Promoting HPV vaccination on social media: interactive effects of threat, efficacy and social cues

ABSTRACT

Background

Women are currently facing a tremendous threat of cervical cancer globally. Social media health campaigns have the potential to shape public health behaviors. This study explores the effects of cervical-cancer-related fear appeal messages with social cues on social media using the extended parallel processing model (EPPM).

Method

We use a 2 (threat: present vs. absent) × 2 (efficacy: present vs. absent) × 2 (social cues: high vs. low) factorial experimental design to examine the effects of fear appeal messages with social cues on behavioral intention to receive the human papillomavirus (HPV) vaccination.

Results

There was a significant main effect of threat on the intention to receive HPV vaccination. Additionally, a significant three-way interactive effect among threat, efficacy, and social cues was detected.

Conclusion

Women exposed to threat messages had a higher intention of HPV vaccination compared to those who were exposed to non-threat messages. Furthermore, with the low number of likes, women who were exposed to messages containing both threat and efficacy tended to have the highest intention of HPV vaccination.

Practice Implications

When conducting fear appeal campaigns on social media, the side effects of number of likes should be recognized. For vaccination promotion campaigns, the efficacy information should be more specific and audience-centered.

KEYWORDS:

• Social media
• health campaign
• fear appeal
• EPPM
• social cues
• cervical cancer

Introduction

Cervical cancer is one of the most common and invasive cancers among women.¹There were 569,000 new cases of cervical cancer and 311,000 deaths related to cervical cancer worldwide in 2018.² In China, 106,430 women are diagnosed with cervical cancer annually, accounting for about 19.71% of the global incidents.^3,4 Compared with other cancers, cervical cancer ranked third in instances of cancers among Chinese women aged 15 to 44 in 2019.²

Cervical cancer is caused by the human papillomavirus (HPV) infection, and currently, the HPV vaccination is the most effective means of prevention.⁵ HPV vaccinations in mainland China were first approved in 2016. However, due to scant knowledge and high price of the HPV vaccination, only about 3% of Chinese women have been vaccinated until 2019.^6–8 It is, therefore, imperative and urgent to promote the HPV vaccination among Chinese women.

In the past decade, social media has become a platform in which a large amount of health information is generated and exchanged.^9–12 Additionally, many social media sites have been used for health promotion and campaigns.^13,14 Extant research has found that the effectiveness of health campaigns on between traditional and social media did not differ.¹⁵ However, some studies have found that health campaign messages from different sources might affect readers’ credibility perceptions, which in turn, could influence the effectiveness of health campaigns.^16,17 In addition, compared to the content in traditional media, social media content includes not only texts and images, but also various social cues with interactive implications hidden behind them, such as user ratings and numbers of friends, likes, comments, and reposts. Previous studies have indicated that social cues could enrich the meaning of the texts and images, which may alter its effectiveness.¹⁵ However, little research has been conducted to examine the roles of social cues in social media health campaigns.¹⁸ The current study seeks to develop the extended parallel process model (EPPM) with social cues to examine the effects of fear appeals and social cues as well as their interactive effects on the intention of Chinese women to receive the HPV vaccination.

Theoretical model and hypothesis development

Extended parallel process model (EPPM)

Witte proposed the extended parallel process model (EPPM), which provides a persuasive message design framework for deciding when fear appeals can be used and how they should be constructed.¹⁹ The model divides the cognitive process that addresses fear appeals into two stages – the threat appraisal and the efficacy appraisal.

In the face of threat as an external stimulus, individuals appraise the severity of and their susceptibility to the stimulus.¹⁹ Severity refers to the magnitude or seriousness of a threat, i.e., the anticipated negative consequences. Susceptibility refers to the expected possibility of experiencing the threat, i.e., the risk of harm perceived by individuals. These two aspects subsequently form the perceived threat, which determines the motivation to respond to fear-arousing information.²⁰ If the threat is perceived as negligible, then fear cannot be stimulated, and the individual loses the motivation to further process the information.¹⁴ In contrast, if the fear is stimulated, then efficacy appraisal would take place. In other words, when the targeted audience is faced with a threat, fear guides them to appraise whether there are possible measures to be taken to protect them from it.

Efficacy appraisal involves response efficacy and self-efficacy: the former refers to the effectiveness of the recommended response in the aversion to the threat, i.e., whether the suggested measure is feasible, whereas the latter refers to the capacity to perform behaviors recommended by messages, i.e., whether the suggested measure could be easily implemented.¹⁴

These two aspects could form individual’s efficacy perception. In this appraisal, if the individual believes that the recommended measures provided by the stimulus are effective and he or she has the capability to take such a measure, the person will engage in danger control and perform the recommended behavior in order to reduce the threat.^21–23 At this point, the individual accepts the recommended strategy and actively responds to the danger. On the contrary, if the individual believes that the strategy would not be effective in helping to cope with the potential threat, or if the individual does not have the capability to perform the recommended health-oriented actions, defensive behaviors are adopted, placing the person into a passive mind-set, whereby no action is taken.

Many studies of fear appeals have documented that EPPM-based messages can enhance recipients’ perceived threat and perceived efficacy, thereby facilitating intentions to perform certain health protective behaviors.^10,24,25 Specifically, both threat and efficacy have main effects and interaction effects. Higher levels of both threat and efficacy, in various combinations, is more persuasive.^10,26 In other words, individuals who receive messages including both threat and efficacy components have a higher degree of behavioral intention than messages containing only one or no component.

Some studies have also used the EPPM to design persuasive messages in promoting vaccination.^17,27,28 These messages highlight the risk of disease (i.e., threat) and provide a way to get vaccinations (i.e., efficacy). This has been found to be effective in increasing the intention to receive vaccination.

Social cues in social media

In addition to the message itself, social media content usually includes some system- and user-generated cues such as the number of followers, likes, comments, and reposts.²⁹ Sundar³⁰ indicated that these cues were design features of technologies that can highlight the underlying affordances and serve as triggers for cognitive heuristics that affect people’s impressions of the quality and credibility of the underlying information. Many studies have found that social cues not only can direct attention to certain topics, but also influence their perceived validity and trustworthiness.^31,32

The most common social cues in social media include the number of comments, reposts, and likes. They can serve as indicators of information engagement.³³ For example, reposting means the audience has read the original post, whereas commenting involves more cognitive efforts on the part of the audience given that it generally requires entering new content that adds value to the original post, and liking represents the popularity of a message.³⁴ In some studies, the number of likes was used as a proxy for social endorsement, indicating the level of public agreement with a certain message.^35,36 Several studies have revealed that the more popular a message (i.e., receives a greater number of likes), the more valuable the message is believed to be. In other words, messages with a greater number of likes tend to affect the recipients more easily and deeply.

Furthermore, studies have found that the number of likes usually plays a significant role in social media health campaigns. Borah and Xiao examined the perceived credibility of social media health information and found that the gain-frame messages with a greater number of likes had a higher level of perceived credibility.³⁶ Jin, Phua, and Lee examined the effects of user-generated messages on attitudes toward breast feeding among female college students.³⁷ They found that strategic messages with a greater number of likes tended to have a higher level of perceived credibility and produce positive attitudes. However, no study has examined the interactive effects between social cues and fear appeals on behavioral intentions.

Based on the above literature review, this research proposes the following hypotheses and question:

H1: Women exposed to information with threat will have a greater intention to be vaccinated against HPV.

H2: Women exposed to information with efficacy will have a greater intention to be vaccinated against HPV.

H3: Women exposed to information with both threat and efficacy will have the highest intention of being vaccinated against HPV.

H4: Women exposed to messages with a greater number of likes will have a greater intention of being vaccinated against HPV.

RQ1: How do threat, efficacy, and the number of likes interact to affect the intention to receive a vaccination against HPV among Chinese women?

Materials and methods

Participants

A snowball sampling method was used to recruit Chinese female participants through the Internet. This method has also been used in previous experimental studies.^14,38 Specifically, the authors first sent the link of the experiment to friends and acquaintances via social media and e-mail, and then asked them to share it by forwarding the link to their friends and acquaintances. A total of 478 women aged 16–45 years old (M= 23.05, SD = 4.52) who had never received an HPV vaccine were selected, due to the higher risk of cervical cancer among women older than 16 and their qualification for all three types of HPV vaccine. In addition, in order to obtain the general attitude of the females toward cervical cancer and HPV vaccination, individuals who were studying or working in the related fields of medicine, biomedicine and health care were excluded from this experiment.

Experiment design

This study used Weibo as the experimental context. Serving as a Chinese version of Twitter, Weibo has become a popular social media platform in China since it was launched in 2009.²⁹ The study was a 2 (threat: present vs. absent) × 2 (efficacy: present vs. absent) × 2 (social cues: high vs. low) factorial experiment that used a between-subjects design. All participants were randomly assigned to one of eight experimental groups. After being exposed to the stimulus, participants were asked to indicate their intention to receive the HPV vaccination.

Procedure

One hyperlink of the eight available was sent to participants via social media. An online informed consent form was presented for the participants to complete. Participants were then randomly exposed to one of the eight posters (i.e., the stimulus) in the web-based questionnaire, which automatically assigned the stimuli to the participants randomly. Participants could only enter the final stage of the questionnaire after viewing the stimulus picture. They were then asked to indicate the perceived threat and efficacy as well as their intention to receive the HPV vaccination.

Stimuli

Posters presenting eight different versions of Weibo posts were created. All the stimulus posters first introduced cervical cancer and the HPV vaccination. In addition, the posters contained different levels of threat, efficacy, and social cues. Social cues were operationalized by manipulating the number of likes (high vs. low). A total of 12,069 likes were used in the high-like condition, and 2 were used in the low-like condition (see Appendix).^32,36,37 The number of likes was adjusted after pilot testing.

Measures

First, in order to check randomization, previous studies were followed.^39,40 Demographic variables were measured, which included age (M= 23.05, SD = 4 52) and education level (Mdn = 3 undergraduate). Additionally, social media use was measured by asking participants to indicate the frequency of their daily social media use within a six month period using a 7-point Likert scale (1 = “not at all” to 7 = “very frequently”) (M = 6.65, SD = .88). Two items were used to measure prior knowledge based on a 7-point Likert scale (1 = “strongly unfamiliar” to 7 = “strongly familiar”):“To what extent are you familiar with 1) cervical cancer and 2) HPV vaccines” (M = 3.78, SD = 1.29, Pearson’ r = .83).

The Risk Behavior Diagnosis Scale (RDBS) was modified and used to measure the threat and efficacy levels of the stimuli.⁴¹ All measures used a 7-point Likert scale (1 = “strongly disagree” to 7 = “strongly agree”).

Threat

Three items were initially used to measure susceptibility. For example, the poster shows that I am at risk for experiencing cervical cancer. The three items on the severity scale. For example, the poster shows that cervical cancer is a serious threat. Scale reliability was measured for all six items (M= 4.70, SD = 1.38, Cronbach’s α = .85).

Efficacy

Response efficacy was measured by three items. For example, the poster shows that the HPV vaccine is important to limiting the spread of HPV. Three items were used to measure self-efficacy. For example, the poster teaches me how to begin HPV vaccination step by step. Scale reliability was measured for all six items (M= 5.04, SD = 1.28, Cronbach’s α = .87).

Behavioral intentions

Three items from a previous study⁴² were modified to measure HPV prevention intentions: a) I intend to begin the HPV vaccination; b) I will try to begin the HPV vaccination; and c) I plan to begin the HPV vaccination (M= 5.59, SD = 1.19, Cronbach’s α = .95).

Data analysis

Basic descriptive analysis, t-tests, and analysis of variance (ANOVA) through software SPSS 23.0 were conducted to check randomization and manipulation as well as address the hypotheses and research questions.

Randomization checks

Following previous studies,^43,44 we conducted ANOVA to compare demographic variables, social media use and prior knowledge between different message conditions. The results revealed that participants in eight groups did not significantly differ in these variables (F_age(7, 470) = 0.46, p = .86; F_education(7, 470) = 0.49, p = .84; F_{social media use} (7, 470) = 1.81, p = .08;F_knowledge = 0.99, p = .44), suggesting that randomization was successful.

Manipulation checks

Since eight stimuli were used in this study, manipulation checks were conducted to ensure participants’ correct perception of the threat and efficacy levels of the stimuli. Specifically, the levels of threat and efficacy were tested. We used independent sample t-tests to examine participants’ perception of threat and efficacy in all instances. The results showed that perceptions of stimulus with threat (M= 5.23, SD = 0.94) scored significantly higher than stimulus without threat (M= 4.16, SD = 1.53), t(478) = 9.18, p< .01. Similarly, the results revealed that perceptions of stimulus with efficacy (M= 5.52, SD = 0.85) scored significantly higher than stimulus without efficacy (M= 4.57, SD = 1.45), t(478) = 8.77, p< .01 (see Table 1). As such, participants’ perceptions of the threat and efficacy of the stimuli were consistent with the levels of threat and efficacy created for this study; the manipulation was successful.

Table 1. Manipulation checks

	Perception	N	Mean	SD	t	df	p
Threat	Present	239	5.23	0.941	9.184	476	.00***
	Absent	239	4.16	1.532
Efficacy	Present	238	5.52	0.849	8.769	476	.00***
	Absent	240	4.57	1.454

*p < .05, **p< .01, and ***p< .001.

Results

To assess the hypotheses and research question, we conducted a three-way analysis of variance (ANOVA) to examine the effects of stimulus messages with different numbers of likes on the participants’ behavioral intention to receive the HPV vaccination. The ANOVA model revealed a significant main effect of threat on participants’ intention to receive HPV vaccination, which indicated that participants exposed to threat messages have higher intention to receive the HPV vaccination (M= 5.72, SD = 1.12), while those who are exposed to non-threatening messages have a lower intention to receive the HPV vaccination (M= 5.38, SD = 1.21), F (1, 465) = 11.74, p < .001, partial η² = .02. Thus, the results supported H1. However, the models showed that the main effects of efficacy and likes on behavioral intention were not significant. Thus, the results failed to support H2 and H4. In addition, H3 proposed that there was a two-way interaction effect between threat and efficacy, but the analysis demonstrated that this interaction on behavioral intentions was not significant, F (1, 465) = .20, p = .65, partial η_p² < .00. Thus, H3 was not supported.

Furthermore, RQ1 asked how threat, efficacy, and the number of likes interact to affect the intention to receive vaccination against HPV. Our findings showed that there was a significant three-way interaction (F (1, 465) = 3.04, p < .05, partial η² = .01). When the number of likes is high, women who are exposed to threat messages without efficacy had the higher intention to receive the HPV vaccination than those who were exposed to other messages. However, with a low number of likes, those who were exposed to messages containing threat and efficacy tended to have a higher intention to receive the HPV vaccination than other messages (See Table 2). The ANOVA test results are displayed in Table 3. The significant interactions are depicted in Figures 1 and 2.

Table 2. Descriptive statistics for intention to receive HPV vaccination

Likes	Threat	Efficacy	Mean	SD	N
High	Present	Present	5.61	1.17	59
		Absent	5.85	1.00	60
		Total	5.73	1.09	119
	Absent	Present	5.48	0.97	60
		Absent	5.29	1.15	60
		Total	5.38	1.06	120
	Total	Present	5.55	1.07	119
		Absent	5.57	1.11	120
		Total	5.56	1.09	239
Low	Present	Present	5.82	1.07	60
		Absent	5.58	1.01	60
		Total	5.70	1.04	120
	Absent	Present	5.36	1.10	59
		Absent	5.38	1.16	60
		Total	5.37	1.13	119
	Total	Present	5.59	1.10	119
		Absent	5.48	1.09	120
		Total	5.54	1.10	239
Total	Present	Present	5.72	1.12	119
		Absent	5.72	1.01	120
		Total	5.72	1.06	239
	Absent	Present	5.42	1.03	119
		Absent	5.33	1.09	120
		Total	5.38	1.213	239
	Total	Present	5.57	1.08	238
		Absent	5.52	1.10	240
		Total	5.55	1.09	478

Table 3. Threat × Efficacy × Likes factorial analysis of variance for intention to receive HPV vaccination

Variables	df	F	p	ηp^2
Threat	1	11.74	.00***	.024
Efficacy	1	.20	.65	.00
Likes	1	.05	.82	.00
Threat × Efficacy	1	.19	.66	.00
Threat × Likes	1	.01	.92	.00
Efficacy × Likes	1	.441	.50	.00
Threat × Efficacy × Likes	1	3.04	.04*	.01
Error	465

*p < .05, **p< .01, and ***p< .001.

Figure 1. Three-way interaction of threat and efficacy on intention to receive HPV vaccination in the high ‘likes’ condition.

The y-axis ranged from 1 to 7, with higher scores representing higher intention to receive HPV vaccination).

Figure 2. Three-way interaction of threat and efficacy on intention to receive HPV vaccination in the low ‘likes’ condition.

The y-axis ranged from 1 to 7, with higher scores representing higher intention to receive HPV vaccination).

Discussion

Through a 2 × 2 × 2 factorial experiment, this study explored the causal relationship between the provision of fear appeal messages on cervical cancer and the intention to receive the HPV vaccination under two different levels of social cues based on the EPPM . First, the main effects of threat and efficacy were examined. The results indicated that threat had a significant main effect on HPV vaccination intention, which was consistent with the results of previous studies.^11,40,45,46 According to the EPPM, when exposed to messages with a high level of threat, people are more likely to accept the recommended health advice to control the potential danger. However, the findings revealed a non-significant main effect of efficacy. It is plausible that the efficacy information on the stimulus picture only provided a way to start the HPV vaccination. Many recipients may be unclear about when people should get the HPV vaccine, how many HPV vaccine injections people should take, and how much the HPV vaccine costs.⁴⁷ Thus, while the messages containing efficacy could increase individuals’ general perceive efficacy of HPV vaccination, the intention to receive the HPV vaccination was not significantly enhanced.

In addition, the study detected a significant three-way interaction among threat, efficacy, and social cues. The results revealed that messages including threat and efficacy had the best persuasive effect only when the number of likes was low, which is in line with the EPPM. This can be explained by the heuristic-systematic model (HSM). According to Chaiken’s HSM,⁴³ an individual will use either heuristic or systematic information processing to evaluate messages in order to arrive at judgments and attitudes.⁴⁴Serving as social cues, a high number of likes highlights the popularity of online messages and leads people to focus on others’ agreement and endorsement to the messages, rather than the message content itself. In other words, when the number of likes is high, people tend to apply the heuristic information processing to assess the message, leading to the difficulty in behavioral change.³⁴

On the contrary, with a low number of likes, people are less likely to process the heuristic cues in the message. Individuals tend to invest more cognitive resources to systematically evaluating the statements and arguments in the messages.^26,48,49 In this regard, people will consider more about how to go to the hospitals or medical centers for the HPV vaccination behind the efficacy information presented on the messages. Moreover, individuals’ cognitive thinking could be motivated to understand the benefits of the HPV vaccination clearly and to eliminate psychological barriers; thus finally enhancing individuals’ behavioral intention to receive the vaccination. Therefore, with a low number of likes, messages with both threat and efficacy components are more effective in promoting the HPV vaccination.

Conclusion

The contributions of our study are twofold. From the theoretical perspective, this study advances the EPPM with social cues by exploring fear appeal messages in social media health campaigns and investigating the potential effectiveness of social cues on health message acceptance. The findings reveal a significant main effect of threat and three-way interaction among threat, efficacy, and social cues. This research provides a new perspective for health campaigns on social media, which motivates scholars to broaden the field’s understanding of the interactive effects of message content and social cues on social media.

In terms of practical implications, the findings of the current study provide useful suggestions for vaccination promotion campaigns on Chinese social media. Specifically, it suggests that fear appeal messages could be effective in promoting HPV vaccination among Chinese women. However, in order to avoid only using heuristic information processing to assess messages, the side effects of the number of likes should be recognized when conducting fear appeal campaigns on social media. In addition, for vaccination promotion campaigns, the efficacy information should be more specific and audience-centered.

Nevertheless, the study is not without limitations. First, in regard to stimuli, the efficacy message in the experiment is inadequate. The efficacy on the stimulus picture only provided information about vaccination appointments, but lacked specific steps to instruct and guide individuals on how to carry out the HPV vaccination. Future studies might conduct a focus group to gain a deeper understanding of general efficacy perception and to design a message with sufficient efficacy. Second, this study merely focused on the effectiveness of fear-appeal messages. However, message framing and message structure that might also promote HPV vaccinations have never been examined. Future studies should use a different persuasive health message framework to develop vaccination promotion campaigns. Third, the current study only investigated a single type of social cues – likes. However, social cues are not presented separately on social media platforms. The experimental settings of social cues should integrate multiple social cues, such as likes and reposts in future studies. The effects of different social cues on fear appeal messages should be further examined as well. Lastly, the current research measured behavioral intentions for HPV vaccinations. While some scholars have indicated a high correlation between, and transformation from, intentions to behaviors,⁵⁰ intentions do not indicate HPV vaccination behavior. Future research should focus on actual behavioral outcomes. In total, further investigation is needed to address the above limitations in different contexts and cultures.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental data

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/21645515.2021.1975449.

Additional information

Funding

This work was supported by the National Social Science Fund Youth Project [19CXW033].

Appendix:

Eight Stimuli Posters