Improving rural White men’s attitudes toward clinical trial messaging and participation: effects of framing, exemplars and trust

Abstract

This study examined whether framing, exemplar presence and exemplar race in social media news posts influence rural White men’s perceptions, attitudes and behavioral intentions toward clinical trial participation, and if so, how individual trust in doctors moderates the effects of these three factors. An experiment with a 2 within (framing: cognitive versus psychological barriers) × 2 within (exemplar: present versus absent) × 2 between (exemplar race: White versus Black) subjects mixed factorial design was conducted among rural White men (N = 208). Twelve social media news posts about clinical trial participation were created for the experiment. Results revealed that respondents had greater behavioral intentions to participate in clinical trials after seeing posts with exemplars present (versus absent). When news posts addressed cognitive barriers (e.g. lacking knowledge about the value of clinical trials), the presence of exemplars enhanced perceived self-efficacy to participate in clinical trials. Participants with lower trust in doctors reported more favorable attitudes to posts with exemplars, and the posts with Black exemplars were perceived as more effective (approaching statistical significance). When communicating about clinical trials to rural White men, health professionals should consider including exemplars and addressing cognitive barriers to participation.

Introduction

Recruiting sizeable and representative participants to clinical trials (CTs) is vital for yielding valid and generalizable outcomes [1]. Despite efforts to promote participation and gender equality in CTs, participation has been reported especially low among rural communities [1–3] and underrepresented for both males and females in different disease categories [4]. Some disease categories with underrepresented male participants in CTs for conditions like obesity, arthritis, mental illness and pulmonary disease [4, 5] belong to the top health concerns of overall rural residents [6–9]. At the intersection of the low participation of rural residents and underrepresented males in CTs, rural men suffer alarming health disparities [10]. For instance, rural men have a lower life expectancy than urban men, rural women and urban women [11]. The underrepresentation of rural men in CTs of the aforementioned disease categories may contribute to their higher excess mortality. Thus, encouraging rural men to participate in CTs can contribute to reducing health disparities.

Considering that rural communities are predominantly White [12], this study exclusively focuses on the rural White male sample. The Pew Research Center’s 2021 Core Trends Survey [13] showed that social media are used by 52% of rural men, most of whom are White (89%). Rural residents have less knowledge of CTs and greater misperceptions about them [1, 14]. Thus, mediated communication can be a feasible way to improve knowledge of CTs [15], and social media platforms are particularly influential in promoting CT engagement [16–18]. Among prior studies done to improve CT recruitment using mediated communication strategies [19–21], none has targeted rural White men.

Among the barriers to CT participation, including barriers that are cognitive (e.g. a lack of knowledge), psychological (e.g. distrust and fear of medical researchers), financial (e.g. costs incurred in participation), logistical (e.g. transportation problems) and structural (e.g. limited accessibility) [22], cognitive and psychological barriers are the biggest obstacles [23]. Studies on rural South Carolinians found cognitive barriers to be of primary concern [24, 25]. Thus, the current article focuses on addressing cognitive and psychological barriers to CT participation.

We use social media news posts as the medium for educating rural White men to help them overcome barriers to CT participation, which can affect their future actual participation behavior. This study examines the influence of framing and exemplification on four outcome variables. Two outcome variables are directly related to the message: perceived message effectiveness (PME) and message attitudes. PME can predict actual message effectiveness, attitudes toward the behavior recommended in the messages, behavioral intention and behavioral change [26, 27]. Moreover, one’s attitude toward a message can impact the attitude toward the health behavior promoted in the message [28], and it thus has been adopted to evaluate health campaign messages in previous studies (e.g. anti-binge drinking [29] and healthy eating [30]).

Another two outcome variables are related to CT participation: self-efficacy and behavior intention, both of which have been shown to be strong predictors of actual behavior according to the health belief model (HBM) and the theory of reasoned action (TRA) [31, 32]. The HBM has been widely used to explain and predict preventive health behaviors [33] and has also been applied to encourage prosocial health behavior (e.g. breast tissue donation [34] and CT participation [35]). The HBM highlights cues to action to adopt the prescribed health behavior [31]. Cues to action are defined as ‘strategies to create readiness’ to engage in prescribed health behavior [33], and they can be internal (e.g. chest pain) or external (e.g. mass media communication) [31]. Previous studies found that cues to action (e.g. behavior recommendations in health news) can increase individual self-efficacy and behavioral intentions [36]. According to the HBM, self-efficacy predicts whether one performs the recommended behavior [31] and the TRA posits that behavior intention determines behavior [32]. The current study uses social media messages addressing different types of barriers to CT participation as cues to action and examines how those messages influence self-efficacy and intention to participate in CTs.

Framing

Framing, which involves intentionally including or excluding certain elements of a message (e.g. certain phrases, sources or images) [37], is often used to draw attention to particular aspects of a news story and make it easier for readers to process and understand complex issues [38–40]. The framing influences what the audience remembers about the message, which can then affect their attitudes and behaviors [41]. Thus, CT news stories that highlight a particular type of barrier may lead the reader to focus on that particular barrier, which can then influence how readers think about CT participation. More specifically, when the news frame offers solutions to common barriers, readers may perceive the barrier as less of an obstacle, increasing self-efficacy and the likelihood of future CT participation. By employing the framing strategy, the news stories in the present study (framed to focus on either psychological or cognitive barriers) should improve message effectiveness and message attitude (by enhancing the message processing [40]). As self-efficacy and attitude influence behavioral intention [32], the frame may also affect readers’ intentions to participate in a future trial.

It is expected that cognitive-barrier-framed news can help readers better understand the purpose and process of participation by focusing on the value and importance of CTs. Prior research has found that rural citizens have less factual knowledge about CTs than urban citizens [24], and a lack of knowledge can impede rural populations’ interest in CT participation [42, 43]. On the other hand, the psychological-barrier-framed messages can help readers feel safer about CTs by focusing on the measures that protect participants. This may also be influential, given that distrust and fear have also been found to obstruct rural participation [2, 44]. Therefore, this study first asks:

Main effect of framing (Research Question 1): How does framing (focused on psychological barriers versus cognitive barriers) influence (a) message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

Exemplar presence

Exemplars are single cases that serve as typical examples of broader concepts [45]. In the news, exemplars are frequently used to humanize larger issues. Because of the nature of news coverage, exemplars in news are often found in the form of quotations from laypeople [46]. As exemplars aid in telling stories and illustrating larger issues, they also can affect individuals’ beliefs about the topic at hand, which can influence related attitudes and behaviors. This is explained by exemplification theory [45, 47] and has been illustrated in many studies [48].

This study argues that the presence of exemplars, in the form of the testimony of a former CT participant, can be a strategic tool for overcoming the barriers of interest (cognitive and psychological) and improve message evaluation, attitude, self-efficacy and behavior intention. The experiences of previous CT participants have been said to affect prospective participants’ understanding of CTs [49], and exemplification in the form of a peer (versus researcher) source has been found to improve PME and favorable message attitude in the context of CT recruitment advertising [21]. Therefore, this study hypothesizes:

Main effect of exemplar presence (Hypothesis 1): The presence (versus absence) of an exemplar will lead to greater (a) perceived message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention.

In addition, this study is interested in examining the combined effects of exemplar presence and the different frame types. Both message features (framing and exemplars) could have significant effects on the outcomes of interest, and the two features could interact with one another to affect the outcomes. To explore this, the next research question asks:

Interaction between framing and exemplar presence (Research Question 2): How do framing and exemplar presence jointly influence (a) message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

Exemplar race

The race of the exemplars is also an important consideration as race influences how much an individual identifies with another person [50]. When individuals identify with whom they see in a message, this can affect their message reception [51, 52]. Serving as a heuristic cue, the race variable can help build trust and message compliance [45]. For instance, anti-human immunodeficiency virus messaging featuring Black speakers (versus White speakers) has been found to be more effective for Black audiences [53], and White audiences have been found more likely to purchase products when the advertisements feature White rather than Black actors [54]. Based on these findings, same-race exemplars may have more profound effects. That said, research has suggested that exemplar race is more salient to members of racial and ethnic minorities, who are more mindful of racial depictions (thereby leading to greater effects) [55]. Exemplar race may not be as influential for the White men in rural areas as they are not racial minorities. To explore the effects of exemplar race, this study asks:

Main effect of exemplar race (Research Question 3): How does exemplar race influence (a) perceived message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

Moreover, to explore how the race of the exemplars interacts with the presence or absence of an exemplar, we ask:

Interaction between exemplar presence and exemplar race (Research Question 4): How do exemplar presence and exemplar race jointly influence (a) message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

Trust in doctors

Trust in doctors is defined as the degree to which the patient has confidence in the doctor’s integrity, ability and judgment [56]. The overall level of trust in doctors results from both interpersonal trust (i.e. individual experiences with doctors) and system trust (i.e. attitudes toward health care system) [57]. Research has demonstrated that patients with greater trust in doctors can experience greater satisfaction with the health care they receive [58], greater self-efficacy in the treatment regimen [59] and greater willingness to participate in CTs [22, 60]. This may occur due to the heuristic processing [61]. That is, trust serves as a shortcut to shape people’s attitudes and decision-making, particularly in the areas where people are less knowledgeable [61]. Because rural White men are generally less knowledgeable about CTs, it is reasonable to assume that their overall trust in doctors may moderate the effect of news framing and exemplars on CT participation. To explore this, the following three research questions ask:

Moderation effect of trust in doctors (Research Question 5–7): How does trust in doctors moderate the influence of framing (RQ5)/exemplar presence (RQ6)/exemplar race (RQ7) on (a) message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

Method

Design and stimuli

The purpose of this study is to examine whether particular message features in social media news posts (i.e. framing, exemplar presence and exemplar race) influence communication outcomes among rural White men (i.e. PME, message attitudes, self-efficacy and behavioral intentions toward CT participation), and if so, how individual trust in doctors moderates the effects of these three factors. With this purpose, this study employed an online experiment with a 2 (framing: cognitive versus psychological) × 2 (exemplar: present versus absent) × 2 (exemplar race: White versus Black) mixed factorial design. Framing and exemplar presence were within-subject factors because within-subject designs (I) can reduce the random noise from respondents’ individual differences and (II) are thus particularly effective in detecting subtle (yet significant) effects that are less likely to transfer across conditions [62]. Exemplar race was a between-subject factor because race’s effect may be too strong and carried over to the other conditions. Because exemplar race was manipulated only under the exemplar-present condition, a total of six conditions were formed (i.e. cognitive/White exemplar present, cognitive/Black exemplar present, cognitive/exemplar absent, psychological/White exemplar present, psychological/Black exemplar present and psychological/exemplar absent). For each condition, we prepared two versions of messages accounting for message variance [63], resulting in 12 experimental stimuli. See Appendix A for the illustration of the experiment flow.

The stimuli were mock Facebook news posts of the Associated Press about generic CT participation, including a headline, link, image and brief textual description. The text included two paragraphs. The first paragraph stated how researchers attempt to promote CT participation by addressing the barriers people may have (i.e. psychological or cognitive). The second paragraph expanded on these points with specific examples. All other message features were kept similar across the four message conditions, including the post design, number of ‘likes’ and comments, text length (ranging from 107 to 129 words) and readability (9/10 grade level). See Appendix B for examples. These were edited by a professional news editor.

Respondents and procedure

We contracted with Qualtrics to send survey invitations to their existing US panelists, who were rural White men, 18–65 years old and English-speaking. The panelists’ personal information was validated by the Qualtrics partners. A total of 208 rural White males completed the online experiment. In addition to using Qualtrics’ targeted panel, we included screeners in our questionnaire, asking respondents to report age, gender, race and the type of region they live in. Through the targeting and screening, we identified a quality sample for the current study. During the data collection process, low quality responses (e.g. speeders, straightliners and responses with missing data) were removed and replaced by the Qualtrics team. Thus, no respondents were deleted during the data analysis. After giving informed consent and passing the screeners, respondents answered questions about their trust in doctors. They were then randomly assigned to either the White or Black exemplar condition. In each condition, respondents saw a total of four messages—two focusing on psychological barriers and two focusing on cognitive barriers. For each framing level, one message included an exemplar, and one did not. The presentation order of the four messages was fully randomized. After each of the four messages, respondents responded to the dependent variable measures in sequence of PME, attitude toward the message, perceived self-efficacy and behavioral intention. Lastly, respondents answered demographic questions.

Respondent ages ranged from 18 to 65 (M = 50.73, SD = 12.94). Over half of the respondents completed high school or some college (n = 116, 55.7%), and some had bachelor’s or graduate degrees (n = 87, 41.9%). From liberal (1) to conservative (7), the sample leaned conservative (M = 4.57, SD = 1.77).

Independent variables

Framing: psychological versus cognitive barriers

Framing referred to the type of barrier information made salient in the post, including two levels, psychological barriers and cognitive barriers. Psychological-barrier-framed posts focused on addressing psychological barriers to CT participation: fear about participation and a lack of trust in doctors. Cognitive-barrier-framed posts focused on overcoming cognitive barriers to CT participation: a deficit of knowledge about how CTs work (e.g. what participation is like) and the importance and benefits of CT participation. These frames were created based on the wording suggestions from Clark et al. [64]. See Table I for stimuli wording. A manipulation check was performed to ensure the two levels of frame focus were distinguished in the news posts. After each post, respondents were asked to indicate what they believed the post emphasized (1 = well-being and safety, 7 = benefits and importance of participation). A paired-samples t-test showed a significant difference between the cognitive-barrier-framed posts (M = 5.14, SD = 1.32) and the psychological-barrier-framed posts (M = 2.81, SD = 1.72), t(207) = −16.62, P < 0.001.

Table I.

Core message manipulations and all measures with associated items

	Exemplar presence	Exemplar absence
Psychological-barrier-frame	‘The researchers really assured me that my health and safety were important to them. I never felt like I was being used. Instead, the researchers seemed to care about my health and well-being. They closely watched me the whole time and made sure to watch out for any possible side effects during the study,’ said Dustin Davenport, a clinical trial participant. ‘They also made sure I knew that my participation was voluntary and that I was free to leave the study at any time, for any reason.’	One big step is letting participants know that their personal health and safety are important to the researchers. Instead of people feeling like they are being used, it is crucial that they know that the researchers care about their well-being. Researchers explain that they closely watch their participants and pay attention to any side effects they might have. Also, it is important that people who sign up know that it is on a volunteer basis. People are free to leave the study at any time, for any reason.
Cognitive-barrier-frame	‘I realized that being in a clinical trial gives me the chance to access some of the best doctors. During the study, they carefully watched my overall health on a regular basis,’ said Dylan Forster, a clinical trial participant. ‘I was able to receive the newest treatment for my chronic disease, maybe with more effectiveness and fewer side effects than what I was taking. I also got extra health care for my disease with no cost from the clinical trial staff.’	First, researchers are letting people know that clinical trial participants get to see some of the best doctors who will carefully watch their overall health during the study. Second, researchers also explain that participants might get cutting-edge treatments for their chronic diseases. These new treatments may work better with fewer side effects than treatments that are currently available. People who enroll in a clinical trial can also expect more care for their health problems from the medical staff, such as regular blood pressure or overall checkups. This extra care is provided at no cost to the participant.
Measures	Items
Trust in doctors	Doctors keep up-to-date on the most modern, current treatments available to protect their patients’ health.
	I trust doctors to protect my health.
	Doctors do not let their personal beliefs bias their decisions about patients’ personal health.
Message effectiveness	The content of this social media post was convincing.
	The social media post said something important to me.
	Reading this social media post helped me feel confident about participating in clinical trials.
	I agree with the information given in this social media post.
	The information in this social media post about clinical trials is believable to me.
Message attitude	The social media post I just saw was: Bad–Good, Unlikable–Likable, Unpleasant–Pleasant, Unattractive–Attractive, Worthless–Valuable, Unappealing–Appealing.
Self-efficacy	For me to participate in a clinical trial would be: Difficult–Easy.
	How certain are you that you could participate in a clinical trial? (Not at all certain–Very certain).
Behavioral intention	I plan on joining a clinical trial.
	I am willing to join a clinical trial.

Exemplar: present versus absent

Exemplar referred to an individual’s account of his CT experiences. When present, the exemplar was shown in the form of a quote in the second paragraph of the post and through an accompanying photo. When absent, the exemplar was replaced with an abstract informational message written in the third person perspective, and the image featured a generic research object, such as a test tube.

Exemplar race: White versus Black

Exemplar race was manipulated through the photo and the name of the man who was quoted in the exemplar-present condition. The photo featured a White or Black man, and their first and last names were chosen from a list of common White and Black male names, respectively.

Trust in doctors: higher versus lower

Trust in doctors assessed the level of people’s confidence and reliance on general doctors measured with a scale of three items from McComas et al. [60] (see Table I). The final score of trust in doctors was the average of the three items (M = 5.15, SD = 1.23, Cronbach’s α = 0.87). Respondents were split into two groups based on the mean score: higher trust in doctors (M = 6.03, SD = 0.52, n = 112) and lower trust in doctors (M = 4.12, SD = 1.01, n = 96).

Dependent variables

PME referred to the target audience’s ratings of the potential impact of persuasive messages [26]. In the present study, PME assessed respondents’ ratings of the potential persuasive effects of the news posts using a scale of five items adapted from Lee et al. [65] (Cronbach’s α across the conditions: 0.93–0.96).

Attitude toward the message assessed respondents’ positive or negative feelings toward the posts. The six-item attitude scale used in this study was adapted from Kang and Lee’s study [29] (Cronbach’s α across the conditions: 0.96–0.98).

Self-efficacy measured people’s belief about their capabilities to participate in CTs with a scale of two items adapted from Lee et al. [65] (Cronbach’s α across the conditions: 0.89–0.94).

Behavioral intention measured the likelihood that respondents would participate in CTs after exposure to the posts. We measured behavior intention using two items from Chen et al. [66] (Cronbach’s α across the conditions: 0.86–0.89).

All measures in this study were derived from previous studies. PME and message attitude were from the communication literature mentioned above. Self-efficacy was adopted from the HBM. Behavioral intention was derived from the TRA. For all measures, respondents rated items on seven-point scales, either Likert scales indicating their agreement (1 = strongly disagree to 7 = strongly agree) or semantic differential scales indicating their positions (e.g. 1 = bad to 7 = good). All items for these measures are also displayed in Table I.

Analysis

A 2 within (framing: cognitive versus psychological) × 2 within (exemplar: present versus absent) × 2 between (exemplar race: White versus Black) × 2 between (trust in doctors: higher versus lower) subjects repeated measures analysis of variance (ANOVA) was performed on each dependent variable (N = 4). The moderation effect of trust in doctors was assessed by checking whether an interaction term was significant (P < 0.05). Follow-up pairwise comparisons were adjusted using least squares difference. All analyses were conducted using SPSS 27 General Linear Models.

Results

The summary of the repeated measures ANOVAs for four outcome variables is presented in Table II.

Table II.

Summary of the results of repeated measures ANOVAs for four outcomes (N = 208)

	Message effectiveness			Message attitude			Self-efficacy			Behavioral intention
Predictors	F(1, 204)	P	ηp 2	F(1, 204)	P	ηp 2	F(1, 204)	P	ηp 2	F(1, 204)	P	ηp 2
Framing (psychological versus cognitive) (RQ1)	3.77	0.054	0.02	2.12	0.15	0.01	0.00	0.99	0.00	0.40	0.53	0.00
Exemplar (presence versus absence) (H1)	2.69	0.10	0.01	3.67	0.06	0.02	2.05	0.15	0.01	6.05	0.02	0.03
Framing × exemplar presence (RQ2)	0.18	0.68	0.00	0.32	0.58	0.00	5.22	0.02	0.03	0.95	0.33	0.01
Exemplar race (White versus Black) (RQ3)	0.68	0.41	0.00	0.09	0.77	0.00	1.01	0.32	0.01	1.31	0.25	0.01
Exemplar race ×exemplar presence (RQ4)	4.98	0.03	0.02	0.38	0.54	0.00	2.19	0.14	0.01	0.48	0.49	0.00
Trust in doctors (higher versus lower)	42.88	0.00	0.17	26.87	0.00	0.12	14.51	0.00	0.07	18.29	0.00	0.08
Trust in doctors ×framing (RQ5)	1.06	0.30	0.01	0.78	0.38	0.00	0.16	0.69	0.00	0.40	0.53	0.00
Trust in doctors ×exemplar presence (RQ6)	0.63	0.43	0.00	8.01	0.01	0.04	0.12	0.73	0.00	2.50	0.12	0.01
Trust in doctors ×exemplar race (RQ7)	4.06	0.045	0.02	2.11	0.15	0.01	0.05	0.83	0.00	0.07	0.79	0.00

Notes. Within-subject factors: framing (cognitive versus psychological) and exemplar (present versus absent). Between-subject factors: exemplar race (White versus Black) and trust in doctors (higher versus lower). Statistically significant point estimates (P < 0.05) are indicated with italic font. ‘×’ refers to the interaction of two variables. Pairwise comparisons were adjusted using least squares difference.

RQ1: How does framing (focused on psychological barriers versus cognitive barriers) influence (a) message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

H1: The presence (versus absence) of an exemplar will lead to greater (a) perceived message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention.

RQ2: How do framing and exemplar presence jointly influence (a) message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

RQ3: How does exemplar race influence (a) perceived message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

RQ4: How do exemplar presence and exemplar race jointly influence (a) message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

RQ5–7: How does trust in doctors moderate the influence of framing (RQ5)/exemplar presence (RQ6)/exemplar race (RQ7) on (a) message effectiveness, (b) message attitude, (c) self-efficacy and (d) behavioral intention?

The main effect of framing (RQ1) was not statistically significant on any of the four outcome variables.

The main effect of exemplar presence (H1) on CT participation intention was statistically significant. Specifically, respondents’ behavioral intention was greater when there was the presence of an exemplar (M = 4.15, SE = 0.11) rather than absence (M = 4.06, SE = 0.11), P = 0.02. However, exemplar presence did not significantly influence PME, message attitude or self-efficacy. Thus, H1d was supported; H1a–c was not.

The interaction between framing and exemplar presence (RQ2) was statistically significant on self-efficacy but not on PME, message attitude or behavioral intention. Further pairwise comparison showed that the presence of an exemplar (M = 4.48, SE = 0.12) was associated with higher self-efficacy than absence (M = 4.32, SE = 0.12) when the message focused on cognitive barriers (P = 0.01), but this effect was not significant when the message focused on psychological barriers (P = 0.49). Figure 1 shows the interaction effect.

Fig. 1.

Interaction between exemplar presence by framing on self-efficacy. The full scale on the y axis is 1–7. Error bars = 95% CI.

The main effect of exemplar race (RQ3) was not statistically significant on any of the four outcome variables.

The interaction between exemplar presence and exemplar race (RQ4) was statistically significant on PME but not on message attitude, self-efficacy or behavioral intention. Pairwise comparison showed that the absence of a White exemplar (M = 4.88, SE = 0.11) was associated with greater PME than presence (M = 4.74, SE = 0.11), P = 0.01, but this effect was not significant for the Black exemplar, P = 0.69. Figure 2 shows this interaction effect.

Fig. 2.

Interaction between exemplar presence by exemplar race on perceived message effectiveness. The full scale on the y axis is 1–7. Error bars = 95% CI.

Trust in doctors did not significantly moderate the influence of framing on the outcome variables (RQ5). Trust in doctors significantly moderated the influence of exemplar presence (RQ6) on message attitude but not on PME, self-efficacy or behavioral intention. Pairwise comparison showed that the presence of an exemplar was associated with more favorable attitudes (M = 4.76, SE = 0.13) than absence (M = 4.57, SE = 0.13) for the respondents with lower trust in doctors (P = 0.001) but not for those with higher trust in doctors (P = 0.50). Figure 3 shows this interaction effect.

Fig. 3.

Interaction between exemplar presence by trust in doctors on message attitude. The full scale on the y axis is 1–7. Error bars = 95% CI.

Trust in doctors significantly moderated the influence of exemplar race (RQ7) on PME but not on message attitude, self-efficacy or behavioral intention. Pairwise comparison showed that the respondents with lower trust in doctors reported greater PME when viewing a Black exemplar (M = 4.59, SE = 0.17) than a White exemplar (M = 4.15, SE = 0.15), P = 0.06. However, the respondents with higher trust in doctors reported greater PME when viewing a White exemplar (M = 5.47, SE = 0.14) than a Black exemplar (M = 5.29, SE = 0.15), P = 0.38. Although the differences were not statistically significant for either group, their patterns were opposite, indicating a significant interaction.Figure 4 shows this interaction effect.

Fig. 4.

Interaction between exemplar race by trust in doctors on perceived message effectiveness. The full scale on the y axis is 1–7. Error bars = 95% CI.

Discussion

This study examined how social media news framing and exemplars could be used to overcome barriers and promote CT participation among rural White men. Findings indicated that respondents had higher intentions to participate in CTs after seeing posts with exemplars (i.e. the testimonies of former CT participants) than posts without exemplars. When news posts addressed cognitive barriers (e.g. lacking knowledge about the value of CTs), the posts with exemplars were associated with greater self-efficacy to participate in CTs than the posts without exemplars. Generally, the absence of White exemplars was perceived as more effective compared with the presence of White exemplars. Respondents with lower trust in doctors reported more favorable attitudes to posts with exemplars and perceived the posts with Black exemplars as more effective than those with White exemplars. The findings of this study can add to the limited experimental literature on CT recruitment message designs and provide pragmatic implications for messaging strategies to reach rural White men.

This study demonstrated that news posts using an exemplary CT participant sharing personal experience that addresses barriers to CTs could boost viewers’ intention to participate in CTs. This result supports exemplification theory by confirming the effectiveness of using exemplars to communicate CT participation among rural White men. Previous studies found that men were more likely to respond to social media recruitment messages that included images of men [67], but using an image of a man did not improve actual participation rate [68]. The present finding can provide a possible explanation for their inconsistent results. That is, merely using an image of a man was perhaps insufficient to elicit behavioral intention and behavioral change even if it increases the viewers’ interest. Instead, the image of the man and his story should be combined to make a compelling exemplar impacting behavioral outcomes.

This study further showed that when viewing posts addressing cognitive barriers, the presence of an exemplar was associated with higher self-efficacy than absence. As predicted, when viewing an exemplar overcoming barriers to CTs, people can visualize how these barriers can be overcome and may feel more competent to do so. In addition to supporting the exemplification effect, this finding also furthers the understanding of cues to action, a less developed construct of the HBM [69, 70]. Our finding showed that social media news posts presenting exemplary CT participants overcoming cognitive barriers could work as a cue to action to increase rural men’s self-efficacy to participate in CTs.

When considering the moderation effect of trust in doctors, the presence of an exemplar generated more favorable message attitudes only for those with lower trust in doctors. A possible explanation would be that when people have lower trust in doctors, giving credit to the exemplar (i.e. previous CT participants) may assuage respondents’ individual bias against doctors because messages featuring a CT participant are more favorable and credible than those featuring a doctor [21]. Thus, viewing posts with and without exemplars can produce a larger difference. However, for those with higher trust in doctors, since they do not have much bias against doctors, the presence or absence of a participant exemplar was less likely to make a significant difference.

While exemplar race has been shown to have significant effects in some cases [53, 54], it was found not influential for the rural White men in this study. This result was in line with Appiah’s [55] finding about TV character race, in which White viewers show no preference for White characters. The author explained this finding using distinctiveness theory [73], which posits that an individual’s distinctive traits will be perceived as more salient to him- or herself than the prevalent traits possessed by the majority of others. Thus, race will be more salient to the members of racial minorities than to the racial majorities in their social environment, and White people comprise a majority in rural settings. This may explain why exemplar race was not a significant factor for rural White men in our study.

However, when considering the moderation effect of trust in doctors, posts with Black exemplars were perceived as more effective than posts with White exemplars for people with lower trust in doctors (approaching statistical significance). A similar finding has also been found in Lee et al.’s study [21]. This finding can possibly be explained by expectancy violations theory (EVT) [71]. EVT predicts that individuals who show characteristics that positively violate social norms or stereotype-based expectations are perceived as more likable than those who matched expectations. For example, research has shown successful Black individuals can be perceived as having more favorable personal qualities (e.g. intelligence) than equally successful White individuals because the success of Black people violates the expected norm that Black people are less likely to succeed (because of racial discrimination) [72]. In the current study, the appearance of Black exemplars might violate the expected norm that exemplars in rural areas would be White. This violation was considered positive because people generally like participant exemplars better than doctor exemplars [21], and the preference for participant exemplars may be particularly salient for those with lower trust in doctors. Consequently, posts with Black exemplars were perceived as more effective than those with White exemplars for rural White men with lower trust in doctors. However, people with higher trust in doctors might not have that strong preference for participant exemplars as people with lower trust in doctors did, and accordingly, the violation of expected exemplar race was less likely to create a significant difference for them.

Practical implications

This study provides several implications for CT recruitment. First, when communicating about CTs to rural White men, health professionals should consider including exemplars and addressing cognitive barriers to participation. Second, when the target audience has relatively low trust in doctors, we recommend that health professionals include Black CT participant stories as exemplars. However, if the target audience has average or higher trust in doctors, there is no need to think about exemplar race.

Limitations and future directions

The use of the experimental method raises a certain degree of artificiality in the study. Since respondents were paid to read the posts, reactions and attention levels may differ from when they view them outside the experimental setting. For example, the attention levels of those who have searched for CTs might be greater than those who never thought about CTs, and actual message effects may differ. Second, even though we adopted a within-subject design to minimize the random effects of individual differences, some contextual factors (e.g. no nearby CT opportunities) could be impossible to override with a brief message. Thus, non-communication strategies should also be developed to promote participation, such as providing travel costs to and from study sites. Additionally, this study did not include an actual control condition in which respondents were not exposed to any messages. Despite the limitations, this study provides theoretical contributions to health and communication theories (i.e. framing theory, exemplification theory and HBM). By applying and developing those theories, we propose effective practical communication strategies for recruiting rural White men. It is also worth mentioning that messaging strategies in this study (including a CT participant exemplar) may potentially be applied to other social media platforms than Facebook. Future research may also test the message effects across different formats, such as a poster in local coffee shops.

Appendix A. Experiment flow

Fig. 5.

 

Notes. Experimental design: 2 within (framing: cognitive versus psychological) × 2 within (exemplar: present versus absent) × 2 between (exemplar race: White versus Black). Black- versus grey-coded texts reflect the between-subject design of exemplar race (Black versus White). Conditions in the same color (all black or all grey) indicate the within-subject design. Screeners include questions about age, gender, race and region. Trust Doc refers to trust in doctors; R, random assignment; Ex Pr, exemplar presence; Ex Ab, exemplar absence; Psych BA, psychological barrier; Cog BA, cognitive barrier; MC, manipulation check; DVs, dependent variables (including message effectiveness, message attitude, self-efficacy and intention to participate in a clinical trial); DG, demographics and v1/v2, version 1 or version 2.

Appendix B. Samples of stimuli

A.1. Psychological barriers presented with Black exemplar

Fig. 6.

 

Image: Freepik.com

A.2. Cognitive barriers presented with White exemplar

Fig. 7.

 

Image: Freepik.com

A.3. Psychological barriers presented without an exemplar

Fig. 8.

 

Image: Freepik.com

Funding

The Washington University Institute of Clinical and Translational Sciences grant (UL1TR002345) from the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH). University of Missouri Research Council Grant (URC-21-030).

Conflict of interest statement

None declared.