
assigning agency to the virus would heighten perceived severity and personal
susceptibility relative to human agency assignment. Results were consistent with
this hypothesis, indicating that virus agency increased perceptions of severity,
personal susceptibility, and reported intentions to seek vaccination relative to
human agency. The image manipulation did not directly affect these factors. The
findings suggest that strategic agency assignment can improve the effectiveness of
educational materials about influenza and other health threats.
Health care providers and health educators alike rely extensively upon printed
educational materials (PEMs) to inform people about health threats, healthy
lifestyles, and treatments for medical conditions (Bull, Holt, Kreuter, Clark, &
Sharff, 2001; Schwartzberg, Cowett, & VanGeest, 2007). Such materials can take the
form of physical leaflets and posters or of digital documents disseminated by e-mail
and the Internet (Leiner, Handal, & Williams, 2004; Taylor, 2008; Wallace, Rogers,
Turner, Keenum, & Weiss, 2006). Despite their widespread use, PEMs usually have
modest effects on behavior (e.g., Harris, Smith, & Veale, 2005). The reasons for these
Downloaded by [University of California Davis] at 17:00 14 December 2012
2 M. S. McGlone et al.
disappointing results are well-known. Many PEMs lack cultural sensitivity (e.g.,
Guidry & Walker, 1999) and are written at reading levels beyond the capabilities of
low-literate and general audiences (Ache & Wallace, 2009; Wallace et al., 2008; Weih
et al., 2008). Format guidelines, development procedures, and evaluation methods
have been advanced in recent years to enhance PEM usability (Hoffman & Worrall,
2004; Holmes-Rovner, 2007; Wright, 2003). Especially promising is the development
of computer-generated PEMs tailored to the unique interests and format preferences
of consumers and patients (e.g., Strecher et al., 2005).
What is often missing from discussions about PEM development is an
explicit recognition that these materials must be both informative and persuasive.
Whittingham, Ruiter, Castermans, Huiberts, and Kok (2008) reported that revising a
health brochure about alcohol abuse on the basis of principles of cognitive psychology
increased message comprehension as intended, but did not affect behavioral intentions.
They concluded that the development of easily comprehended messages “should
be complemented by determinant-specific change methods if behavior change is to
occur” (p. 414). PEMs should, for example, be attributed to credible sources, make
use of recognized experts, offer evidence-based treatment explanations, and use fear
appeals to motivate action (Cohen, Colligan, & Berger, 1985). Selander, Troein,
Finnegan, and Rastam (1997) have called for a greater emphasis on the rhetoric of
PEMs, highlighting the need to give due consideration to logos, ethos, and pathos.
There is a clear need for research on specific design elements that can enhance
the persuasiveness of PEMs and complement research focused on instructional
design (Kools, 2007). In this spirit, we examined a message strategy grounded in
psycholinguistics that was intended to enhance the motivational force of PEMs.
Specifically, we evaluated the effect of linguistic assignments of agency to either a
health threat, the H1N1 virus, or to individuals at risk. The study was carried out in
October 2009. At that time the virus was spreading quickly, frightening news reports
about the pandemic were being issued daily, and a vaccine had been developed but
not yet released to the public. Study participants were college-aged adults, a relevant
population given that H1N1 has disproportionately affected people under 25 years of
age (Centers for Disease Control & Prevention, 2009).
Languages permit different lexical and grammatical choices for describing an
event, and by their choices speakers convey different beliefs and attitudes about
the event to audiences. One of these choices is the assignment of agency – i.e., the
ascription of action or change to one or more event-related entities (Dowty, 1991).
Although agency is typically assigned to animate, living entities (e.g., the dog barked),
assignments to inanimate and abstract agents are common and easy to overlook
(McGlone & Pfiester, 2009). For example, English and most other languages permit
speakers to figuratively assign the agency for time passage to humans (e.g., we are
approaching the weekend) or to an event (the weekend is approaching). Conventional
discourse about viral transmission permits an analogous set of alternatives for
assigning agency, either to humans (e.g., thousands of young adults have contracted
the virus) or to the virus itself (the virus has infected thousands of young adults). The
availability of the different assignment choices reflects the multiple causality of
transmission, by human interaction on the one hand and viral replication on the
other. Less obvious, however, is the implicit effect that assignment choice may have
on people’s perceptions of viral threat and how to protect themselves from it.
We explored the effects of human and viral agency assignments in a printed fact
sheet about H1N1 on people’s concerns about the virus and beliefs about vaccination
Downloaded by [University of California Davis] at 17:00 14 December 2012
Agency Assignment and H1N1 Transmission 3
efficacy. Agency assignment linguistically encodes the causality for an event, and
assigning causality to a non-human agent implicitly conveys human passivity in the
event (Dowty, 1991), with potential negative implications (McGlone & Pfiester, 2009).
We hypothesized that consistently assigning agency to the virus would encourage
readers to conceive the virus as a predator seeking out human victims, and thus
heighten the perceived severity of the health threat it posed, as well as their perceived
susceptibility to it. In contrast, assigning agency to humans was expected to shape
a conception of the virus not as a combatant but rather as a consequence of social
contact within readers’ control, which in turn would result in lower perceived threat
and susceptibility.
How will the hypothesized effects of agency influence people’s intentions to
engage in precautionary behaviors for avoiding infection, such as vaccination?
Theories about the effect of fear-arousing messages, such as the Extended Parallel
Process Model, suggest that people perform two distinct appraisals of fear appeals
(Witte, 1994). They initially assess the perceived severity of and susceptibility to
the threat. If they believe it to be severe and feel personally vulnerable to it, they
then evaluate the efficacy of the recommended course of action in reducing the
threat (response efficacy) and their ability to enact the response (self-efficacy). High
perceived severity and susceptibility, when accompanied by high efficacy, should lead
the person to control the danger by adopting the recommendations (e.g., by getting
vaccinated). In contrast, when severity and susceptibility loom large and response or
self-efficacy is low, they are more focused on controlling their fear by suppressing
thoughts about the danger, attacking the message, and other defensive responses. This
framework leads to the expectation that a PEM about H1N1 will be most effective in
promoting precautionary behavior when it induces perceptions of severity, personal
susceptibility, and high response-efficacy and self-efficacy.
Research to date has focused on agency assignments made by text. In the
present study, we advanced our understanding of agency assignment strategies by
examining the effects of text- and image-based assignments. For example, the H1N1
virus can be represented literally with a photograph of the virus as seen under an
electron microscope or with an image that metaphorically highlights its agency (e.g., a
depiction of the virus as a vicious monster). Visual imagery conveys agency with less
subtlety or nuance than linguistic assignment (Jackendoff, 1990), and imagery can
evoke fear that is more immediate and enduring than linguistic descriptions in certain
health message contexts (Cameron & Chan, 2008). In the present study, we compared
the effects of a literal image of H1N1 with an agentic image that represented the virus
as a creature with sharp teeth. Because linguistic grammatical structure and visual
imagery are processed through distinct cognitive and neural mechanisms (Kosslyn,
Ganis, & Thompson, 2001; Pinker & Jackendoff, 2005), and there was no rationale a
priori for why linguistic and image-based representations of agency would interact,
we expected their effects to be additive, not multiplicative.
Method
Participants
A total of 246 undergraduate students at a large public university in the United States
were recruited from undergraduate courses in communication for a study of “beliefs
and attitudes about the H1N1 virus.” Data collection commenced on October 6,
2009, and concluded on October 24, 2009, approximately a month before the H1N1
Downloaded by [University of California Davis] at 17:00 14 December 2012
4 M. S. McGlone et al.
vaccine became available to students on the university campus. Data were collected
in small groups of 5–20 noninteracting students in a single experimental session held
in a campus classroom. Upon arrival at the session, student volunteers were told that
participation in the study entailed reading a fact sheet about the H1N1 virus and then
completing a questionnaire.
Experimental Design and Stimulus Materials
Participants were randomly assigned to one of four experimental conditions.
Participants in each condition received a booklet comprised of a single-page fact sheet
clipped to a questionnaire. The information in the sheet was taken from the Centers
for Disease Control and Prevention H1N1 website and was attributed to a fictitious
association of university health centers. Four versions of the sheet were created by
independently manipulating two characteristics of its composition in a 2 × 2 factorial
design: agency assignment language (human agency vs. virus agency) and virus image
(literal vs. agentic). For illustrative purposes the fact sheets for the virus agency
language/agentic image and human agency/literal image experimental conditions are
presented in Figures 1 and 2.
For the agency assignment language manipulation, text used to describe viral
transmission employed idiomatic expressions that assign the agency for viral
transmission either to the virus itself or to humans. The 22 idiomatic expressions that
encoded human agency and virus agency are shown in Table 1. The human agency
and virus agency versions of the sheet consisted of 505 and 508 words, respectively,
indicating that efforts to hold message length constant were largely successful. The
readability levels of the human and flu agency versions were calculated with an online
readability score tool, using five formulas: the Flesch-Kincaid Grade Level, Gunning-
Fog Score, Coleman-Liau Index, SMOG Index, and Automated Readability Index
(Child, 2008). The average grade levels obtained across these five measures were 8.2
and 8.6 for the virus and human agency versions, respectively.
The virus image manipulation was carried out by altering a single image in the top
left column of the fact sheet. The agentic image was selected from an online repository
of clipart images (http://www.fundraw.com). It depicted H1N1 as a lamprey-like
creature with a mouth and sharp teeth (see Figure 1). This image was chosen over
others because it portrayed the virus as a vicious living thing but did not include
cartoonish animal features (e.g., eyes, feet) or human emotional expressions (e.g.,
anger, scorn) that, in our estimation, could have detracted from its fear-arousing
potential. The literal image was a high resolution photograph of the virus downloaded
from the Centers for Disease Control and Prevention website (http://www.cdc.gov/
H1N1flu/images.htm; see Figure 2).
Questionnaire and Measures
Dependent measures and manipulation checks were incorporated into a questionnaire.
The initial sections included 7-point Likert-type items designed to assess participants’
perceptions of susceptibility to H1N1 infection and the severity of infection; their
affective experiences while reading the fact sheet; their perceptions of the effectiveness
of the vaccine (response efficacy); their abilities to become vaccinated (self-efficacy);
and their intentions to get vaccinated. In addition, participants rated their subjective
likelihood of getting the H1N1 vaccine on a 0–100% probability scale. These items were
adapted loosely from previous Extended Parallel Process Model research (Witte, Meyer,
Downloaded by [University of California Davis] at 17:00 14 December 2012
Agency Assignment and H1N1 Transmission 5
Table 1. Agency assignment language embedded in the human agency and virus
agency versions of the H1N1 virus fact sheet
Position Human agency Virus agency
Headline We’re Approaching Flu Season … The Flu Season is Approaching …
Caption Thousands of Americans May Die
from H1N1
H1N1 May Kill Thousands Of
Americans
Body … new influenza virus that people
spread through coughing or
sneezing.
… new influenza virus that spreads
itself from person to person
through coughing or sneezing
Body You can also contract it when you
touch …
It can also infect you when you
touch …
Body … half of the U.S. population will
contract H1N1 …
… H1N1 will infect half of the U.S.
population …
Body People with H1N1 exhibit symptoms
such as fever …
Victims of H1N1 exhibit symptoms
such as fever …
Body Many people will get sick … The virus will make many people
sick …
Body … some people will get seriously ill … … it will make some people seriously
ill …
Body … and a small percentage of those
who contract it will die.
… and will kill a small percentage of
those it infects
Body In the U.S., we will officially start the
flu season in October …
In the U.S., the flu season will
officially start in October …
Body … but people can contract H1N1 in
any month of the year
… but H1N1 can strike in any
month of the year
Body Younger adults are more likely to
contract the virus than older adults
The virus is more likely to attack
younger than older adults
Body Because younger adults are especially
susceptible …
Because the virus prefers younger
adults …
Body … university communities are at risk
for widespread viral outbreaks
… university communities are at risk
for widespread viral attacks
Text box The H1N1 virus is more prevalent
among younger than older adults
The H1N1 virus prefers infecting
younger than older adults
Body Only an H1N1 vaccine can prevent
you from contracting this new virus
Only an H1N1 vaccine can prevent
you from being infected by this
new virus
Text box H1N1 cases may be on the rise, but
effective vaccines are on the way!
H1N1 may be on the prowl, but
effective vaccines are on the way!
Heading Things You Can Do To Avoid
Getting H1N1
Things You Can Do To Outsmart
H1N1
Body … there are things you can do to
avoid getting sick
… there are things you can do to
avoid becoming its prey
Body You can protect against H1N1 by
washing your hands often …
You can outwit H1N1 by washing
your hands often …
Body … and avoid close contact with sick
people who may have contracted a
flu virus
… and avoid close contact with sick
people who may have become
infected by a flu virus
Footer Crowds Can Promote H1N1
Contraction
The H1N1 Virus Loves Crowds
Downloaded by [University of California Davis] at 17:00 14 December 2012
6 M. S. McGlone et al.
& Martell, 2001). We also assessed the possibility of elevated fear leading participants to
respond defensively by derogating the message. Thirteen 11-point (–5 to +5) semantic
differential items probing participants’ perceptions of the fact sheet’s credibility were
included to measure any potential message or source derogation. To probe participants’
suspicions, they were asked to speculate about the researchers’ objectives in an open-
ended question. Memory-based manipulation checks were included that probed
participants’ recognition memory for the agency assignment language and the virus
image presented in the fact sheet. In five dichotomous questions, pairs of sentences were
presented conveying the same proposition using human and virus agency assignment,
only one of which appeared in the fact sheet any given participant had read. Participants
were asked to choose the sentence they had seen previously; their recognition accuracy
in this task served as an index of their sensitivity to our linguistic agency manipulation.
One multiple-choice item measured memory for the virus image. Several filler items
were included as well to obscure the purpose of this memory check. The questionnaire
concluded with a set of standard demographic and health status measures and with
an invitation to be notified when the H1N1 vaccine became locally available. This
invitation took the form of a tear sheet, allowing us to link the participant’s responses in
the fact sheet to their desire to learn more about the vaccine.
Procedure
The experiment proceeded in three phases. In the initial study phase, participants
were randomly assigned to one of the four experimental conditions and given a packet
consisting of the condition-appropriate version of the fact sheet and the questionnaire.
Figure 1. Fact sheet version presented to participants assigned to the virus agency/agentic
image experimental condition. (Color figure available online.)
Downloaded by [University of California Davis] at 17:00 14 December 2012
Agency Assignment and H1N1 Transmission 7
The experimenter instructed participants to remove the sheet from the packet and study
its contents for 5 minutes. At the end of this period, the experimenter collected the sheets
from participants and instructed them to begin work on the questionnaire. Participants
took 15 minutes on average to complete the questionnaire. When they were finished,
they returned the completed questionnaire to the researcher. Those who elected to
receive an e-mail notification about vaccine availability turned in their notification tear
sheets in a stack separate from the completed questionnaires. In the final debriefing
phase, participants were told the purpose and hypotheses of the experiment and offered
the opportunity to pose any questions they had to the experimenter.
Results
Of the 246 participants who completed the study, data from 11 students who had
already contracted the H1N1 virus were discarded because patients with a confirmed
case of H1N1 are often told they no longer need vaccination. The data of an
additional 13 students who had correct suspicions about our research objectives
were also discarded. Thus, final analyses were based on a sample of 222 students. A
majority of research participants (71.9%) were female. White race was reported by
64.9%. Hispanic cultural identity was claimed by 17.1% of participants of all races.
The mean age was 20.4 years (SD = 2.82). A general health self-rating of “excellent,”
“very good,” or “good” was provided by 95.7% of participants, 55.9% had a regular
source of medical care, and 64.4% had received a seasonal flu vaccine in the previous
12 months. Initial analyses did not reveal any significant main effects or interactions
Figure 2. Fact sheet version presented to participants assigned to the human agency/literal
image experimental condition. (Color figure available online.)
Downloaded by [University of California Davis] at 17:00 14 December 2012
8 M. S. McGlone et al.
involving the aforementioned demographic variables, so subsequent analyses
collapsed across these factors.
Factor Analysis of Items and Scale Reliabilities
Questionnaire items were subjected to a principal component factor analysis with
varimax rotation in order to obtain discrete factors with minimal overlap. This
procedure resulted in eight factors: source credibility (5 items, explaining 27.0%
of the variance), information quality (3 items, 13.1%), self-efficacy (3 items, 9.2%),
fear arousal (3 items, 8.5%), severity (3 items, 5.9%), susceptibility (3 items, 5.5%),
response efficacy (3 items, 5.1%), and vaccination intention (2 items, 3.7%). The two
items comprising the vaccination intention factor were measured on different scales (a
7-point Likert-type scale and a 0–100% probability estimate), and consequently were
standardized to compute the factor’s internal consistency and scale scores. Factor
reliabilities ranged from α = 0.93 (source credibility) to 0.77 (response efficacy). The
items comprising each factor were averaged into an index for subsequent analysis.
These items, the factor reliabilities, and index means (and standard deviations) are
reported in Table 2.
Manipulation Checks
Analysis of questionnaire items probing recognition for information in the fact sheet
indicated that participants attended to the manipulated language and imagery while
reading the sheet. Participants assigned to human agency or virus agency language
versions of the sheet achieved 84.2% and 87.6% accuracy, respectively, on the five
wording recognition probes, t(220) = .67, p = .252. The single item probing recognition
of the image presented in the sheet was accurately answered by 98.4% of participants
assigned to the literal image version and 97.5% of those assigned to the agentic version.
Removing participants who did not achieve 100% accuracy on the memory probes did not
significantly alter our findings, so we report analyses on the basis of the complete sample.
Source Credibility and Information Quality
Agency assignment language significantly influenced the source credibility index,
F(1, 218) = 4.58, p = .033, η 2 = .022. Specifically, participants who read the virus agent
fact sheet version had higher index ratings (M = 3.00, SD = 1.66) than did those who
read the human agent version (M = 2.49, SD = 1.80). Higher ratings on this index
indicate higher degrees of perceived responsibility, fairness, and ethical intent of the
information presented in the sheet. However, the language manipulation did not
reliably influence perceptions of the informativeness or believability of the sheet, as
measured by the information quality index, F(1, 218) = 1.51, p = .223, η 2 = .001. No
main effects or interactions involving virus image were obtained for either index,
p > .15 in all cases. Thus, the agency language manipulation influenced perceptions
of the fact sheet’s credibility, but the imagery manipulation did not affect the sheet’s
perceived credibility nor its informativeness.
Fear Arousal, Severity, and Susceptibility
We hypothesized that the language of virus agency would create greater concern
among readers about the health risk posed by H1N1 than would semantically
Downloaded by [University of California Davis] at 17:00 14 December 2012
Table 2. Principle components factor analysis (with varimax rotation) of respondents’ perceptions of the H1N1 virus and vaccine
Survey question
summaries
Factor 1
Source
credibility
Factor 2
Info quality
Factor 3
Self-efficacy
Factor 4
Fear
arousal
Factor 5
Severity
Factor 6
Susceptibility
Factor 7
Response
efficacy
Factor 8
Intent to get
vaccine
Factor
scale
M (SD) a
Coefficient
α
Fact sheet is
responsible
.77 .33 .01 .00 –.02 –.02 .07 .11
Fact sheet is sincere .81 .22 .14 .10 .07 .06 .10 .07
Fact sheet is ethical .86 .21 .07 .12 .06 –.05 .09 .09
Fact sheet is moral .88 .12 .13 .10 .06 –.06 .15 .08
Fact sheet is fair .86 .13 .12 .10 .10 –.01 .12 .10 2.74
(1.74)
.93
Fact sheet is
informative
.25 .82 .03 .10 –.02 –.03 .13 .17
Fact sheet is
believable
.37 .80 .09 .03 .07 –.01 .00 .11
Fact sheet is
professional
.32 .83 .09 –.04 .01 .09 .00 –.01 3.10
(1.83)
.87
Able to get vaccine .14 .11 .85 –.02 .08 .04 .13 .11
Easy to get vaccine .06 .15 .87 .01 .10 .11 .12 .06
Nothing to prevent
vaccination
.18 –.08 .75 .08 .09 –.03 .21 .13 5.02
(1.22)
.82
Fact sheet
frightened me
.11 .11 –.03 .84 .08 .04 .03 .04
Fact sheet made
me worry
.13 –.06 .06 .86 .07 –.01 .08 .05
Fact sheet made me
nervous, tense
.05 .02 .03 .88 .01 .03 .03 –.01 3.24
(1.04)
.84
(Continued)
9
Downloaded by [University of California Davis] at 17:00 14 December 2012
Survey question
summaries
Factor 1
Source
credibility
Factor 2
Info quality
Factor 3
Self-efficacy
Factor 4
Fear
arousal
Factor 5
Severity
Factor 6
Susceptibility
Factor 7
Response
efficacy
Factor 8
Intent to get
vaccine
Factor
scale
M (SD) a
Coefficient
α
Virus poses serious
risk
.07 .12 .10 .04 .88 .12 .04 .05
Virus infection is
harmful
.06 .05 .25 .06 .80 .23 .03 .03
Virus is a severe
threat
.09 –.12 –.05 .09 .79 .20 .21 .07 5.04
(1.05)
.82
At risk for infection –.01 –.07 .01 .03 .31 .83 –.01 –.01
Possible I will get
virus
–.12 .13 .19 .03 .15 .80 .07 .08
Susceptible to virus .07 –.01 –.05 .01 .08 .84 .12 –.11 5.35
(0.96)
.81
Vaccine will
prevent infection
.14 .03 .14 .10 .09 .02 .80 .16
Less likely to die if
vaccinated
.08 .02 .12 .01 .01 .11 .81 .11
Vaccine effective in
ending threat
.20 .10 .20 .02 .17 .05 .73 .13 4.80
(1.14)
.77
Will get vaccine
when available
.17 .12 .19 –.01 .06 .01 .24 .88
Chance of getting
vaccinated
.20 .12 .12 .08 .09 –.07 .19 .89 4.69
(1.63)
.89
Note. Boldface indicates that the item loaded high on a particular factor and was thus counted as part of that factor.
a Factor scales were computed by averaging item ratings. The first two scales had a 5-point range, whereas the remaining had a 7-point range.
Table 2. Continued
10
Downloaded by [University of California Davis] at 17:00 14 December 2012
Agency Assignment and H1N1 Transmission 11
equivalent human agency language. Consistent with our predictions, ratings on the
fear arousal index (measuring the degree of apprehension participants experienced
while reading the fact sheet) were higher for readers of the virus agent version
(M = 3.45, SD = 1.07) than for those who read the human agent version (M = 3.02,
SD = 0.97), F(1, 219) = 9.34, p = .003, η 2 = .041. Ratings on the severity index followed
a similar pattern, with higher ratings for the virus agent (M = 5.45, SD = 0.94) than
for the human agent version (M = 4.62, SD = 0.99), F(1, 218) = 41.51, p < .0001,
η 2 = .16. Moreover, agency assignment language also affected participants’ perceived
susceptibility to H1N1, F(1, 218) = 7.18, p = .008, η 2 = .032. However, the image used
to depict the virus moderated the influence of language on this index, as indicated
by an unanticipated Agency Assignment Language × Virus Image interaction, F(1,
218) = 10.00, p = .002, η 2 = .04. Pairwise comparisons revealed a significant difference
in susceptibility ratings for the virus agent (M = 5.64, SD = 0.98) and human agent
(M = 4.92, SD = 0.77) versions when the image was literal (Tukey’s honestly significant
difference = 3.65, p < .05) but not when it was agentic (M = 5.43, SD = 0.94; M = 5.37,
SD = 0.99; Tukey’s honestly significant difference = 1.07, p > .15). No other main
effects or interactions involving virus image were obtained for any of these indices,
p > .18 in all cases. In sum, participants felt more apprehensive about the virus,
perceived it as more severe, and perceived themselves as more susceptible to it when
the fact sheet was encoded in virus agency language than when it was encoded in
human agency language; however, the effect on susceptibility was observed only when
the image in the fact sheet was literal.
Self-Efficacy, Response Efficacy, and Intention to Get Vaccinated
Mean ratings on the response efficacy (M = 5.13, SD = 1.14) and self-efficacy
(M = 5.35, SD = 0.96) indices were relatively high (measured on a 7-point scale),
indicating that participants overall believed vaccination to be effective in preventing
H1N1 and that they would have little difficulty acquiring the vaccine once it became
available. Agency assignment language significantly affected ratings of response
efficacy, F(1, 218) = 10.17, p = .002, η 2 = .045, but not of self-efficacy, F(1, 218) = 2.46,
p = .118, η 2 = .01. Response efficacy ratings were higher among readers of the virus
agent version (M = 5.40, SD = 1.05) than among readers of the human agent version
(M = 4.96, SD = 1.17). Last, the language manipulation did reliably influence readers’
reported vaccination intention, F(1, 218) = 10.93, p = .001, η 2 = .048. As predicted,
the rated likelihood of getting vaccinated was higher among readers of the virus
agent version (M = 5.04, SD = 1.46) than among readers of the human agent version
(M = 4.32, SD = 1.72). No main effects or interactions involving virus image were
obtained for any of these indices, p > .14 in all cases. Overall, agency language did
appear to influence readers’ perceptions of the effectiveness of the H1N1 vaccine and
their intention to seek vaccination, but it had no appreciable effect on their perceived
ability to acquire it.
Request for Notification of Vaccine Availability
Overall, approximately 2 in 3 participants (65.8%) in our sample took up our
invitation to be notified when the H1N1 vaccine became locally available. Binary
logistic regression was used to determine the significant predictors of the invitation
acceptance rate. The predictor variables were agency assignment language, virus
Downloaded by [University of California Davis] at 17:00 14 December 2012
12 M. S. McGlone et al.
image, and the eight factors derived from the principal components analysis. A Hosmer
and Lemeshow (2000) goodness-of-fit test indicated that the model adequately fit the
data overall, χ 2 (8) = 10. 2 3, p = .249. The model was able to correctly classify 87.0%
of those who requested notification and 33.3% of those who did not, for an overall
success rate of 68.8%. Table 3 shows the unstandardized regression coefficient,
standard error, Wald χ 2 , odds ratio, and 95% confidence interval for each predictor.
Although neither of the manipulated variables reliably predicted notification request
rate, the susceptibility, response efficacy, and vaccination intention factors emerged
as significant partial predictors. Inverting the odds ratio for each predictor indicates
that a 1-point scale increase (all were measured on 7-point scales) was associated with
an increase in the likelihood of requesting notification of 65.79% for susceptibility,
156.25% for response efficacy, and 136.99% for vaccination intention. Thus,
participants were most interested in finding out about the vaccine’s availability when
they considered themselves at high risk for infection, they believed the vaccine would
reduce the threat of infection, and reported a strong intention to get the vaccine when
it became available.
Discussion
Medical discourse has often been criticized for its reliance on rhetorical devices that
obscure the human element in diagnosis and treatment (Broom & Tovey, 2007; Donnelly,
1986; Korsch, Gozzi, & Francis, 1968). For example, Anspach (1988) identified
several “depersonalizing” descriptive conventions in medical case histories, such as the
separation of biological processes from the people experiencing them (e.g., “The tibial
fracture was caused by a bicycle accident”), diagnostic statements in the passive voice
that omit human agents (e.g., “Cervical bleeding was observed”), and the assignment of
agency to technology (e.g., “Angiography revealed an arterio-venous malformation”).
These customs of clinical communication have been described as “language maladies”
that hinder the development of patient-centered medicine (Donnelly, 1997).
Although we acknowledge the potential of depersonalizing language to harm
patient care, our results suggest that downplaying the human element in certain health
Table 3. Binary logistic regression predicting likelihood of request for notification of
H1N1 vaccine availability
Predictor b SE Wald p Odds ratio 95% CI
Agency assignment
language
–0.09 0.35 0.07 .80 0.91 0.46–1.81
Virus imagery –0.19 0.32 0.37 .55 0.83 0.45–1.53
Source credibility –0.09 0.12 0.51 .48 0.92 0.72–1.17
Information quality –0.03 0.11 0.12 .75 0.97 0.79–1.19
Self-efficacy 0.14 0.14 0.93 .34 1.15 0.87–1.52
Fear arousal 0.14 0.16 0.81 .37 1.13 0.84–1.57
Severity –0.21 0.18 1.33 .25 0.81 0.57–1.16
Susceptibility 0.42 0.20 4.57 .03 1.52 1.04–2.23
Response efficacy –0.44 0.17 6.45 .01 0.64 0.45–0.94
Vaccination
intention
–2.25 0.80 7.88 .01 0.73 0.58–0.91
Downloaded by [University of California Davis] at 17:00 14 December 2012
Agency Assignment and H1N1 Transmission 13
messages may actually be beneficial in prevention-oriented messages. 1 Specifically, we
found that linguistically assigning the agency for H1N1 transmission to the virus in
an informational sheet led readers to perceive the sheet’s content as more impartial,
the virus as more severe, and themselves as more susceptible to it than assigning
transmission agency to humans. Readers of the virus agent version also perceived the
vaccine as more effective and reported stronger intentions to get vaccinated than did
readers of the human agent version. Manipulation of the image (literal or agentic)
accompanying the textual information did not affect perceptions of severity or vaccine
efficacy, although it did unexpectedly moderate the influence of agency language on
readers’ concerns about their susceptibility to infection. We consider these effects on
readers’ beliefs about the virus and their behavioral intentions regarding vaccination
in turn.
The manipulation of agency assignment did not affect perceptions of fact sheet
content as “informative” or “believable,” but it did influence the extent to which it
was viewed as “fair,” “sincere,” “responsible,” and “ethical,” a set of descriptors we
interpreted as collectively indicating perceptions of the sheet’s credibility. Although
the observed advantage of the virus agent version in credibility was not hypothesized,
it is consistent with analyses of the linguistic cues that signal “discourse stance”—i.e.,
a communicator’s orientation (e.g., emotional, experiential, political) toward a topic
raised in ongoing discourse (Berman, Ragnarsdottir, & Stromqvist, 2002; van Hell,
Verhoeven, Tak, & van Oosterhout, 2005). Sentence constructions that de-emphasize
human agency, such as agentless passive voice (e.g., “The manuscript was reviewed
in three weeks”) and assignments to nonliving entities (e.g., “This essay criticizes
American foreign policy”) generally serve a rhetorical distancing function in
exposition and, in the specific context of medical discourse (e.g., “Rectal temperature
was measured”), are commonly used to convey an orientation of clinical objectivity
(Atkinson, 1999; Berman, 2005; Lamb, 1991). Thus, some of the cases in which
agency was assigned to a nonliving entity in the virus agent version may have been
interpreted by readers as vestiges of a clinical perspective on H1N1 transmission and
prevention.
According to the Extended Parallel Process Model, audiences evaluate messages
about a health threat by separately appraising the perceived threat (comprising
severity and one’s personal susceptibility to the threat) and their efficacy in protecting
themselves from it. These judgments in turn determine whether audience members
will take measures to control the danger associated with the threat (optimally) or
instead control only their fear of the threat (problematically). As we predicted, agency
assignment reliably affected readers’ perceptions of the threat the H1N1 virus posed.
Assigning transmission agency to the virus induced greater apprehension during
perusal of the fact sheet, higher perceived severity, and elevated concerns about
H1N1 susceptibility relative to human agency, although the last effect was observed
only when the image accompanying the text was literal. These effects derive from a
conception of the virus as an autonomous aggressor shaped by the language of virus
agency. In contrast, encoding H1N1 transmission in terms of human agency cast the
1 It is notable that criticisms of “depersonalizing” language in medical discourse have
focused on post-treatment description messages (e.g., case histories), rather than the prevention-
oriented messages at issue in our research. Whether message timing influences the perception of
depersonalization is an important issue for future research. We thank an anonymous reviewer
for pointing it out to us.
Downloaded by [University of California Davis] at 17:00 14 December 2012
14 M. S. McGlone et al.
illness as a consequence of careless but controllable human behavior, and thereby
made it appear less threatening.
It is important to note that the heightened threat produced by virus agency did
not exceed the critical point at which perceived severity and susceptibility outweigh
efficacy, causing people to shift from danger control to fear control processes (Witte
& Allen, 2000). Self-efficacy judgments were not affected by agency assignment and,
despite aggravating perceived severity and susceptibility, virus agency enhanced
perceived response efficacy rather than reducing it. Virus agency also boosted
reported intentions to seek vaccination relative to human agency, although it did not
directly increase the frequency with which participants requested notification about
when the vaccination would become available on campus. Note, however, that the
response rate was high in general (65.8% across experimental conditions) and near
ceiling among participants who felt highly susceptible to H1N1, firmly believed in
the vaccine’s response efficacy, and reported strong intentions to get vaccinated, all
factors influenced by agency assignment language.
The manipulation of agency imagery (literal vs. agentic) had no effect on
perceptions of severity, self-efficacy, response efficacy, or reported intentions to
seek vaccination. Manipulation check findings indicate that the vast majority of
participants were able to remember the image that appeared on their sheets, so these
null effects cannot be attributed to participants’ failure to attend to the images; agency
imagery simply did not augment the effects of agency assignment language as we had
predicted. However, we did not subject these images to pilot-testing to determine their
evocative properties a priori. Whether the hypothesized effects might be obtained with
different visual depictions of agency is an issue for future research.
Future research is also needed to address other limitations of the present study.
First, there are several questions about the external validity of our study that merit
further exploration. We examined the effect of agency assignments on perceptions of
the H1N1 virus in a convenience sample of predominantly female college students
reading printed educational materials. The generalizability of the observed effects to
other health threats (e.g., herpes, obesity), other populations (e.g., children, adults
with lower vaccination rates 2 ), and other media (e.g., television, Internet) are issues
we will address in subsequent research. Second, we examined agency assignment’s
influence on reported intentions to seek vaccination, but not its influence on the
actual rate of vaccination on respondents. The direct influence of this manipulation
on vaccination-seeking behavior merits future research. Third, the language
manipulation was not homogeneous, but rather was comprised of different cue
words and expressions conveying virus or human agency. Thus, it is not clear that
the agency assignment effects were attributable to these cues in the aggregate or to
specific cues that had significant impact in contrast with others with little or none
(Clark, 1973; McGlone, Bortfeld, & Kobrynowicz, 2007). A fourth, related concern
is a possible confound between agency language and intensity, such that virus agent
cues (e.g., infect) were more threatening than were human agent cues (e.g., contract)
by virtue of being more intense. We acknowledge this possibility, but also note that
2 The reported level of influenza vaccination within the past 12 months in our sample
(66.4%) was higher than the national average for people in their age range (40.4% among per-
sons over 18 according to the Centers for Disease Control and Prevention; see http://www.cdc.
gov/flu/professionals/ vaccination/coverage_0910estimates. htm). In this respect, our sample is
not completely representative of the U.S. population most at risk for acquiring H1N1.
Downloaded by [University of California Davis] at 17:00 14 December 2012
Agency Assignment and H1N1 Transmission 15
the discourse of disease transmission places humans in a defensive role semantically
as well as medically, and the language of defensive action tends to be less intense
than that for offensive action (McEwen & Greenberg, 1970). Nevertheless, potential
differences in the intensity of language cues qualify our conclusions about the
influence of agency assignment.
Last, our fact sheet materials can be faulted for a lack of specificity that may
have distorted participants’ understanding of how to prevent or treat flu symptoms.
For example, all versions of the fact sheet informed participants that in the event
they contracted the flu, a doctor “can prescribe antiviral drugs that effectively treat
influenza.” This statement is accurate only within a limited time frame—specifically,
an antiviral treatment regimen is most likely to be effective if begun within 48 hours
of infection. Our omission of this qualification consequently may have induced undue
optimism among participants about medical intervention, which in turn may have
dampened the effect of our agency manipulation on perceptions of response efficacy.
PEMs and other health educational materials must be both informative and
persuasive, especially when addressing evidence-based recommendations and
supporting prescribed treatment plans. Our findings suggest that designers can
improve the persuasive effect of these materials by making deliberate decisions
about the language used to encode viral transmission and treatment facts. Agency
assignment in language commonly occurs outside of conscious awareness and news
reports, pamphlets, and Internet briefs about influenza viruses typically include a blend
of statements assigning transmission agency to humans and to the virus. However,
conscious and consistent assignment of agency to the virus provides a straightforward
means for enhancing PEM impact by elevating perceptions of severity and response
efficacy.
References
Ache, K. A., & Wallace, L. S. (2009). Are end-of-life patient education materials readable?
Palliative Medicine Journal, 23, 545–548.
Anspach, R. R. (1988). Notes on the sociology of medical discourse: The language of case
presentation. Journal of Health and Social Behavior, 29, 357–375.
Atkinson, D. (1999). Language and science. Annual Review of Applied Linguistics, 19, 193–214.
Berman, R. A. (2005). Developing discourse stance in different text types and languages. Journal
of Pragmatics, 37, 105–124.
Berman, R. A., Ragnarsdottir, H., & Stromqvist, S. (2002). Discourse stance. Written Language
and Literacy, 5, 255–290.
Broom, A., & Tovey, T. (2007). The dialectical tension between individuation and
depersonalization in cancer patients’ mediation of complementary, alternative, and
biomedical cancer treatments. Sociology, 41, 1021–1039.
Bull, F. C., Holt, C. L., Kreuter, M. W., Clark, E. M, & Scharff, D. (2001). Understanding the
effects of printed health education materials: Which features lead to which outcomes?
Journal of Health Communication, 6, 265–279.
Cameron, L. D., & Chan, C. K. Y. (2008). Designing health communications: Harnessing the
power of affect, imagery, and self-regulation processes. Social and Personality Psychology
Compass, 1, 262–282.
Centers for Disease Control & Prevention. (2009). 2009 H1N1 early outbreak and disease
characteristics. Retrieved from http://www.cdc.gov/H1N1FLU/surveillanceqa.htm
Charnock, D., Shepperd, S., Needham, G., & Gann, R. (1999). DISCERN: An instrument for
judging the quality of written consumer health information on treatment choices. Journal
of Epidemiology and Community Health, 53, 105–111.
Downloaded by [University of California Davis] at 17:00 14 December 2012
16 M. S. McGlone et al.
Child, A. (2008). Check text readability. Retrieved from http://www.addedbytes.com/code/
readability-score
Clark, H. H. (1973). The language-as-fixed effect fallacy: A critique of language statistics in
psychological research. Journal of Verbal Learning and Verbal Behavior, 12, 335–359.
Cohen, A., Colligan, M. J., & Berger, P. (1985). Psychology in health risk messages for workers.
Journal of Occupational Medicine, 27, 543–551.
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16,
297–334.
Donnelly, W. J. (1986). Medical language as symptom: Doctor talk in teaching hospitals.
Perspectives in Biology and Medicine, 30, 81–94.
Donnelly, W. J. (1997). The language of medical case histories. Annals of Internal Medicine, 127,
1042–1048.
Dowty, D. (1991). Thematic proto-roles and argument selection. Language, 67, 547–619.
Guidry, J. J., & Walker, V. D. (1999). Assessing cultural sensitivity in printed cancer materials.
Cancer Practice Journal, 7, 291–296.
Harris, M., Smith, B., & Veale, A. (2005). Printed patient education interventions to facilitate
shared management of chronic disease: A literature review. Internal Medicine Journal, 35,
711–716.
Herschorn, S., Becker, D., Miller, E., Thompson, M., & Forte, L. (2004). Impact of a health
education intervention in overactive bladder patients. Canadian Journal of Urology, 11,
2430–2437.
Hill-Briggs, F., & Smith, A. S. (2008). Evaluation of diabetes and cardiovascular disease print
patient education materials for use with low-health literate populations. Diabetes Care
Journal, 31, 667–671.
Hoffmann, T., & Worrall, L. (2004). Designing effective written health education materials:
considerations for health professionals. Disability & Rehabilitation, 26, 1166–1173.
Holmes-Rovner, M. (2007). International Patient Decision Aid Standards (IPDAS): Beyond
decision aids to usual design of patient education materials. Health Expectations, 10,
103–107.
Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression. New York, NY: Wiley.
Jackendoff, R. (1990). Semantic structures. Cambridge, MA: MIT Press.
Kools, M. (2007). A focus on the usability of health education materials. Patient Education and
Counseling, 65, 275–276.
Korsch, B. M., Gozzi, E., & Francis, V. (1968). Gaps in doctor–patient communication: Doctor-
patient interaction and patient satisfaction. Pediatrics, 42, 855–871.
Kosslyn, S. M., & Ganis, G., & Thompson, W. L. (2001). Neural foundations of imagery. Nature
Reviews Neuroscience, 2, 635–642.
Lamb, S. (1991). Acts without agents: An analysis of linguistic avoidance in journal articles on
men who batter women. American Journal of Orthopsychiatry, 61, 250–257.
Leiner, M., Handal, G., & Williams, D. (2004). Patient communication: A multidisciplinary
approach using animated cartoons. Health Education Resources, 19, 591–595.
McEwen, W. J., & Greenberg, B. S. (1970). The effects of message intensity on receiver
evaluations of source, message, and topic. Journal of Communication, 20, 340–350.
McGlone, M. S., Bortfeld, H., & Kobrynowicz, D. (2007). Laying it on thin: Analogical cue
frequency in the manipulation of choice. Personality and Social Psychology Bulletin, 33,
721–731.
McGlone, M. S., & Pfiester, R. A. (2009). Does time fly when you’re having fun or do you?
Affect, agency, and embodiment in temporal communication. Journal of Language and
Social Psychology, 28, 3–31.
Pinker, S., & Jackendoff, R. (2005). The faculty of language: What’s so special about it?
Cognition, 95, 201–236.
Ruiter, R. A. C., Abraham, C., & Kok, G. (2001). Scary warnings and rational precautions: A
review of the psychology of fear appeals. Psychology and Health, 16, 613–630.
Downloaded by [University of California Davis] at 17:00 14 December 2012
Agency Assignment and H1N1 Transmission 17
Schwartzberg, J. G., Cowett, A., VanGeest, J., & Wolf, M. S. (2007). Communication techniques
for patients with low health literacy: A survey of physicians, nurses, and pharmacists.
American Journal of Health Behavior, 31, S96–S104.
Selander, S., Troein, M., Finnegan, J., Jr., & Rastam, L. (1997). The discursive formation
of health. A study of printed health education material used in primary care. Patient
Education and Counseling, 31, 181–189.
Seligman, H. K., Wallace, A. S., DeWalt, D. A., Schillinger, D., Arnold, C. L., Shilliday, B. B.,
… Davis, T. C. (2007). Facilitating behavior change with low-literacy patient education
materials. American Journal of Health Behavior, 31, S69–S78.
Strecher, V. J., Marcus, A., Bishop, K., Fleisher, L., Stengle, W., Levinson, A., … Nowak, M.
(2005). A randomized controlled trial of multiple tailored messages for smoking cessation
among callers to the cancer information service. Journal of Health Communication, 10,
105–118.
Taylor, J. D. (2008). The impact of electronic mail versus print delivery of an exercise program
on muscular strength and aerobic capacity in people with type 2 diabetes. Journal of
Strength and Conditioning Research, 22, 1696–1704.
van Hell, J. G., Verhoeven, L., Tak, M., & van Oosterhout, M. (2005). To take a stance: A
developmental study on the use of pronouns and passives in spoken and written texts in
different genres. Journal of Pragmatics, 37, 238–272.
Wallace, L. S., Keenum, A. J., Roskos, S. E., Blake, G. H., Colwell, S. T., & Weiss, B. D.
(2008). Suitability and readability of consumer medical information accompanying
prescription medication samples. Patient Education and Counseling, 70, 420–425.
Wallace, L. S., Rogers, E. S., Turner, L. W., Keenum, A. J., & Weiss, B. D. (2006). Suitability of
written supplemental materials available on the Internet for nonprescription medications.
American Journal of Health System Pharmacy, 63, 71–78.
Weih, M., Reinhold, A., Richter-Schmidinger, T., Sulimma, A. K., Klein, H., & Kornhuber, J.
(2008). Unsuitable readability levels of patient information pertaining to dementia and
related diseases: a comparative analysis. International Psychogeriatrics, 20, 1116–1123.
Whittingham, J. R., Ruiter, R. A., Castermans, D., Huiberts, A., & Kok, G. (2008). Designing
effective health education materials: Experimental pre-testing of a theory-based brochure
to increase knowledge. Health Education Resources, 23, 414–426.
Witte, K. (1994). Fear control and danger control: A test of the Extended Parallel Process
Model. Communication Monographs, 61, 113–134.
Witte, K., & Allen, M. (2000). A meta-analysis of fear appeals: Implications for effective public
health campaigns. Health Education and Behavior, 27, 591–615.
Witte, K., Meyer, G., & Martell, D. (2001). Effective health risk messages. Newbury Park, CA:
Sage.
Wright, P. (2003). Criteria and ingredients for successful patient information. Journal of
Audiovisual Media in Medicine, 26, 6–10.
