Framing Messages for Vaccination Supporters

Abstract

Efficiently communicating information on vaccination is crucial to maintaining a high level of immunization coverage, but it implies finding the right content for the right audience. Provaccination individuals, who represent the majority of the population, and who have been neglected in the literature, could play an important role relaying provaccination messages through informal discussions, if only these messages are (a) found plausible, (b) remembered, and (c) shared. We conducted 7 experiments on 2,761 provaccination online participants (United States and United Kingdom), testing whether the valence of a statement (positive or negative) and its rhetorical orientation (pro- or antivaccine) affected these 3 steps. Participants deemed more plausible, were more willing to transmit (and actually transmitted more), but did not remember positively framed statements better. Provaccination rhetorical orientation had little or no effect. Overall, the framing effects observed were dramatic: one framing made participants very eager to transmit a statement, while another made them reluctant to transmit it at all. The framing effects also influenced vaccination attitudes, with participants exposed to positively framed statements reporting more positive attitudes toward vaccination. Since messages have to be framed one way or the other, the framing effects demonstrated here should be considered when designing public health messages. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Full text

Framing Messages for Vaccination Supporters
Sacha Altay and Hugo Mercier
Jean Nicod Institute, Paris, France
Efficiently communicating information on vaccination is crucial to maintaining a high level of immu-
nization coverage, but it implies finding the right content for the right audience. Provaccination
individuals, who represent the majority of the population, and who have been neglected in the literature,
could play an important role relaying provaccination messages through informal discussions, if only these
messages are (a) found plausible, (b) remembered, and (c) shared. We conducted 7 experiments on 2,761
provaccination online participants (United States and United Kingdom), testing whether the valence of
a statement (positive or negative) and its rhetorical orientation (pro- or antivaccine) affected these 3 steps.
Participants deemed more plausible, were more willing to transmit (and actually transmitted more), but
did not remember positively framed statements better. Provaccination rhetorical orientation had little or
no effect. Overall, the framing effects observed were dramatic: one framing made participants very eager
to transmit a statement, while another made them reluctant to transmit it at all. The framing effects also
influenced vaccination attitudes, with participants exposed to positively framed statements reporting
more positive attitudes toward vaccination. Since messages have to be framed one way or the other, the
framing effects demonstrated here should be considered when designing public health messages.
Public Significance Statement
Subtle changes in the way information about vaccination is presented can affect how people perceive
it. Participants found more plausible, and were much more likely to share, positively framed
statements about vaccination than negatively framed statements. These framing effects should be
taken into account when designing public health messages.
Keywords: vaccination, communication, framing, transmission, plausibility
Despite the clear scientific consensus on the safety and benefits
of vaccines, a significant part of the population remains vaccine
hesitant (i.e., holding doubts and concerns about the benefits and
safety of vaccination; Larson, Jarrett, Eckersberger, Smith, &
Paterson, 2014; MacDonald & the SAGE Working Group on
Vaccine Hesitancy, 2015; Yaqub, Castle-Clarke, Sevdalis, & Cha-
taway, 2014). Most efforts dedicated to increasing vaccine accep-
tance have aimed at vaccine-hesitant individuals, examining
whether provaccination messages attenuate their misgivings (e.g.,
Chanel, Luchini, Massoni, & Vergnaud, 2011; Horne, Powell,
Hummel, & Holyoak, 2015). Typically, these messages are meant
to be part of health campaigns, or to be delivered by medical
professionals. However, information about vaccination also
spreads through other means, with significant portions of the
public relying, in part, on informal sources such as friends, col-
leagues, and family (Freed, Clark, Butchart, Singer, & Davis,
2011). In the area of public opinion more generally, the role of
peers to relay messages from the media or other official sources is
well known (e.g., Katz & Lazarsfeld, 1955). In science commu-
nication, a recent study has shown that the more people know
about the scientific consensus on global warming, the more likely
they are to discuss it with their peers, which leads to a positive
“pro-climate social feedback loop” (Goldberg, van der Linden,
Maibach, & Leiserowitz, 2019, p. 14804). In the case of vaccina-
tion, peers might play an equally important role of information
transfer—for instance, relatives’ opinions on vaccines are a strong
predictor of vaccination attitudes (Brunson, 2013; Opel & Mar-
cuse, 2013) and interpersonal communication is commonly used to
address health problems (Knapp & Daly, 2002).
In most countries, the majority of the population supports vac-
cination at least to the extent that they and their children get
vaccinated (Kennedy, Basket, & Sheedy, 2011). We could there-
fore expect this majority to effectively help spread provaccination
messages, a strategy that has been successfully implemented by
Schoeppe et al. (2017) in small communities. Unfortunately, most
XSacha Altay and XHugo Mercier, Department of Cognitive Studies,
Jean Nicod Institute, Paris, France.
Financial support for this study was provided entirely by the grant
ANR-17-EURE-0017 FrontCog and ANR-10-IDEX-0001-02 PSL. The
funding agreement ensured the authors’ independence in designing the
study, interpreting the data, writing, and publishing the report. Sacha
Altay’s PhD thesis is funded by the Direction Générale de l’Armement
(DGA). We are grateful to Aurélien Allard for statistical advice. We would
also like to thank Camille Williams and Brent Strickland for valuable
feedback and suggestions.
Correspondence concerning this article should be addressed to Sacha
Altay, Departement d’étude cognitive, Institut Jean-Nicod, 29 rue d’Ulm,
75005 Paris, France. E-mail: sacha.altay@gmail.com
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Experimental Psychology: Applied
© 2020 American Psychological Association 2020, Vol. 2, No. 999, 000
ISSN: 1076-898X http://dx.doi.org/10.1037/xap0000271
1

provaccination members of the public are only willing or able to
offer relatively weak arguments to defend their opinions, relying
on personal anecdotes rather than proper evidence (Faasse, Chat-
man, & Martin, 2016; Fadda, Allam, & Schulz, 2015). Provacci-
nation individuals are also less likely to engage in online commu-
nication (McKeever, McKeever, Holton, & Li, 2016), leading to an
overrepresentation of antivaccination comments online (Feinberg
et al., 2015; Lei et al., 2015; Pereira et al., 2013). This could reflect
the fact that most vaccinees passively accept vaccination, and often
lack information about its scientific backing and benefits (Das &
Das, 2003; Nichter, 1995, p. 617; Yaqub et al., 2014; for a more
detailed classification of attitudes toward vaccination, see Stree-
fland, Chowdhury, & Ramos-Jimenez, 1999). Moreover, studies
have shown that even vaccinees often have concerns regarding the
safety of vaccination (e.g., Casiday, Cresswell, Wilson, & Panter-
Brick, 2006).
Many efforts at convincing staunchly antivaccine individuals
have failed (Dubé, Gagnon, MacDonald, N. E., & the SAGE
Working Group on Vaccine Hesitancy, 2015; Sadaf, Richards,
Glanz, Salmon, & Omer, 2013) or have even backfired (e.g.,
Nyhan, Reifler, Richey, & Freed, 2014). These failures may reflect
the lack of trust antivaccine individuals place in the medical
establishment (Salmon et al., 2005; Yaqub et al., 2014). As a
result, members of the public may be in a good position to
convince vaccine-hesitant individuals (at least fence-sitters; Leask,
2011), if only they could muster convincing arguments. Thus,
improving the argumentative arsenal of provaccination individuals
could indirectly help convince vaccine hesitant individuals—and it
might further assuage the remaining concerns of provaccine indi-
viduals.
So far, the effectiveness of provaccination arguments has been
measured in terms of persuasiveness. However, if arguments are to
be further spread by provaccine individuals, these individuals
should not only accept the arguments, but they should also be able
to memorize them and be willing to transmit them to others.
Although the field of cultural evolution has long paid attention to
memory and willingness to transmit in theoretical (Boyer, 2018;
Morin, 2015) and empirical work (for a recent review, see Miton
& Charbonneau, 2018; for applications to vaccination, see Jimé-
nez, Mesoudi, & Tehrani, 2020; Jiménez, Stubbersfield, & Teh-
rani, 2018; Miton & Mercier, 2015), only one of these factors
(memorization) has been examined in the case of vaccination
(Pluviano, Watt, & Della Sala, 2017).
Studies on information transmission suggest that seemingly
minor changes in the presentation of a piece of information can
influence its likelihood of being accepted, remembered, and fur-
ther transmitted. Framing is the best studied of these superficial
changes: logically equivalent descriptions can lead to very differ-
ent decisions when they are framed differently (Kahneman &
Tversky, 1979; Tversky & Kahneman, 1981). One of the most
robust framing effects is valence framing, where negatively framed
information is judged as truer (Fessler, Pisor, & Navarrete, 2014),
is better memorized (Bebbington, MacLeod, Ellison, & Fay,
2017), and is transmitted more often (Heath, 1996) than positively
framed information (a glass half empty is a negative frame while
a glass half full would be a positive frame). However, there are
relevant exceptions to this dominance of negative frames. When
people are asked whether to approve medical treatments, empha-
sizing survival rates, by contrast with mortality rates, leads to more
positive attitudes toward the treatment (Levin, Schnittjer, & Thee,
1988; Marteau, 1989; Wilson, Kaplan, & Schneiderman, 1987).
This preference for positive frames is also found in cases such as
rating basketball players (Levin, 1987), judging the quality of
ground beef (Levin & Gaeth, 1988), support for condom use
(Linville, Fischer, & Fischhoff, 1993), or evaluating industry proj-
ect teams (Dunegan, 1993, 1995). Anecdotally, a survey on vac-
cine communication found that positive messages were rated as
more convincing and more believable than negative messages
(Association of State and Territorial Health Officials, 2009), but
the conclusions we can draw from this study are limited since the
information available in the two frames was not equivalent.
Framing is unavoidable: a message has to be framed one way or
the other. This is relevant for common provaccination messages:
those regarding side effects and those regarding medical consen-
sus. Since overestimating the dangerousness of vaccine side effects
appears to be a major reason for vaccine refusal (Betsch & Wicker,
2012; Omer, Salmon, Orenstein, deHart, & Halsey, 2009), provac-
cination campaigns often stress the safety of vaccines (Dittmann,
2001; van der Linden, Clarke, & Maibach, 2015; World Health
Organization, 2016). Such messages can be framed positively (e.g.,
999 out of 1,000 people do not have any severe side effects after being
injected with a vaccine) or negatively (e.g., 1 individual out of 1,000
has severe side effects after being injected with a vaccine). For
example, the World Health Organization opted for a negative frame
when tweeting “Severe side effects are extremely rare” (https://twitter
.com/WHO/status/1192071297230147584?s!20). Since knowledge
that the medical consensus on the efficacy and safety of vaccination
is positively correlated with favorable beliefs on vaccination (Dixon
&Clarke,2013),messagesaboutmedicalconsensushavealsobeen
used. These messages (e.g., 90% of medical scientists agree that
vaccines are safe; van der Linden et al., 2015) have managed to
reduce belief in the autism-vaccine link and to increase public support
for vaccination. Here as well, the message has to be framed either
positively (90% of medical scientists agree that vaccines are safe) or
negatively (10% of medical scientists do not agree that vaccines are
safe). This type of simple framing has been classified as attribute
framing, and it is considered “the most straightforward test of the
influence of positive and negative framing” (for a typology of framing
effects, see Levin, Schneider, & Gaeth, 1998, p. 158).
Even though frames are logically equivalent, it has been sug-
gested that they leak information, such as information about the
speaker’s beliefs (McKenzie, 2004; McKenzie & Nelson, 2003;
Sher & McKenzie, 2006). Specifically, framing valence influences
perceived rhetorical orientation—that is, what conclusion the in-
formation appears to support. A speaker who uses the positive
frame might be perceived as being provaccine, while one who uses
the negative frame might be perceived as being antivaccine. As a
result, people who want to be perceived as either pro- or antivac-
cine may show a preference for either one of the frames, by being
more eager to transmit messages that appear to reflect their posi-
tion (Chung & Darke, 2006; Hennig-Thurau, Gwinner, Walsh, &
Gremler, 2004; Sundaram, Mitra, & Webster, 1998). Rhetorical
orientation could even affect the perceived plausibility of the
messages. Studies in political science have shown that survey
respondents often provide answers on factual questions as a way of
expressing their political opinions (Bullock, Gerber, Hill, & Hu-
ber, 2015). Similarly, the plausibility ratings of statements related
to vaccination could be used by participants to communicate their
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2ALTAY AND MERCIER

attitudes toward vaccination, with statements having a provaccine
rhetorical orientation being deemed more plausible by provaccine
individuals, and vice versa.
The goal of the study was to test whether framing influences the
steps necessary for a statement to be used by a provaccine indi-
vidual in informal discussions: (a) evaluation of message plausi-
bility (Experiments 2 and 5), (b) memorization of the message
(Experiment 3), (c) willingness to transmit the message further
(Experiments 4 and 6), and finally (d) the propensity to use the
statements to convince someone who refuses to vaccinate (Exper-
iment 7). Experiment 1 aimed at establishing that our predictions
about the direction of rhetorical orientation of the experimental
statements (in favor or against vaccination) were accurate. Exper-
iments 5 and 6 were replications of Experiments 2 and 4 using a
different sample. In each experiment, we compared statements that
varied in valence (positively vs. negatively framed) and in rhetor-
ical orientation (whether they are associated with a pro- or an
antivaccine position). However, statements with a rhetorical ori-
entation in favor of vaccination also had a positive valence (“90%
of medical scientists think that vaccines are safe” and “999 out of
1,000 don’t have any severe side effects after being injected with
a vaccine”). To disentangle the effects of valence framing and
rhetorical orientation, we added a third variant, that was negatively
framed but whose rhetorical orientation was likely to be perceived
as provaccine because it began with the word only (e.g., “Only 1
individual out of 1,000 has severe side effects after being injected
with a vaccine”). We hypothesized that (a) only would convey an
evaluative judgment indicating that the given frequency is consid-
ered to be low, and (b) that this evaluative judgment would be
more likely attributed to a vaccination supporter than to someone
holding negative views on vaccination.
The literature on the pragmatics of framing effects (McKenzie,
2004; McKenzie & Nelson, 2003; Sher & McKenzie, 2006) leads
us to expect that the choice of frame made by a person will
influence what ideological position others attribute to them:
Hypothesis 1: Positively framed statements with a rhetorical
orientation in favor of vaccination will be more strongly
associated with provaccination individuals than statements
with a negative valence and a rhetorical orientation against
vaccination.
Hypothesis 2: Negatively framed statements with a rhetorical
orientation in favor of vaccination (i.e., the statements with
only) will be more strongly associated with provaccination
individuals than statements with a negative valence and a
rhetorical orientation against vaccination.
The literature on the negativity bias (e.g., Baumeister, Brat-
slavsky, Finkenauer, & Vohs, 2001) leads us to expect that:
Hypothesis 3: Negatively framed statements will be judged
truer than positively framed statements.
Hypothesis 4: Negatively framed statements will be better
memorized than positively framed statements.
Hypothesis 5: Negatively framed statements will be more
transmitted than positively framed statements.
The literature on attribute framing (e.g., Levin et al., 1998) leads
us to expect that:
Hypothesis 6: Positively framed statements will be judged
truer than negatively framed statements.
Hypothesis 7: Exposition to positively framed statements will
create more positive attitudes toward vaccination than expo-
sition to negatively framed statements.
The literature in political science (e.g., Bullock et al., 2015)
leads us to expect the rhetorical orientation to have the following
effects:
Hypothesis 8: Statements with a rhetorical orientation in favor
of vaccination will be judged truer than statements with a
rhetorical orientation against vaccination.
Hypothesis 9: Statements with a rhetorical orientation in favor
of vaccination will be more shared than statements with a
rhetorical orientation against vaccination.
Some of these predictions are conflicting (e.g., H
3
and H
6
),
others are unique (e.g., H
7
). The present investigation aims at
putting these theories to the test in the case of information about
vaccines.
Experiment 1
In this first experiment we measured the perceived ideological
position of individuals sharing the statements. We predicted that
positively framed statements (with a rhetorical orientation in favor
of vaccination) would be attributed to provaccination individuals
(H
1
), and that the addition of the word only at the beginning of
negatively framed statements would modify the statements’ per-
ceived rhetorical orientation from antivaccination to provaccina-
tion (H
2
).
Participants
For Experiments 1 to 4, participants residing in the United States
were recruited through Amazon Mechanical Turk (MTurk). Sam-
ples recruited on MTurk have proven suitable for similar experi-
ments (Paolacci & Chandler, 2014). A power analysis for a chi-
square test with an estimated effect size of w!0.5 (we expected
a large effect size since information leakage in attribute framing
can have drastic effects; Sher & McKenzie, 2008), an "level of
5%; and a power of 80%, suggested that we needed 52 participants
per condition. Since we have six conditions, we needed a mini-
mum of 312 participants. Because we did not know how many
participants we would need to exclude and considering that we
could have overestimated the expected effect size, we recruited
480 participants, who were paid $0.30. Eleven participants failed
to complete the survey, and we removed 50 participants catego-
rized as being against vaccination, leaving 419 participants (184
women, M
age
!35.03, SD !11.05; Public Support Index (from
1!negative attitudes to 7 !positive attitudes): M!6.25, SD !
0.83). Approval from the ethics committee was obtained for all
experiments (CER-Paris Descartes; No. 2019-03-MERCIER; all
data presented in this article are available at https://osf.io/ksa5b).
Design and Procedure
The overall design and procedures were similar for all experi-
ments. Participants signed a consent form, took part in the exper-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
FRAMING MESSAGES FOR VACCINATION SUPPORTERS

iment itself, then filled in a Public Support Index questionnaire
developed by van der Linden et al. (2015) to assess respondents’
position on vaccination. The index is composed of seven state-
ments supporting vaccination that participants have to rate on a
scale from 1 (Strongly disagree)to7(Strongly agree). Participants
who mostly agreed with these statements (i.e., average score
higher than 4) were considered to be in favor of vaccination, others
against vaccination. Finally, in a debriefing, participants were
provided with accurate information about vaccination, and then
filled in demographic questions.
In Experiment 1, for the experimental task participants were
provided with one of the six statements from Table 1, and asked to
answer the question “Would this person [the person who uttered
the statement] more likely to be: in favor of vaccination or against
vaccination?”
Results and Discussion
Table 1 provides the percentage of participants who thought the
individual uttering each statement was more likely to be pro- or
antivaccination.
As predicted by H
1
, positively framed statements with a rhetor-
ical orientation in favor of vaccination were more likely to be
attributed to a provaccination individual than negatively framed
statements with a rhetorical orientation against vaccination (#
2
(1,
N!285) !91.13, p$.001, %!1.37).
As predicted by H
2
, negatively framed statements with a rhe-
torical orientation in favor of vaccination were more likely to be
attributed to a provaccination individual than negatively framed
statements with a rhetorical orientation against vaccination #
2
(1,
N!276) !98.84, p$.001, %!1.49.
Statements perceived as having a provaccination rhetorical ori-
entation were more likely to be attributed to a provaccination
individual than statements perceived as having an antivaccination
rhetorical orientation #
2
(1, N!419) !119.84, p$.001, %!
1.23.
Experiment 2
In the second experiment we measured how valence framing
and rhetorical orientation influence the perceived plausibility of
the statements. The literature on the negativity bias predicts that
negatively framed statements will be judged truer than positively
framed statements (H
3
), while the literature on attribute framing
leads us to expect the opposite (H
6
). The literature on political
science leads us to expect that statements with a rhetorical orien-
tation in favor of vaccination will be judged truer than statements
with a rhetorical orientation against vaccination (H
8
).
Participants
A power analysis for an analysis of variance (ANOVA) with an
estimated effect size of f!0.325 (the literature led us to expect
an effect size between medium and large, see, Fessler et al., 2014), an
"-level of 5%, and a power of 80% suggested that we needed a
minimum of 77 participants per condition. Since we have six condi-
tions, we needed a minimum of 462 participants. Considering that we
would need to exclude participants, we recruited 480 participants who
Table 1
The Messages Used in All the Experiments, Along With Whether They Are Framed Positively or Negatively, Their Presumed
Rhetorical Orientation, and the Results of All the Experiments (Experiment 1 [U.S.]: Perception as Provaccine; Experiments 2 [U.S.]
and 5 [UK]: Plausibility; Experiment 3 [U.S.]: Recall Rate; Experiments 4 [U.S.] and 6 [UK]: Willingness to Transmit)
Rhetorical
orientation Perception as
provaccine
Plausibility (1–7) Willingness to transmit
(1–5)
Statement Frame U.S. UK Recall rate U.S. UK
Side effects
999 people out of 1,000 don’t
have any severe side effects
after being injected with a
vaccine.
&In favor of
vaccination
99% (N!73) M!4.87
SD !1.95
N!71
M!5.09
SD !1.71
N!65
69.64% (N!56) M!3.82
SD !1.11
N!72
M!3.75
SD !1.07
N!68
Only 1 individual out of 1,000
has some severe side effects
after being injected with a
vaccine.
'In favor of
vaccination
91% (N!67) M!4.88
SD !1.55
N!69
M!4.23
SD !1.56
N!65
80.00% (N!55) M!2.83
SD !1.16
N!70
M!3.02
SD !1.20
N!68
1 individual out of 1,000 has
some severe side effects after
being injected with a vaccine.
'Against
vaccination
46% (N!71) M!3.92
SD !1.77
N!66
M!3.94
SD !1.67
N!68
83.64% (N!110) M!2.56
SD !1.22
N!63
M!2.90
SD !1.30
N!71
Medical consensus
90% of medical scientists think
that vaccines are safe.
&In favor of
vaccination
97% (N!72) M!5.97
SD !1.37
N!69
M!5.86
SD !1.11
N!71
78.00% (N!50) M!3.87
SD !1.01
N!77
M!3.52
SD !1.03
N!63
Only 10% of medical scientists
don’t think that vaccines are
safe.
'In favor of
vaccination
90% (N!67) M!4.10
SD !1.98
N!67
M!3.79
SD !1.82
N!65
73.33% (N!60) M!2.62
SD !1.31
N!63
M!2.66
SD !1.38
N!70
10% of medical scientists don’t
think that vaccines are safe.
'Against
vaccination
42% (N!69) M!3.11
SD !1.95
N!66
M!3.54
SD !1.84
N!69
81.08% (N!111) M!2.03
SD !1.14
N!67
M!2.34
SD !1.05
N!65
Note. U.S. !United States; UK !United Kingdom.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4ALTAY AND MERCIER

were paid $0.30. Five participants failed to complete the survey, and
we removed 67 participants categorized as being against vaccination,
leaving 408 participants (223 women, M
age
!36.19, SD !11.29;
score on the Public Support Index: M!6.17, SD !0.84).
Design and Procedure
Overall design and procedure were identical to Experiment 1.
For the experimental task, each participant had to read one of the
statements from Table 1 and answer “How confident are you that
the statement is TRUE or FALSE?” on a 7-point Likert scale
ranging from 1 (I’m absolutely certain this statement is FALSE) to
7(I’m absolutely certain this statement is TRUE). The question is
identical to the one used by Fessler et al. (2014).
Results and Discussion
Plausibility scores of each statement are presented in Table 1. A
three-way ANOVA was conducted to compare the main effects of
rhetorical orientation (pro- or antivaccine), valence (positive or
negative), and type of statement (side effects or consensus) on the
plausibility ratings. There was a small effect of valence F(1,
404) !17.22, p$.001, (
2
!0.04, with participants rating
positive statements (M!5.41, SD !1.77) as more plausible than
negative statements (M!4.01, SD !1.92). The effect of rhetor-
ical orientation was small F(1, 404) !19.36, p$.001, (
2
!0.04,
with participants rating provaccination statements (M!4.96,
SD !1.85) as more plausible than antivaccination statements
(M!3.52, SD !1.90). The type of statement had no influence on
plausibility ratings F(1, 404) !0.66, p!.42, (
2
!0.001.
Contrary to the predictions of the negativity bias hypothesis
(H
3
), but in accordance with the literature on attribute framing
(H
6
), positively framed statements were found to be more plausible
than negatively framed ones. We also found that statements with a
rhetorical orientation in favor of vaccination were judged truer
than statements with a rhetorical orientation against vaccination,
supporting H
8
.
Experiment 3
In the third experiment we tested whether valence framing and
rhetorical orientation of the statements would affect memorization.
The literature on the negativity bias led us to expect that negatively
framed statements would be better memorized than positively
framed statements (H
4
).
Participants
A power analysis for logistic regression with an estimated odds
ratio of 2 (corresponding to an effect size between small and
medium (Chen, Cohen, & Chen, 2010), as found in a previous
experiment, see Bebbington et al., 2017), an "-level of 5%, and a
power of 80% suggested that we needed a minimum of 88 partic-
ipants per condition. Since we have three conditions (across the 6
dialogues), we needed a minimum of 264 participants. Considering
that we could have overestimated the expected effect size, we
recruited 280 participants who were paid $0.40. Four participants
failed to complete the survey, and we removed 33 participants
categorized as being against vaccination based on their answers to
the Public Support Index. We also removed 22 participants whose
answers in the free-recall task bore no relationship with the para-
graph to be memorized, leaving 221 participants (99 women,
M
age
!35.37, SD !10.81). The sample size is smaller compared
to the other experiments but the number of data points is similar,
since each participant had to read two statements instead of one.
Due to constraints on the organization of the dialogue to be
memorized, we could not counterbalance the three factors per-
fectly, so that negatively framed statements with a rhetorical
orientation against vaccination (1/1000 and 10%) received more
responses.
Materials
The statements were integrated in small dialogues, such as the
following:
Denis: Can you come to the game on Monday?
Raymond: I cannot, I have to take my kid to the pediatrician,
he’s getting his vaccines.
Denis: Do you know that [1 individual out 1,000 has
severe side effects after being injected with a
vaccine?]
Raymond: Maybe, but [90% of medical scientists think that
vaccines are safe.]
A statement on vaccines side effects was always associated with
a statement on the medical consensus with a different frame. In
half of the dialogues a statement on vaccine side effects was
presented first (as in the example) and in the other half statements
on the medical consensus was presented first. Six different dia-
logues were used (1/1,000 and 90%; 90% and 1/1,000; 10% and
999/1,000; 999/1,000 and 10%; only 1/1,000 and 10%; only 10%
and 1/1,000).
Design and Procedure
The overall design and procedure were similar to those of
Experiments 1 and 2. The experimental task consisted in asking
participants to read the dialogue and, on the next page, write what
they remembered of it. Such free-recall tasks are frequently used in
the literature on cultural evolution (e.g., Eriksson & Coultas,
2014).
Results and Discussion
Each statement was attributed a score of one when the partici-
pant correctly recalled it (including the correct statistic) and a
score of zero otherwise. The memorization scores of each state-
ment are presented in Table 1.
We conducted a logistic regression with participants as a ran-
dom effect to compare the main effects of valence, rhetorical
orientation, and statement type on the memorization score and
found no significant effects (valence (positive compared to nega-
tive): Z!0.002, SE !0.30, p!.99; rhetorical orientation (in
favor of vaccination compared to against vaccination): Z!'1.05,
SE !0.29, p!.29; type of statement (side effects compared to
scientific consensus): Z!0.09, SE !0.23, p!.93). The hypoth-
esis that negatively framed statements would be better remem-
bered was not supported (H
4
).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
5
FRAMING MESSAGES FOR VACCINATION SUPPORTERS

Experiment 4
In the fourth experiment we measured the effect of valence
framing and rhetorical orientation on the willingness of partici-
pants to share the statements. The negativity bias hypothesis led us
to expect that participants would be more willing to share nega-
tively framed statements than positively framed statements (H
5
).
The literature in political science led us to expect that participants
would be more willing to share statements with a rhetorical ori-
entation in favor of vaccination (H
9
).
Participants
A power analysis for ANOVA with an estimated effect size of
f!0.325 (based on previous findings we expected to find an effect
size between medium and large; Heath, 1996), an "-level of 5%,
and a power of 80% suggested that we needed a minimum of 77
participants per condition. Since we have six conditions, we
needed a minimum of 462 participants. Considering we would
need to exclude participants, we recruited 480 participants who
were paid $0.30. Six participants failed to complete the survey, and
we removed 62 participants categorized as being against vaccina-
tion, leaving 412 participants (195 women, M
age
!36.69, SD !
11.99; Public Support Index: M!6.14, SD !0.86).
Design and Procedure
The overall design and procedure were similar to those of
Experiments 1 and 2. The experimental task consisted in asking
participants to read one of the statements and to answer “How
likely would you be to pass along this statement to other people”
on a 5-point Likert scale ranging from 1 (very unlikely to pass
along)to5(very likely to pass along). This task is commonly used
in the literature on cultural evolution (e.g., see van Leeuwen,
Parren, Miton, & Boyer, 2018), and has proven to be correlated
(.44) with actual sharing patterns (Mosleh, Pennycook, & Rand,
2020).
Results and Discussion
The willingness to transmit ratings of each statement are pre-
sented in Table 1. A three-way ANOVA was conducted to com-
pare the main effects of valence, rhetorical orientation, and state-
ment type on the willingness to transmit rating. There was a
medium effect of valence F(1, 408) !63.62, p$.001, (
2
!0.13,
with participants more likely to transmit positively framed state-
ments (M!3.85, SD !1.05) than negatively framed statements
(M!2.51, SD !1.24). The effect of the rhetorical orientation was
small F(1, 408) !9.17, p!.003, (
2
!0.02 with participants
more willing to transmit provaccine statements (M!3.33, SD !
1.27) than antivaccine statements (M!2.30, SD !1.21). The type
of statement had a very small influence on willingness to transmit
rating F(1, 408) !4.77, p!.03, (
2
!0.007, with participants
being slightly more willing to transmit statements on vaccine side
effects (M!3.08, SD !1.28) than on the consensus (M!2.90,
SD !1.39).
Again, our results go against the predictions of the negativity
bias hypothesis (H
6
), with people being more willing to trans-
mit positively framed statements. Participants were also more
willing to transmit statements whose rhetorical orientation they
supported (H
9
).
Experiment 5
In the fifth experiment we replicated Experiment 2 on the
perceived plausibility of the statements with a different population
sample: in the United Kingdom rather than the United States (for
a review of the differences and commonalities between the two
countries, see Wheelock, Thomson, & Sevdalis, 2013). Despite the
differences between the two populations in attitudes toward vac-
cination, we expected to observe the same patterns.
Participants
Four hundred twenty-five participants were recruited through
Prolific Academic and paid £0.20. We removed 22 participants
categorized as being against vaccination, leaving 403 participants
(274 women, M
age
!35.89, SD !11.69; score on the Public
Support Index: M!6.19, SD !0.78).
Results and Discussion
The plausibility ratings of each statement are presented in Table
1. A three-way ANOVA was conducted to compare the main
effects of the three independent variables on plausibility ratings.
There was a medium effect of valence, F(1, 399) !53.50,
p$.001, (
2
!0.11 with participants rating positive statements
(M!5.49, SD !1.48) as more plausible than negative state-
ments (M!3.87, SD !1.74). The effects of rhetorical orien-
tation, F(1, 399) !1.78, p!.18, (
2
!0.003 and of type of
statement, F(1, 399) !0.001, p!.97, (
2
$.001 were not
significant. Experiment 5 thus replicates the effects of valence,
but not of rhetorical orientation. These results again support H
6
in opposition to H
3
, but they do not support H
8
.
The aggregated results of Experiments 2 and 5 are displayed in
Figure 1.
Experiment 6
The sixth experiment is an exact replication of Experiment 4
with a different population sample: crowdsourcing participants
recruited in the United Kingdom (through Prolific Academic)
rather than the United States.
Participants
Four hundred twenty-six participants were recruited and paid
£0.20. We removed 21 participants categorized as being against
vaccination, leaving 405 participants (290 M
age
!36.88, SD !
12.80; Public Support Index: M!6.09, SD !0.78).
Results and Discussion
Willingness to transmit ratings attributed to each statement are
presented in Table 1. A three-way ANOVA was conducted to
compare the main effects of the three independent variables on the
willingness to transmit ratings. There was a medium effect of
valence, F(1, 401) !30.50, p$.001, (
2
!0.07, with participants
being more willing to transmit positively framed statements (M!
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
6ALTAY AND MERCIER

3.64, SD !1.05) than negatively framed statements (M!2.73,
SD !1.26). The effect of rhetorical orientation was not signifi-
cant, F(1, 401) !2.2, p!.14, (
2
!0.004. The effect of type of
statement was small F(1, 401) !10.64, p!.001, (
2
!0.02, with
participants being more willing to transmit statements on vaccine
side effects (M!3.22, SD !1.25) than on the consensus (M!
2.83, SD !1.27).
Like Experiment 5, Experiment 6 replicates the effects of va-
lence, but not of rhetorical orientation. In both experiments, par-
ticipants were slightly more likely to share statements on vaccine
side effects than on consensus. These results again contradict
against H
5
, but they also do not support H
9
.
The aggregated results of Experiments 4 and 6 are displayed in
Figure 2.
Cross-Experiments Analysis
Finally, we assessed the influence of being exposed to the
statements on vaccine attitudes by merging the data of all the
experiments except Experiments 2 and 7 in which participants
were exposed to multiple statements. We did not exclude partici-
pants based on their attitudes toward vaccination since it was the
main variable of interest in this analysis. The total sample was
composed of 2,273 participants (1,274 women, M
age
!36.15,
Figure 1. Aggregated results of Experiments 2 and 5 (N!811): estimated plausibility (y-axis) as a function
of the type of statements (x-axis). The boxes represent the middle 50% of the scores; the lines that divide the
boxes are the median; upper and lower whiskers represent scores outside the middle 50% (except outliers); and
diamonds represent the means.
!
p$.05.
!!
p$.01.
!!!
p$.001.
Figure 2. Aggregated results of Experiments 4 and 6 (N!817): willingness to transmit (y-axis) as a function
of the type of statements (x-axis). The boxes represent the middle 50% of scores for the group; the lines that
divide the boxes are the median; upper and lower whiskers represent scores outside the middle 50% (except
outliers); and diamonds represent the means. ns !not significant.
!!
p$.01.
!!!
p$.001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
7
FRAMING MESSAGES FOR VACCINATION SUPPORTERS

SD !11.81; initial score on the Public Support Index: M!5.84,
SD !1.31).
We found that participants exposed to statements with a positive
valence reported more positive attitudes toward vaccination (M!
5.94, SD !1.22) than participants exposed to statements with a
negative valence (M!5.79, SD !1.35; Welch’s t(1655.4) !2.70,
p!.007, d!0.12; Figure 3). Participants exposed to statements with
apositiverhetoricalorientationwerenotmorelikelytoreportpositive
attitudes toward vaccination (M!5.85, SD !1.33) than participants
exposed to statements with a negative rhetorical orientation (M!
5.83, SD !1.27; Welch’s t(1578.4) !.28, p!.78).
These results are consistent with participants saying that they
find positively framed statements more plausible, and they are
important from a practical standpoint, given that participants re-
ported a greater willingness to share positively framed statements
(i.e., the most persuasive statements). These results also offer
evidence in favor of H
7
.
Experiment 7
In this last experiment, we measured the propensity of partici-
pants to use our statements in order to convince someone who
refuses to vaccinate, a few minutes after reading them. We com-
pared the arguments provided by participants who had read some
of our statements to the arguments of participants who had not read
the statements (control condition). This experiment offers a more
ecological setting than the memorization task of Experiment 2.
Participants
We increased the sample size compared to Experiment 2 to
make sure we would be sufficiently powered. Five hundred forty-
eight participants were recruited through Prolific Academic and
paid £0.20. We removed 24 participants categorized as being
against vaccination and two participants who failed at the attention
check, leaving 522 participants (358 women, M
age
!36.59, SD !
12.29; Public Support Index: M!6.52, SD !0.78).
Materials
We excluded statements with only because we wanted to focus
on the statements’ valence, which appeared to play a more impor-
tant effect on willingness to transmit. The statements were inte-
grated in vignettes such as the following:
Imagine the following situation:
You’re listening to a program on the radio in which doctors and
scientists are interviewed about vaccination. At some point, you hear:
“90 % of medical scientists think that vaccines are safe.”
A few minutes later you hear:
“1 individual out of 1,000 has some severe side effects after being
injected with a vaccine.”
A statement on vaccine side effects was always associated with
a statement on the medical consensus with a different frame. In
half of the dialogues the statement on vaccine side effects was
presented first (as in the example) and in the other half the
statement on the medical consensus was presented first.
Design and Procedure
The overall design and procedure were similar to those of
Experiment 2, with the exceptions that we introduced a control
group receiving no arguments, and we added a distraction task
lasting approximately four minutes. Half of the participants read
two statements on vaccination as described above; the other half
did not. The experimental task consisted in asking participant to
convince someone to vaccinate. Participants were presented with
the following text:
Imagine the following situation:
You are chatting with a friend of yours, and at some point she tells
you: “Vaccines are not safe, I won’t vaccinate myself or my kids.”
Concerned for your friend and her children, you try to convince her
that her opinion is misguided by answering:
[Free text entry]
Coding
We coded participants’ answer in the following way:
Side effects. The participant mentioned that very few people
suffer from side effects or that most people do not suffer from any
side effects.
Scientific consensus. The participant mentioned that most
medical scientists (or doctors) are in favor of vaccination or think
it is safe, or that very few medical scientists (or doctors) are not in
favor of vaccination or think it is unsafe.
Vaccines safety. The participant mentioned the fact that vac-
cines are safe, but did not mention scientific consensus or the side
effects.
Figure 3. Influence of valence framing on vaccine attitudes across Ex-
periments 1, 3, 4, 5, and 6 (N!2,273): participants’ attitudes toward
vaccination (y-axis) as a function of the valence of the statements’ partic-
ipants have been exposed to (x-axis). The boxes represent the middle 50%
of scores for the group; the lines that divide the boxes are the median;
upper and lower whiskers represent scores outside the middle 50% (except
outliers); and diamonds represent the means.
!!
p$.01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8ALTAY AND MERCIER

Herd immunity. The participants mentioned that vaccination
is not only useful for individuals but for society at large or others.
Diseases are riskier than vaccines. The participant argued
that diseases that vaccinate protect against are more dangerous
than the vaccines themselves.
Scientific evidence. The participant mentioned scientific ev-
idence in favor of vaccination such as scientific articles or exper-
iments.
Vaccines offer individual protection. The participant men-
tioned the protective nature of vaccines for individuals.
Disease are dangerous. The participant mentioned that some
diseases vaccines protect against are dangerous.
Not vaccinating is dangerous. The participant mentioned
that not vaccinating is dangerous (without explaining why).
Anecdotal evidence. The participant used anecdotal evidence
relying on firsthand experience to support vaccination.
No argument. What the participant wrote could not fit into
any of the categories above because it was not an argument.
Twenty percent of responses (104/522) were recoded by an
independent coder blind to our hypotheses. Coders agreed on
92.31% of the observations ()!0.85, SE $.01; 95% CI [0.83,
0.87]), the strength of agreement is considered to be “almost
perfect” (Landis & Koch, 1977).
Results and Discussion
All of the arguments used by participants can be found in Table 2.
In comparison to the control condition, participants in the ex-
perimental conditions were more likely to use arguments that
relied on the low rate of side effects (58 vs. 1, Z!0.21, p$.001)
or the scientific consensus on vaccination (22 vs. 2, Z!0.08, p$
.001).
In the experimental conditions, participants who received the
side effects statement positively framed were not more likely to
use it as an argument that participants who received it negatively
framed (31 vs. 27, Pearson’s chi-square test with Yates’ continuity
correction: #
2
(1, N!58) !.14, p!.70). Most participants
reported the side effects framed negatively (43 vs. 15, chi-square
goodness of fit test: #
2
(1, N!58) !13.52, p$.001, %!0.48),
but most of them added only (27/43) or other rhetorical devices
(8/43) to modify the rhetorical orientation of the statement.
Participants who received the consensus statement positively
framed were more likely to use it as an argument than participants
who received it negatively framed (19 vs. 3, Pearson’s chi-square
test with Yates’ continuity correction: #
2
(1, N!22) !10.91, p$
.001, %!0.70). All the mentions of the scientific consensus were
positively framed (22/22), as the three participants who had been
presented with the consensus negatively framed reframed it posi-
tively. When the participants made use of the statements that had
been provided—whether on the scientific consensus or the side
effects—nearly all gave them a rhetorical orientation in favor of
vaccination (76/80).
In the experimental conditions, participants who mentioned at
least one of the statements used on average more arguments (M!
2.36, SD !0.96) than participants who did not (M!1.35, SD !
0.61; Welch’s t(521) !7.85, p$.001, d!1.36), suggesting that
these statements reinforced their argumentative arsenal.
These results suggest that it is important to frame the consensus
statement positively because it increases the rate at which it is then
used by readers in their arguments. However, it is not clear that it
would be beneficial to positively frame the statement on the side
effects. Framing these statements negatively while maintaining a
provaccination rhetorical frame (e.g., with only, which many par-
ticipants spontaneously did) could minimize the effort required for
participants to appropriate the argument.
General Discussion
Our studies tested the effects of framing on the estimated
plausibility, memorization, and willingness to transmit of provac-
cination statements among provaccination individuals in the
United States and United Kingdom. The statements had two con-
tents, relating either to vaccine side effects or to the medical
consensus on vaccination. As established by Experiment 1, the
statements’ valence could be positive or negative, and the state-
ments could be perceived as strongly provaccination or as being
ambiguous between being pro- and antivaccination.
Contrary to previous studies (Bebbington et al., 2017; Fessler et
al., 2014), negatively framed statements, compared to positively
framed statements, were not better memorized (Experiment 3),
were deemed less plausible (Experiments 2 and 5), and were less
appealing to transmit (Experiments 4 and 6). These effects in
favor of positively framed statements ranged from small to
large. The size of these effects and the conflicting direction of
our results with the literature justified our decision to replicate
Experiments 2 and 4.
The effects of rhetorical orientation were broadly as expected.
Rhetorical orientation had no effect on memory, but participants
deemed more plausible (Experiment 2), and were more willing to
transmit (Experiment 4), statements whose rhetorical orientation
better fitted their own views, compared to statements that didn’t.
While these effects ranged from small to medium for participants
residing in the United States, there was no significant effect of
rhetorical orientation on plausibility or willingness of transmission
with the sample from the United Kingdom. It is unclear what drove
this potential difference between the United States and United
Kingdom populations. Nonetheless, this discrepancy does not af-
fect the practical conclusions that can be drawn from this study,
since the same statements were found to be more plausible and
more transmissible in the two populations.
Table 2
Types of Arguments Used by Participants in Experiment 7 as a
Function of the Condition
Arguments
Experimental
conditions
(N!264)
Control
condition
(N!258)
Side effects 58 1
Scientific consensus 22 2
Vaccines are safe 62 83
Herd immunity 45 60
Diseases are riskier than vaccines 66 41
Scientific evidence support vaccination 41 40
Vaccines offer individual protection 21 56
Diseases are dangerous 15 25
Not vaccinating is dangerous 27 31
Anecdotal evidence 8 8
No arguments 48 41
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
9
FRAMING MESSAGES FOR VACCINATION SUPPORTERS

Visual inspection of the results on plausibility and willingness to
transmit in Figures 1 and 2 reveal the striking effects of framing.
Comparing the most favorable frame (positive with provaccination
rhetorical orientation) to the most negative frame (negative with
ambiguous rhetorical orientation), we move from statements
deemed very plausible to statements deemed barely plausible, and
from statements participants are keen to transmit to statements
participants would rather not transmit.
These strong framing effects have at least two significant prac-
tical consequences. First, positively valenced statements prompted
participants to adopt a more positive attitude toward vaccination.
Second, in the case of statements about the scientific consensus,
positively valenced statements were more likely to be used by
participants to argue against someone who opposes vaccination.
Moreover, framing effects were also clearly perceptible in the way
participants framed the arguments they chose to use, with nearly
all arguments being framed with a provaccination rhetorical ori-
entation.
In relation with the literature on framing, our results confirm
that the choice of frame induces information leakage (McKenzie,
2004; McKenzie & Nelson, 2003; Sher & McKenzie, 2006, 2008),
as frames informed readers about the attitudes of the speakers
toward vaccination. Moreover, our results join a growing body of
research in questioning the generality of the negativity bias (for a
literature review, see Levin et al., 1998; for more recent findings,
see Berriche & Altay, 2019; Stubbersfield, Dean, Sheikh, Laland,
& Cross, 2019; van Leeuwen et al., 2018). It is thus important to
test, on a case-by-case basis, the effects of positive or negative
framing, and to acknowledge that framing could differently affect
memorization, plausibility evaluations, willingness to share, and
actual sharing behaviors. We thus encourage future research to
combine different measures and not to presume that framing will
have the same effect on multiple measures.
Some limitations of our study should be pointed out. First, we
obviously do not claim that our manipulation will be sufficient to
overcome vaccine hesitancy on its own—indeed, it is not even its
direct goal. However, we suggest that seemingly superficial fea-
tures of vaccine communication might have a substantial impact
on how some provaccine individuals process the communication,
making it more or less likely that they transmit it further. In
particular, some frames—in the case at hand, positive frames—
appear to make provaccine participants more likely to accept and
transmit provaccine messages, which are intrinsically desirable
outcomes. e.g., information campaigns often frame information
about vaccines side effects negatively, albeit with a provaccination
rhetorical orientation, as in the example above: “Severe side ef-
fects are extremely rare.” A positive frame (e.g., “The overwhelm-
ing majority of people do not have any side effects”) might be
more easily further transmitted by provaccination audiences.
Most research on vaccination-related messaging has focused on
its direct effects on vaccine-hesitant individuals, largely disregard-
ing vaccination supporters (Vanderslott, 2019). Yet, in most coun-
tries, the majority of the population is provaccine, suggesting that
these individuals could play a significant role in propagating
provaccination messages through informal communication. The
present study reveals that messages aimed at reinforcing the argu-
mentative tools of provaccination individuals must consider the
way the messages are framed. Finally, our research highlights the
importance of conducting experiments on the specific messages
we wish to provide to the population, rather than extrapolating
purely from past results, since the major framing effects we ob-
served went in the opposite direction from what was expected from
some of the existing literature.
References
Association of State and Territorial Health Officials. (2009). Communicating
effectively about vaccines: New communication resources for health offi-
cials. Retrieved from http://www.astho.org/Programs/Immunization/
Communicating-Effectively-About-Vaccines--Summary-of-a-Survey-of-U-
S--Parents-and-Guardians/
Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001).
Bad is stronger than good. Review of General Psychology, 5, 323–370.
http://dx.doi.org/10.1037/1089-2680.5.4.323
Bebbington, K., MacLeod, C., Ellison, T. M., & Fay, N. (2017). The sky
is falling: Evidence of a negativity bias in the social transmission of
information. Evolution and Human Behavior, 38, 92–101. http://dx.doi
.org/10.1016/j.evolhumbehav.2016.07.004
Berriche, M., & Altay, S. (2019). “Sister, I love you!” Phatic posts are
more successful on Facebook than health misinformation. Retrieved
from https://psyarxiv.com/nj2sr/
Betsch, C., & Wicker, S. (2012). E-health use, vaccination knowledge and
perception of own risk: Drivers of vaccination uptake in medical stu-
dents. Vaccine, 30, 1143–1148. http://dx.doi.org/10.1016/j.vaccine.2011
.12.021
Boyer, P. (2018). Minds make societies: How cognition explains the world
humans create. New Haven, CT: Yale University Press.
Brunson, E. K. (2013). The impact of social networks on parents’ vacci-
nation decisions. Pediatrics, 131, e1397–e1404.
Bullock, J. G., Gerber, A. S., Hill, S. J., & Huber, G. A. (2015). Partisan
bias in factual beliefs about politics. Quarterly Journal of Political
Science, 10, 519–578.
Casiday, R., Cresswell, T., Wilson, D., & Panter-Brick, C. (2006). A
survey of U. K. parental attitudes to the MMR vaccine and trust in
medical authority. Vaccine, 24, 177–184. http://dx.doi.org/10.1016/j
.vaccine.2005.07.063
Chanel, O., Luchini, S., Massoni, S., & Vergnaud, J.-C. (2011). Impact of
information on intentions to vaccinate in a potential epidemic: Swine-
origin Influenza A (H1N1). Social Science & Medicine, 72, 142–148.
http://dx.doi.org/10.1016/j.socscimed.2010.11.018
Chen, H., Cohen, P., & Chen, S. (2010). How big is a big odds ratio?
Interpreting the magnitudes of odds ratios in epidemiological studies.
Communications in Statistics Simulation and Computation, 39, 860–
864. http://dx.doi.org/10.1080/03610911003650383
Chung, C. M., & Darke, P. R. (2006). The consumer as advocate: Self-
relevance, culture, and word-of-mouth. Marketing Letters, 17, 269–279.
http://dx.doi.org/10.1007/s11002-006-8426-7
Das, J., & Das, S. (2003). Trust, learning, and vaccination: A case study of
a North Indian village. Social Science & Medicine, 57, 97–112. http://
dx.doi.org/10.1016/S0277-9536(02)00302-7
Dittmann, S. (2001). Vaccine safety: Risk communication—A global per-
spective. Vaccine, 19, 2446–2456. http://dx.doi.org/10.1016/S0264-
410X(00)00470-9
Dixon, G. N., & Clarke, C. E. (2013). Heightening uncertainty around
certain science: Media coverage, false balance, and the autism-vaccine
controversy. Science Communication, 35, 358–382. http://dx.doi.org/10
.1177/1075547012458290
Dubé, E., Gagnon, D., MacDonald, N. E., & the SAGE Working Group on
Vaccine Hesitancy. (2015). Strategies intended to address vaccine hes-
itancy: Review of published reviews. Vaccine, 33, 4191–4203. http://dx
.doi.org/10.1016/j.vaccine.2015.04.041
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
10 ALTAY AND MERCIER

Dunegan, K. J. (1993). Framing, cognitive modes, and image theory:
Toward an understanding of a glass half full. Journal of Applied Psy-
chology, 78, 491–503. http://dx.doi.org/10.1037/0021-9010.78.3.491
Dunegan, K. J. (1995). Image theory: Testing the role of image compati-
bility in progress decisions. Organizational Behavior and Human Deci-
sion Processes, 62, 79–86. http://dx.doi.org/10.1006/obhd.1995.1033
Eriksson, K., & Coultas, J. C. (2014). Corpses, maggots, poodles and rats:
Emotional selection operating in three phases of cultural transmission of
urban legends. Journal of Cognition and Culture, 14, 1–26. http://dx.doi
.org/10.1163/15685373-12342107
Faasse, K., Chatman, C. J., & Martin, L. R. (2016). A comparison of
language use in pro- and anti-vaccination comments in response to a
high profile Facebook post. Vaccine, 34, 5808–5814. http://dx.doi.org/
10.1016/j.vaccine.2016.09.029
Fadda, M., Allam, A., & Schulz, P. J. (2015). Arguments and sources on
Italian online forums on childhood vaccinations: Results of a content
analysis. Vaccine, 33, 7152–7159. http://dx.doi.org/10.1016/j.vaccine
.2015.11.007
Feinberg, Y., Pereira, J. A., Quach, S., Kwong, J. C., Crowcroft, N. S.,
Wilson, S. E., . . . the Public Health Agency of Canada/Canadian
Institutes of Health Research Influenza Research Network (PCIRN)
Program Delivery and Evaluation Group. (2015). Understanding public
perceptions of the HPV vaccination based on online comments to
Canadian news articles. PLoS ONE, 10, e0129587. http://dx.doi.org/10
.1371/journal.pone.0129587
Fessler, D. M., Pisor, A. C., & Navarrete, C. D. (2014). Negatively-biased
credulity and the cultural evolution of beliefs. PLoS ONE, 9, e95167.
http://dx.doi.org/10.1371/journal.pone.0095167
Freed, G. L., Clark, S. J., Butchart, A. T., Singer, D. C., & Davis, M. M.
(2011). Sources and perceived credibility of vaccine-safety information
for parents. Pediatrics, 127, S107–S112. http://dx.doi.org/10.1542/peds
.2010-1722P
Goldberg, M. H., van der Linden, S., Maibach, E., & Leiserowitz, A.
(2019). Discussing global warming leads to greater acceptance of cli-
mate science. Proceedings of the National Academy of Sciences of the
United States of America, 116, 14804–14805. http://dx.doi.org/10.1073/
pnas.1906589116
Heath, C. (1996). Do people prefer to pass along good or bad news?
Valence and relevance of news as predictors of transmission propensity.
Organizational Behavior and Human Decision Processes, 68, 79–94.
http://dx.doi.org/10.1006/obhd.1996.0091
Hennig-Thurau, T., Gwinner, K. P., Walsh, G., & Gremler, D. D. (2004).
Electronic word-of-mouth via consumer-opinion platforms: What moti-
vates consumers to articulate themselves on the internet? Journal of
Interactive Marketing, 18, 38–52. http://dx.doi.org/10.1002/dir.10073
Horne, Z., Powell, D., Hummel, J. E., & Holyoak, K. J. (2015). Countering
antivaccination attitudes. Proceedings of the National Academy of Sci-
ences of the United States of America, 112, 10321–10324. http://dx.doi
.org/10.1073/pnas.1504019112
Jiménez, Á. V., Mesoudi, A., & Tehrani, J. (2020). No evidence that
omission and congruity biases affect the perception and recall of
vaccine-related information. PLoS ONE, 15(3), e0228898. http://dx.doi
.org/10.1371/journal.pone.0228898
Jiménez, Á. V., Stubbersfield, J. M., & Tehrani, J. J. (2018). An experi-
mental investigation into the transmission of antivax attitudes using a
fictional health controversy. Social Science & Medicine, 215, 23–27.
http://dx.doi.org/10.1016/j.socscimed.2018.08.032
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of
decision under risk. Econometrica, 47, 263–292. http://dx.doi.org/10
.2307/1914185
Katz, E., & Lazarsfeld, P. F. (1955). Personal influence: The part played
by people in the flow of mass communications. Glencoe, IL: Free Press.
Kennedy, A., Basket, M., & Sheedy, K. (2011). Vaccine attitudes, con-
cerns, and information sources reported by parents of young children:
Results from the 2009 HealthStyles survey. Pediatrics, 127, S92–S99.
Knapp, M. L., & Daly, J. A. (2002). Handbook of interpersonal commu-
nication. Thousand Oaks, CA: Sage.
Landis, J. R., & Koch, G. G. (1977). The measurement of observer
agreement for categorical data. Biometrics, 33, 159–174.
Larson, H. J., Jarrett, C., Eckersberger, E., Smith, D. M., & Paterson, P.
(2014). Understanding vaccine hesitancy around vaccines and vaccina-
tion from a global perspective: A systematic review of published liter-
ature, 2007–2012. Vaccine, 32, 2150–2159. http://dx.doi.org/10.1016/j
.vaccine.2014.01.081
Leask, J. (2011). Target the fence-sitters. Nature, 473, 443–445. http://dx
.doi.org/10.1038/473443a
Lei, Y., Pereira, J. A., Quach, S., Bettinger, J. A., Kwong, J. C., Corace, K.,
. . . the Public Health Agency of Canada/Canadian Institutes of Health
Research Influenza Research Network Program Delivery and Evaluation
Group. (2015). Examining perceptions about mandatory influenza vac-
cination of healthcare workers through online comments on news stories.
PLoS ONE, 10, e0129993. http://dx.doi.org/10.1371/journal.pone
.0129993
Levin, I. P. (1987, May). Associative effects of information framing on
human judgments. Paper presented at the annual meeting of the Mid-
western Psychological Association, Chicago, IL.
Levin, I. P., & Gaeth, G. J. (1988). Framing of attribute information before
and after consuming the product. Journal of Consumer Research, 15,
374–378.
Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not
created equal: A typology and critical analysis of framing effects.
Organizational Behavior and Human Decision Processes, 76, 149–188.
http://dx.doi.org/10.1006/obhd.1998.2804
Levin, I. P., Schnittjer, S. K., & Thee, S. L. (1988). Information framing
effects in social and personal decisions. Journal of Experimental Social
Psychology, 24, 520–529. http://dx.doi.org/10.1016/0022-1031(88)
90050-9
Linville, P. W., Fischer, G. W., & Fischhoff, B. (1993). AIDS risk percep-
tions and decision biases. Mahwah, NJ: Erlbaum.
MacDonald, N. E., & the SAGE Working Group on Vaccine Hesitancy.
(2015). Vaccine hesitancy: Definition, scope and determinants. Vaccine,
33, 4161–4164. http://dx.doi.org/10.1016/j.vaccine.2015.04.036
Marteau, T. M. (1989). Framing of information: Its influence upon deci-
sions of doctors and patients. British Journal of Social Psychology, 28,
89–94. http://dx.doi.org/10.1111/j.2044-8309.1989.tb00849.x
McKeever, B. W., McKeever, R., Holton, A. E., & Li, J.-Y. (2016). Silent
majority: Childhood vaccinations and antecedents to communicative
action. Mass Communication & Society, 19, 476–498. http://dx.doi.org/
10.1080/15205436.2016.1148172
McKenzie, C. R. M. (2004). Framing effects in inference tasks—And why
they are normatively defensible. Memory & Cognition, 32, 874–885.
http://dx.doi.org/10.3758/BF03196866
McKenzie, C. R. M., & Nelson, J. D. (2003). What a speaker’s choice of
frame reveals: Reference points, frame selection, and framing effects.
Psychonomic Bulletin & Review, 10, 596–602. http://dx.doi.org/10
.3758/BF03196520
Miton, H., & Charbonneau, M. (2018). Cumulative culture in the labora-
tory: Methodological and theoretical challenges. Proceedings of the
Royal Society B: Biological Sciences, 285, 20180677.
Miton, H., & Mercier, H. (2015). Cognitive obstacles to pro-vaccination
beliefs. Trends in Cognitive Sciences, 19, 633–636. http://dx.doi.org/10
.1016/j.tics.2015.08.007
Morin, O. (2015). How traditions live and die. New York, NY: Oxford
University Press.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
11
FRAMING MESSAGES FOR VACCINATION SUPPORTERS

Mosleh, M., Pennycook, G., & Rand, D. (2020). Self-reported willingness
to share political news articles in online surveys correlates with actual
sharing on Twitter. PLoS ONE, 15, e0228882.
Nichter, M. (1995). Vaccinations in the Third World: A consideration of
community demand. Social Science & Medicine, 41, 617–632. http://dx
.doi.org/10.1016/0277-9536(95)00034-5
Nyhan, B., Reifler, J., Richey, S., & Freed, G. L. (2014). Effective
messages in vaccine promotion: A randomized trial. Pediatrics, 133,
e835–e842. http://dx.doi.org/10.1542/peds.2013-2365
Omer, S. B., Salmon, D. A., Orenstein, W. A., deHart, M. P., & Halsey, N.
(2009). Vaccine refusal, mandatory immunization, and the risks of
vaccine-preventable diseases. The New England Journal of Medicine,
360, 1981–1988. http://dx.doi.org/10.1056/NEJMsa0806477
Opel, D. J., & Marcuse, E. K. (2013). Window or mirror: Social networks’
role in immunization decisions. Pediatrics, 131, e1619–e1620. http://
dx.doi.org/10.1542/peds.2013-0531
Paolacci, G., & Chandler, J. (2014). Inside the Turk. Current Directions in
Psychological Science, 23, 184–188. http://dx.doi.org/10.1177/
0963721414531598
Pereira, J. A., Quach, S., Dao, H. H., Kwong, J. C., Deeks, S. L., Crow-
croft, N. S., . . . the Public Health Agency of Canada/Canadian Institutes
of Health Research Influenza Research Network (PCIRN) Program
Delivery, Evaluation Group. (2013). Contagious comments: What was
the online buzz about the 2011 Quebec measles outbreak? PLoS ONE, 8,
e64072. http://dx.doi.org/10.1371/journal.pone.0064072
Pluviano, S., Watt, C., & Della Sala, S. (2017). Misinformation lingers in
memory: Failure of three pro-vaccination strategies. PLoS ONE, 12,
e0181640. http://dx.doi.org/10.1371/journal.pone.0181640
Sadaf, A., Richards, J. L., Glanz, J., Salmon, D. A., & Omer, S. B. (2013).
A systematic review of interventions for reducing parental vaccine
refusal and vaccine hesitancy. Vaccine, 31, 4293–4304. http://dx.doi
.org/10.1016/j.vaccine.2013.07.013
Salmon, D. A., Moulton, L. H., Omer, S. B., DeHart, M. P., Stokley, S., &
Halsey, N. A. (2005). Factors associated with refusal of childhood
vaccines among parents of school-aged children: A case-control study.
Archives of Pediatrics & Adolescent Medicine, 159, 470–476. http://dx
.doi.org/10.1001/archpedi.159.5.470
Schoeppe, J., Cheadle, A., Melton, M., Faubion, T., Miller, C., Matthys, J.,
& Hsu, C. (2017). The immunity community: A community engagement
strategy for reducing vaccine hesitancy. Health Promotion Practice, 18,
654–661. http://dx.doi.org/10.1177/1524839917697303
Sher, S., & McKenzie, C. R. (2006). Information leakage from logically
equivalent frames. Cognition, 101, 467–494. http://dx.doi.org/10.1016/
j.cognition.2005.11.001
Sher, S., & McKenzie, C. R. M. (2008). Framing effects and rationality. In
N. Chater & M. Oaksford (Eds.), The probabilistic mind: Prospects for
Bayesian cognitive science (pp. 79–96). Oxford, UK: Oxford University
Press.
Streefland, P., Chowdhury, A. M. R., & Ramos-Jimenez, P. (1999). Pat-
terns of vaccination acceptance. Social Science & Medicine, 49, 1705–
1716. http://dx.doi.org/10.1016/S0277-9536(99)00239-7
Stubbersfield, J. M., Dean, L. G., Sheikh, S., Laland, K. N., & Cross, C. P.
(2019). Social transmission favours the ‘morally good’ over the ‘merely
arousing’. Palgrave Communications, 5, 70. http://dx.doi.org/10.1057/
s41599-019-0269-y
Sundaram, D.S., Mitra, K., & Webster, C. (1998). Word-of-mouth com-
munications: A motivational analysis. Advances in Consumer Research,
25, 527–531.
Tversky, A., & Kahneman, D. (1981). The framing of decisions and the
psychology of choice. Science, 211, 453–458. http://dx.doi.org/10.1126/
science.7455683
van der Linden, S. L., Clarke, C. E., & Maibach, E. W. (2015). Highlight-
ing consensus among medical scientists increases public support for
vaccines: Evidence from a randomized experiment. BMC Public Health,
15, 1207. http://dx.doi.org/10.1186/s12889-015-2541-4
Vanderslott, S. (2019). Exploring the meaning of pro-vaccine activism
across two countries. Social Science & Medicine, 222, 59–66.
van Leeuwen, F., Parren, N., Miton, H., & Boyer, P. (2018). Individual
choose-to-transmit decisions reveal little preference for transmitting
negative or high-arousal content. Journal of Cognition and Culture, 18,
124–153. http://dx.doi.org/10.1163/15685373-12340018
Wheelock, A., Thomson, A., & Sevdalis, N. (2013). Social and psycho-
logical factors underlying adult vaccination behavior: Lessons from
seasonal influenza vaccination in the U.S. and the U. K. Expert Review
of Vaccines, 12, 893–901. http://dx.doi.org/10.1586/14760584.2013
.814841
Wilson, D. K., Kaplan, R. M., & Schneiderman, L. J. (1987). Framing of
decisions and selections of alternatives in health care. Social Behaviour,
2, 51–59.
World Health Organization. (2016). Vaccine safety communication: Guide
for immunization programme managers and national regulatory author-
ities. Manila, Philippines: WHO Regional Office for the Western Pa-
cific.
Yaqub, O., Castle-Clarke, S., Sevdalis, N., & Chataway, J. (2014). Atti-
tudes to vaccination: A critical review. Social Science & Medicine, 112,
1–11. http://dx.doi.org/10.1016/j.socscimed.2014.04.018
Received December 4, 2019
Revision received February 14, 2020
Accepted February 22, 2020 !
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12 ALTAY AND MERCIER