People Wear Masks When They Ski: Comparing Congruent and Incongruent Behavioral Context Appeals

Abstract

When advocating for a behavior, persuasive messaging typically focuses on the context that behavior is performed in, such as mask-wearing during the COVID-19 pandemic. However, situating the advocated behavior in a different context, termed an incongruent context appeal, may persuade by increasing attention, novelty, and memorability. The current study tested this supposition in a message experiment. Participants (N = 324) were randomized to view an incongruent context (e.g., skiing) or a congruent context (i.e., COVID-19) appeal advocating for mask wearing. The incongruence appeal had a direct, positive effect on mask wearing intentions and indirect, positive effects via two serial mediation pathways: time spent with the message increased attention through novelty and memorability. Findings suggest that an incongruent context appeal is an effective strategy for persuading audiences in information-saturated environments like the COVID-19 pandemic.

Persuasive messaging typically advocates for behavior in the context it is to be performed. For example, a communicator advocates for wearing a mask during a pandemic by focusing on pandemic-prevention mask wearing (Epton et al., 2022; Wang et al., 2021). The logic behind this approach is that the appeal (wear a mask during a pandemic) is consistent with the desired behavior (wear a mask during a pandemic). We label this a congruent context appeal. In congruent context behavioral advocacy, the message can simply convey that a behavior is desired (e.g., “wear a mask” with a picture of someone wearing one) or advocate for the benefits and barriers of performing the behavior in the context (e.g., wearing a mask reduces the spread of COVID-19) (Zanjani et al., 2011).

An alternative approach is what we label an incongruent context appeal. In this strategy, communicators advocate for a particular behavior by discussing it in an alternative context. For example, attempting to increase mask wearing during the pandemic by arguing that people wear masks in other contexts such as skiing, hunting, or snowmobiling. During the COVID-19 pandemic, public health communicators experimented with incongruent context behavioral advocacy given the saturation of the communication environment and resistance to advocacy within certain groups. As an illustration, the Department of Public Health in Montana developed an incongruent context appeal that advocated for pandemic mask wearing by noting that Montanans wear masks in numerous other situations (e.g., skiing; Depping, 2020).

Scholars have long been interested in the concepts of congruence and incongruence. Since Osgood and Tannenbaum (1955) proposed the principle of congruence, many studies have been done to refine and enrich underlying theoretical frameworks (e.g., Festinger, 1964; Heckler & Childers, 1992). The congruence effect has emerged as a focus for scholars because of its persuasive power. Previous research has shown that a congruent message can generate more favorable attitudes, improve attention to the message and message memory, and increase behavior intention (Chang, 2005, 2009; Choi & Rifon, 2012; Godinho et al., 2017; Segev et al., 2014; Updegraff et al., 2007).

However, Yang and Nan (2020) argue that the congruence effect in health messages is not consistently supported. Some studies have found that incongruent messages perform better than congruent messages (Lee & Jeong, 2014; Uskul & Oyserman, 2010; van Kesteren et al., 2012). The persuasive power of incongruence has been explained from the perspective of attention (Meyers-Levy & Tybout, 1989), critical thinking (Heckler & Childers, 1992b), novelty (Yoon, 2013), and memorability (Lange & Dahlén, 2003). For example, thematic incongruence can lead consumers to engage in more extensive processing to resolve the incongruence (Dahlen et al., 2008).

Given the inconsistent persuasive effect of (in)congruence appeals, scholars have argued for additional research investigating endogenous and exogenous variables that affect (in)congruence effects (Kononova & Yuan, 2015; Yoon, 2013). There is a pressing need for studies that explicate the contexts under which congruence or incongruence is more effective and the mechanisms through which they move audiences. Moreover, previous research about context congruence has primarily focused on advertising and marketing. Therefore, it is valuable to test (in)congruence effect in health communication settings (Segev et al., 2014; Yang et al., 2021). The current study engages these questions by theorizing the relationship between congruent/incongruent context and mask-wearing intention and testing four potential mechanisms of effect: time spent on message, critical reflection, perceived novelty, and perceived memorability.

Congruence & Communication

Defining Congruence

In 1955, Osgood and Tannenbaum proposed that the principle of congruence is that “changes in evaluation are always in the direction of increased congruity with the existing frame of reference” (P. 43). Congruity theory predicts that when individuals receive a positive or negative message about an object, their attitude toward that object changes depending in part on their perception of the particular object (Anderson & Fishbein, 1965; Osgood & Tannenbaum, 1955). Later research demonstrated that individuals are psychologically reluctant to change their attitudes and beliefs to align with message positions that are incongruent with their existing attitudes and beliefs (Festinger, 1964; Lee & Jeong, 2014).

Heckler and Childers (1992) argued that relevancy and expectancy are two dimensions of congruence. Relevancy refers to the material’s closeness to the theme's meaning and reflects how information present in the stimulus aids or impedes the clear identification of the topic or primary message being delivered. Expectancy emphasizes the degree to which an item or piece of information fits into a preset pattern or structure generated by the theme (Heckler & Childers, 1992). For example, when congruence theory is applied to the virtual environment, individuals will have a certain degree of expectancy to the physical products based on their website experiences since they perceived the relevancy between online introduction and physical products (Lee & Jeong, 2014).

Based on processing fluencing theories, de Droog et al. (2012) identified two types of congruence: conceptual congruence and perceptual congruence. These theories assume that information may be processed fluently through conceptual fluency and perceptual fluency. Conceptual fluency means the ease with which a concept (e.g., an argument, idea, or script) is recognized; perceptual fluency refers to the physical features of a stimulus (e.g., shape and color) which could improve information processing (de Droog et al., 2012; Lee & Labroo, 2004).

Previous research has explicated congruence in multiple ways. Researchers have examined the congruence between attitude and online reviews (Walther et al., 2012); character and product (de Droog et al., 2012, 2014); picture and text (Howlett et al., 2017; Shi et al., 2017); emotional state and emotion evoked by message (Na et al., 2018); and brand and influencer (Fleck et al., 2012; Kim & Kim, 2021; Olson & Thjomoe, 2011). Context refers to “the parts of a discourse that surround a word or passage and can throw light on its meaning” (Merriam-Webster, 2023). Thus, context congruence is the degree to which surrounding words or images share an identical theme with editorial content or persuasive advocacy (Zanjani et al., 2011). Shifts in context can significantly alter perception; for example, different reactions are elicited when the same message is delivered to the same audience in diverse media situations (Segev et al., 2014). Conversely, in the same context, audiences’ perceptions may differ whether the message content is congruent with the context. In the current study, we examine the congruence between message content and behavior advocacy context.

Congruence Effects

Chang (2002) conceptualized congruence effects as a stronger and more reliable source of persuasive effects. Multiple studies supported the persuasive power of congruence, especially in the advertising area. For example, when a consumer encounters an ad on a social media feed, the congruence between ad content and feed theme lead to higher involvement and a more positive attitude (Huang & Yoon, 2022; Moorman et al., 2013). An eye-tracking experiment confirmed that congruity between context and content could positively impact consumers' attention and memorability of online product recommendations (Luan et al., 2018).

Multiple factors can influence the effect. Strong congruent messages can trigger more favorable attitudes than weak messages (Updegraff et al., 2007). The effect also relates to issue involvement. When a message is shown in a congruent context, individuals who are highly involved with the issue will have a more favorable attitude toward the advertisement and show greater purchase intentions than individuals who are less involved with the issue (Segev et al., 2014).

Scholars are working to identify the underlying mechanism of congruence effects. Sherman et al. (2006) found that a congruent message leads to a stronger intention to perform behaviors and, subsequently, increased behavior change. An empirical study found that the relationship between ad-content congruence and ad effectiveness was mediated by advertising value among individuals with high skepticism levels (Yang et al., 2021). Later studies demonstrated that strategically matching culture to message frame has no effect on time spent on the message (Hervet et al., 2011), but can improve message memory for health recommendations and adoption of health behavior (Brick et al., 2016; Moorman et al., 2015). Kononova and Yuan’s study (2015) found that congruent messages are more memorable and can influence audiences’ attitudes; accordingly, the authors call for additional studies about potential factors that influence congruence effects.

Incongruence Effects

Despite previous studies showing the effect of congruence, scholars continue to debate the effectiveness of congruence and conduct research comparing congruent and incongruent messages. In 1982, Mandler developed incongruity theory which postulated that when information is not congruent with the semantic network, the information could be received with a more favorable attitude. Several studies confirmed the existence of an incongruence effect and proved its ability to improve behavior change intention (Goodstein, 1993; Heckler & Childers, 1992; Lee & Labroo, 2004). For example, conventionally, scholars believe that placing a cosmetic advertisement in a fashion magazine can improve the effectiveness of the ad since the content and context are congruent (e.g., Moorman et al., 2015). However, Dahlén et al. (2008) found that participants generated a higher ad attitude toward the same cosmetic ad when it was presented in an outdoor sports magazine than a fashion magazine. In the same vein, incongruence can lead to an increase in attention, more sophisticated brand association processing, greater ad and brand memory, and improved ad attitudes (Torn & Dahlén, 2017).

One explanation of the incongruence effect is that incongruent messages draw greater attention and engagement and evoke increased reflective thinking. Empirical studies have demonstrated that incongruity draws greater attention than congruity and has the potential for a positive evaluation (Lee & Schumann, 2004; Meyers-Levy & Tybout, 1989). Heckler and Childers (1992) conducted an experiment and discovered when information is incongruent with prior expectations or schemata, individuals will engage in more effortful or elaborative processing. Similarly, ads placed in a thematically incongruent magazine were found to evoke more thoughts and matching processes compared to ads in thematically congruent magazines (Dahlén et al., 2008).

Memorability is another possible explanation for incongruence effects, though past research has yielded conflicting findings. Some studies have demonstrated that information congruent with existing knowledge is better remembered than less congruent information (Anderson, 1981). However, other studies have found that incongruent information is better remembered (e.g., van Kesteren et al., 2012). Incongruent information could increase memorability because moderately discrepant pieces of information result in a more elaborate network of associations between stimulus and memory information (Halkias & Kokkinaki, 2014; Heckler & Childers, 1992). Perceived memorability of the message could also be impacted by incongruence. Lange and Dahlén (2003) measured the relation between perceived memorability and incongruent ads. Results showed that ad-brand incongruency enhanced brand memorability and brand attitude.

Incongruent messages often look different than other message strategies which could enhance perceived novelty. Novelty is an element of incongruence (Hirschman, 1980; Mandler, 1982). One aspect of novelty seeking is pursuing information potentially ‘incongruent’ with an existing belief (Lee & Schumann, 2004), which explains why sometimes individuals read stimuli that are incongruent with their long-term disposition. Researchers have found support for this idea, however, further studies are needed to understand the perception and effect of incongruence on novelty (Dahlen et al., 2008; Yoon, 2013).

Incongruency and Behavioral Context

The current study considers congruence in terms of behavioral context. From this standpoint, congruence is achieved when a message advocates for a particular behavior in the context of that behavior – what we label congruent context appeals. For example, advocating individuals wear a mask to prevent the spread of COVID-19 via messages that focus on pandemic-related imagery and benefits. Most persuasive messaging, and certainly public health messaging, is congruent in this sense and likely demonstrates, congruent context effects.

In terms of behavioral context, incongruent messaging occurs when the message context is not congruent with the behavioral context of the target behavior – what we label incongruent context appeals. For example, advocating individuals wear a mask to curb the spread of COVID-19 by arguing people wear masks in other contexts such as skiing, fishing, and snowmobiling. When incongruent context messages exert impact on intentions and behaviors, they demonstrate incongruent context effects.

Previous research has demonstrated mixed findings that incongruent messages could improve behavior change intention and attention to messages. However, Heckler and Childers (1992) identified a potential moderator that could explain past results: the presence of existing schema. They found that incongruent messages yielded greater attention for individuals with established cognitive schema. Based on schema theory, the term “existing schema” refers to a cognitive structure that represents organized knowledge about a particular concept or type of stimulus (Wicks, 1992). These schemas are derived from previous experiences and serve as frameworks for understanding and interpreting new information. Specifically, when new information becomes available, individuals attempt to assimilate it into the existing cognitive framework they have used in the past to interpret information about similar situations (Axelrod, 1973). Empirical studies have demonstrated that people tend to exhibit greater attentiveness and responsiveness to stimuli that are inconsistent with their existing schemas (Loftus & Mackworth, 1978; Warlaumont, 1997). COVID prevention provides a relevant example, as individuals were mandated to perform certain behaviors (e.g., mask-wearing, quarantine) despite resistance to the message and behavior. Consequently, when individuals are presented with a new message related to the same behavior, they may react based on their pre-existing schema towards the behavior. This resistance can lead to a negative perception of congruent context, as it has already been processed and deemed negative. In contrast, incongruent context could increase the time spent with the message, as it is not automatically processed by the existing schema.

Generally, attention to messages is positively related to the time spent with information (Avery & Park, 2018; Brown & Richardson, 2012). Followed previous study (e.g., Dahlén et al., 2008; Moore et al., 2005), the current study use time spent on message as a proxy for attention. Consequently, in line with Heckler and Childers (1992), we hypothesize that incongruent context appeals will increase intentions and time spent with the message because, in the current study, participants should have existing schema for COVID prevention.

• H1: In the situations where people have existing schemas, an incongruent context appeal will increase mask wearing intentions.

• H2: In the situations where people have existing schemas, an incongruent context appeal will increase time spent with the message.

We theorize that congruent/incongruent context effects are similar to previous context effects in that communicators likely choose a message approach based on the perceived cognitions, needs, and prior behaviors of the target audience. Incongruent context messages may be optimal in situations where the communicator believes that shifting the context could increase behavioral intentions by increasing time spent with the message. In these situations, an incongruent context triggers additional time with the message as audiences seek to integrate the new context with their existing schema.

In a sense, incongruent contexts may facilitate persuasion akin to narratives. Research on narrative persuasion has demonstrated that, compared to didactic and expository messages (Bruner; 1986; Newman, 1843), stories can increase immersion and facilitate shifts in perspective. Concerning the latter, narratives can increase time spent on message and perspective change even for individuals with pre-existing oppositional schema (Niederdeppe et al., 2011). Critical reflection, defined as the process of questioning one’s assumptions and beliefs, along with examining the basis or reasons behind them, leads to a transformation in beliefs, attitudes, or behavior (Gardner et al., 2006; Jung, 2012; Mezirow, 1998). Full engagement with a narrative message can temporarily suspend or reduce an individual’s ability to engage in critical reflection on their thoughts and emotions. However, after processing the message, critical reflection may increase due to the heightened attention given to it. Individuals who refrain from engaging in critical thinking become more receptive to exploring alternative viewpoints and perspectives (Lillie et al., 2021). Similarly, the increased time spent processing an incongruent context message may lead individuals to exhibit a greater tendency for critical reflection.

Novelty-seeking behavior often involves actively seeking out and paying more attention to information that may be incongruent with an individual’s existing beliefs or attitudes (Hirschman, 1980). Consequently, people may spend more time on a message and perceive it as more novel (Lee & Schumann, 2004). When individuals stay focused on a particular message instead of diverting their attention elsewhere, they tend to retain the information in their working memory for a longer period of time and engage in more elaborate or cognitively demanding processing (Heckler & Childers, 1992; Torn & Dahlén, 2017).

For both narrative messages and incongruent appeals, the audience has increased engagement with the message which influences the perception of the message, reflection on the content, and behavior changes. Based on past research, we theorize that messages with an incongruent context also increase time spent on message for individuals with existing schema which, in turn, increases critical reflection, perceived novelty, and perceived memorability.

Therefore, we made the following hypotheses:

• H3: In the situations where people have existing schemas, an incongruent context appeal will indirectly increase mask wearing intention by increasing time spent with the message which then increases critical reflection (H3a), perceived message novelty (H3b), and perceived message memorability (H3c).

Method

Design

Participants (U.S. adults 18 and older) were recruited national-wide by Qualtrics Panels as part of a larger study that surveyed U.S. adults every week for 36 consecutive weeks. Each week of the study included a unique embedded message experiment. The data for this manuscript were collected from July 3 through July 9, 2020. The sample was stratified by sex (male, female) and education (50% had a high school education or less). After completing a brief pretest, participants were randomly assigned to view an incongruent context poster or congruent context poster. Following exposure to a poster, participants completed a posttest, including measures of mask wearing intention, critical reflection, novelty, and memorability. The protocol was approved and monitored by an IRB. Data for this project is stored on a public repository and can be viewed and accessed here.

Participants

Participants (N = 324) ranged from 18 – 87 years of age (M = 39.52, SD = 17.12). Just over half of the participants identified as male (50.6%) and just over half had a high school degree or less (51.5%). In terms of race/ethnicity, 73.9% of the sample were White, 11.2% were Black/African American, 6.7% were Asian/Pacific Islander, 15.1% were Hispanic, 2.1% were Native American, and 6.1% self-described as other (participants could check more than one racial/ethnic category). For political parties, 48.1% of participants identified as Republicans and 51.9% are Democrats.

Power was calculated using G*Power version 3.1 (Faul et al., 2007; Faul et al., 2009). For a one-way ANCOVA (α = .05; power = .95), a sample size of 84 was identified to detect large effects (f = .40), a sample of 210 to detect medium effects (f = .25), and a sample of 1302 to detect small effects (f = .10). The obtained sample was 324, which provided strong power (.95) to detect effects as small as f = .20.

Stimuli

Participants were randomly assigned to receive one of two state-based poster campaigns (from Montana and Minnesota). Montana posters utilized incongruent context appeals and Minnesota posters utilized congruent context appeal. Each condition had two posters from the campaign. These posters are designed to promote mask-wearing behavior. Posters from the Montana depicted people wearing masks when they ski, hunt, or snowmobile with the slogan “Montanans wear face coverings all the time”. The key visual of the Minnesota posters is a man wearing a mask and the slogan “Mask up, Minnesota!” (see all stimuli materials in Appendix A).

Measures

Time spent on message

Qualtrics survey system automatically recorded the time between participants opening the stimuli and leaving the page. The time was measured in seconds (M = 18.84, SD = 19.43).

Mask wearing intention

Mask wearing intention was measured with a single item: “how likely it is that you will wear a mask in public to slow the spread of coronavirus?”. The item was assessed on a 7-point scale ranging from extremely unlikely (1) to extremely likely (7) (M = 6.06, SD = 1.68).

Critical Reflection

Critical reflection was measured based on Kember’s (2000) reflective thinking scale. The reflective thinking scale consists of four items assessed via a five-point scale ranging from strongly disagree (1) to strongly agree (5) (M = 3.20, SD = 1.23, α = .90). Items include “The message made me think carefully about things I believed.”, The message challenged ideas I had about the world”, “The message encouraged me to reflect on my beliefs”, and “The message helped me to think carefully about complicated ideas”.

Perceived Novelty

Perceived novelty was measured using four items scored on a seven-point semantic differential scale (Authors, under review). Novelty was measured with the semantic differentials old-new, tiresome-stimulating, monotonous-exciting, and something I’ve encountered too many times-something I still look forward to encountering on a scale from 1 to 7, with higher scored indicating greater novelty (M = 4.29, SD = 1.64, α = .88).

Perceived Memorability

Four items graded on a seven-point semantic differential scale were used to measure perceived memorability. (Authors, under review). Memorability was measured with the semantic differentials forgettable-unforgettable, will not stick in my head-will stick in my head, is not memorable-is memorable, and I won’t think about it again-I will think about it again (M = 4.48, SD = 1.87, α = .94).

Confirmatory Factor Analysis (CFA) was utilized to test a model with two latent variables, each with four items. The model was a good fit for the data, Satorra-Bentler (S-B) scaled χ² = 42.37, p = .35 (Satorra & Bentler, 2010), RMSEA = 04., 90% confidence interval: .02, .06, Model AIC = 512.59, CFI = .99, NNFI = .99, SRMR = .02.

Results

Incongruence Context Appeals versus Congruence Context Appeals

One-way MANCOVA was conducted to determine whether there is a difference between a congruent context appeal and an incongruent context appeal on mask-wearing intention and time spent on message, covarying pretest mask intentions. Results indicate that there was a significant difference in mask-wearing attention and time spent on the message based on context, Pillai’s trace = .17, F(2, 320) = 33.81, p < .001. Compared with the congruent context appeal, the incongruent context appeal significantly increased intentions to wear a mask and time spent with the message (support for H1 and H2, see Table 1).

Table 1.

Main Effects of Context on Outcomes

	Incongruence	Congruence	r	F
Posttest Mask Intentions	6.20 (1.52)	5.91 (1.82)	0.09	7.76**
Time Spent with Message	26.55 (23.20)	11.23 (10.19)	0.39	60.07***
Critical Reflection	3.29 (1.11)	3.12 (1.14)	0.08	2.34
Perceived Message Novelty	4.40 (1.65)	4.25 (1.62)	0.05	1.00
Perceived Message Memorability	4.44 (1.98)	4.52 (1.77)	−0.02	.09
N	161	163

Note. Means (standard deviations in parentheses).

^**p < .01

^***p < .001

Serial Mediation Models: Critical Reflection, Perceived Message Novelty, Perceived Message Memorability

SPSS Amos version 29 was used to test H3 via structural equation modeling (SEM). The study employed Anderson and Garbing’s (1988) two-step approach to SEM analysis. In the first step, a measurement model was tested with all latent variables, which were covaried with each other. In the second step, a structural model was tested with the expected associations between the variables. This approach helped identify whether issues with model fit were due to measurement problems, indicated by a poorly fitting measurement model, or associations not matching the data. Model fit was evaluated based on the comparative fit index (CFI) and the root mean square error of approximation (RMSEA). Acceptable model fit was indicated by CFI ≥ .90, RMSEA ≤ .08, and χ²/df ≤ 3.00, although CFI ≥ .95 and RMSEA ≤ .05 were preferred (Hu & Bentler, 1999; Schermelleh-Engel et al., 2003).

Before testing the model, the study variables were analyzed using bivariate correlations to examine their zero order relationships (Table 2). The measurement model achieved acceptable fit, χ²/df = 1.79, CFI = .98, RMSEA = .049. SEM identified a significant difference between incongruence context appeal and congruence context appeal, χ² = 147.530, df = 82, p < .001 (see Figure 1). Time spent on message was positively related to critical reflection, perceived message novelty, and perceived message memorability. However, only critical reflection presented a significant relationship with mask-wearing intention.

Table 2.

Correlation Matrix

	Variable	1	2	3	4	5	6
1	Incongruence context	–
2	Time spent on message	.40±	–
3	Critical reflection	.08	.13	–
4	Novelty	.04	.22±	.46±	–
5	Memorability	−.02	.18	.51±	.77±	–
6	Mask-wearing intention	.09	.09	.31±	.29±	.31±	–

^±p ≤ .001.

Figure 1.

Structural model.

Note. *p<.05. **p<.01. ***p<.001. Structural Model: x²/df=1.80, CFI=.98, RMSEA=.051. Direct effect =.080, 95%CI = −.022; .179, p = .127. Total indirect = .100, 95%CI = .034; .178, p = .003.

In addition to a combined model, and in line with Hayes (2018), we also tested via three serial mediation models. While the combined model tested for unique mediational variance, the serial mediation models examined whether each variable was a significant mediator, independent of the others. The measurement models achieved acceptable fit (Figures 2). The results revealed a significant indirect effect of impact of time spent on message and critical reflection on mask-wearing intention (β = 0.104, p < .05). For path through time spent on message and perceived novelty, the indirect effect was significant, β = 0.286, p < .001. The indirect effect of impact of time spent on message and perceived memorability on mask-wearing intention was positive and significant (β = 0.290, p < .01). Mediation analysis summary is presented in Table 3. Critical reflection, perceived message novelty, and perceived message memorability were positively related to time spent on message and mask-wearing intention. Therefore, H3 was supported.

Figure 2.

a. Measurement Model: x²/df=1.20, CFI=.997, RMSEA=.025; Structural Model: x²/df=1.053, CFI=.999, RMSEA=.013.

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

b. Measurement Model: x²/df=1.628, CFI=.991, RMSEA=.044; Structural Model: x²/df=1.503, CFI=.991, RMSEA=.40

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

c. Measurement Model: x²/df=2.10, CFI=.990, RMSEA=.058; Structural Model: x²/df=2.032, CFI=.989, RMSEA=.057.

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

Table 3.

Mediation analysis summary

Relationship	Direct Effect	Indirect Effect	Confidence Interval		p
			Lower Bound	Upper Bound
Incongruence appeal -> Time -> Critical reflection -> Intention	.220	.104	.029	.175	.015
Incongruence appeal -> Time -> Novelty-> Intention	.246	.286	.139	.437	.000
Incongruence appeal -> Time -> Memorability -> Intention	.101	.290	.124	.455	.001

Discussion

The current study compared the persuasive power of congruent context and incongruent context appeals. The incongruent context appeal resulted in greater mask wearing intention and more time spent with the message compared to the congruence context appeal. Time spent with the message, critical reflection, novelty, and memorability served as mechanisms of the incongruent appeal’s effect on intentions.

Mask messaging during the COVID-19 pandemic has overwhelmingly focused on congruent context appeals. Messages have advocated for mask wearing by discussing the benefits masking has on reducing the spread of COVID-19 (Gillman et al., 2022; Wang et al., 2021). In the current study, an incongruent context appeal was more effective at shaping intentions. This aligns with Mandler’s (1982) incongruency theory, proposing that incongruent messages are more effective than congruent ones. Previous research has demonstrated that incongruent messages motivate greater cognitive processing of the message and more time spent on the message (Dahlen et al., 2008; Lee & Schumann, 2004; Meyers-Levy & Tybout, 1989).

More time with the message influenced behavioral intentions through three key mechanisms: critical reflection, novelty, and memorability. The findings indicate that incongruent context appeals stimulate individuals to engage in more effortful message processing and reflect their own behavior critically, thereby facilitating changes in attitudes and behaviors, particularly when they possess pre-existing schemas.

A key finding is that incongruent context increased critical reflection which, in turn, increased behavioral outcomes. Communication researchers have devoted significant time to theorizing critical thinking as a construct, but the majority of that research has focused on critical thinking about a message rather than critical reflection. Critical reflection focuses on the cognitions of the individual rather than the content of the message. Given the results of the current study, future research should continue to explicate critical reflection as a mechanism of message effects. Importantly, communication researchers should identify message features that trigger critical reflection, especially for individuals with existing schema. In addition, communication scholars should fully explicate the theoretical process that unfolds from message exposure to critical reflection. The current study postulated that critical reflection begins after the message recipient is finished with the message, but that is a hypothesis in need of validation.

After exposure to an incongruent context appeal, the more time an individual spent on the message, the more novel and memorable they perceived it to be. In other words, an incongruent context appeal indirectly increased mask-wearing intention through a serial mediation pathway of time spent with the message and perceived message novelty or perceived message memorability. The result is consistent with van Kesteren et al.'s (2012) finding that an incongruent message has the ability to enhance message memorability. The findings also parallel Sheinin et al.'s (2011) argument that novelty influences receivers' attitude and intention.

Further, findings from the current study suggest that message designers may benefit from incongruent context appeals during crisis situations. Throughout the COVID-19 pandemic, concerns about information overload and message fatigue were common because of the volume of messaging about the pandemic (Gardikiotiset al., 2021; Jia et al., 2021). Audiences may ignore messaging when overloaded, and the influence of a single message, therefore, wanes (van Scoy et al., 2021). An incongruent context appeal can capture individuals’ attention in part because such appeals are atypical and novel. Further, incongruent context appeals may normalize a behavior by relating that behavior to other areas of individuals’ lives, aiding in information processing when overloaded.

Limitations

The current study provides several important contributions to understanding incongruence and persuasive messaging. However, it has limitations that should be noted. The stimuli were real-world messages from COVID-19 prevention campaigns in Montana and Minnesota. On the one hand, this is a strength of the study because it provides greater external validity. On the other hand, using existing messaging limits experimental control. Though the current results are robust, it is important to note that the contrast was not a tightly controlled message experiment. Other features of the posters, like the social norm and geographical location, may influence the results. Future research should investigate this possibility by carrying out a tightly controlled experiment. Since it is a national survey, it is possible that people from Montana or Minnesota perceived the poster different from participants who located in other area. The current analysis found support for a sequential relations from incongruence to time followed by critical reflection, novelty, memorability. However, the design does not allow researchers to rule out the possibility that alternative sequential models are accurate as the variables in the posttest were all assessed at the same time.

The current study used self-report designs, and the amount of time spent on a message was used to measure attention. However, we acknowledge the limitations of this approach and suggest that future research should explore other methodological approaches, such as using eye-tracking, to better quantify attention. Additionally, mask wearing intention served as the primary outcome of the current study. Intentions are an important indicator of future behavior, but do not always lead to behavior. It will be important for future research to assess if incongruence impacts actual behavior.

Conclusion

Recent studies argue that the effect of congruence is not consistent, and that incongruent messages could be more effective than congruent messages in certain conditions (Dahlen et al., 2008; Yang & Nan, 2020). The current study theorized congruence/incongruence in relation to behavioral context. The results provide evidence for an incongruent context effect. We tested four possible mediators, and the findings supported three. Yet, there is still more work to be done on this important topic. Additional research is needed to fully explicate and theorize situations where incongruent context messages are more effective.