Critical Considerations for Using Cultural Targeting and Tailoring in Health Communication Interventions

Abstract

One approach to addressing observed health disparities that is frequently discussed in the literature is adapting health messages to the cultural identities of groups who experience an undue burden of disease. The extant research on the cultural tailoring and targeting (CTT) of health messages generally indicates that such adaptations are effective. However, the empirical basis for this conclusion does not provide definitive evidence that CTT is always necessary nor demonstrate that culturally adapted messages are always more effective than more general message appeals. Despite extensive literature on CTT, important questions remain about the necessary components, methodology, and evaluation of CTT research. In this essay, we present a set of criteria for assessing the existing research base for CTT and ensuring that future CTT research is valid, replicable, rigorous, and robust. Key considerations include identifying when CTT is necessary, conducting rigorous formative research, using appropriate experimental designs, designing message interventions in a way that enables generalization, and evaluating mediators in order to build explanatory theories of CTT.

Introduction

Cultural tailoring and targeting (CTT) generally refers to the modification of intervention components to make them congruent with the characteristics, norms, values, preferences, and beliefs of target populations (Griffith et al., 2023; Joo & Liu, 2021). CTT has been an important area of health communication research for more than two decades (Kreuter et al., 2003; Resnicow et al., 1999), when investigators began to more systematically assess the potential for tailoring or targeting on cultural elements to increase the effectiveness of health communication interventions (Kreuter et al., 2005; Resnicow et al., 2005). Interest in CTT is particularly driven by its potential to increase equity and improve outcomes for underserved populations (Joo & Liu, 2021). However, the evidence supporting the effectiveness of cultural tailoring and targeting is inconsistent (Huang & Shen, 2016), with reviews finding evidence of effectiveness under some circumstances and for some health topics but not others (Nierkens et al., 2013). Additionally — despite the substantial volume of research on CTT that has accumulated to date — many gaps remain in understanding how CTT works and how to make CTT interventions more effective and efficient.

In this essay, we outline criteria for evaluating existing CTT research and provide recommendations to guide future CTT research. We begin by defining CTT and presenting a brief overview of the evidence base for CTT, before moving into a discussion of key considerations for research, including identifying when CTT is appropriate, approaches to formative research that ensure the basis for CTT has been empirically established, designing studies to enable strong inferences (particularly in regard to comparison group selection), increasing the generalizability of CTT research results so they can contribute to cumulative knowledge and be used to inform subsequent work, and attending to mechanisms of effect to increase understanding of how and why CTT works. For each consideration, illustrative examples are included to concretize and contextualize the principles discussed. Additionally, it should be noted that while many aspects of an intervention can be culturally tailored or targeted (e.g., use of culturally relevant meals in a weight loss program), this essay specifically focuses on CTT of health messaging and communication interventions.

What is Cultural Tailoring and Targeting?

CTT has three primary components: culture, tailoring, and targeting. Below, we operationalize each of these terms to anchor the subsequent discussion.

Message tailoring.

We follow the conceptualization of message tailoring from Rimer and Kreuter (2006, p. 184), which states:

Tailored health communication (THC) is any combination of information and behavior change strategies intended to reach one specific person based on information unique to that person, related to the outcome of interest, and derived from an individual assessment.

The emphasis in this definition is on messages being created uniquely for each person in the target population using information that is unique to that individual. The processes used to create such specific message-to-individual matching are often computational (Strecher, 1999), and generally require individuals to supply data regarding their socio-demographics, attitudes, beliefs, and other characteristics necessary to generate a profile that can guide selection of key communication components, including message content, context, message source, and the channels through which the message is delivered (Griffith et al., 2023). Message tailoring can be expensive and technically complex, but the investment in resources to create individualized messages can pay dividends in engagement, recall, and acceptance, thereby increasing likelihood of message effectiveness.

Message targeting.

Targeting refers to the process of modifying and delivering messages to a specified group of people with shared characteristics (Griffith et al., 2023). The targeted group is identifiable a priori (e.g., young Hispanic women), and is larger in number than the individual in tailoring applications, but smaller than the population within which the targeted group resides (e.g., all young women). Targeted messaging directs the same message to all members of the identified subgroup based on the assumption that the message chosen for the specified group (i.e., young Hispanic women) will match less well than a message individualized for each member of the group, but better than a general message meant to be applicable to all members of the larger population (i.e., all young women). Work by Davis and Resnicow demonstrates that even when targeting is narrow, assuming homogeneity may be unwarranted (Davis & Resnicow, 2012): for example, their research shows that tailoring to variation in ethnic identity among African Americans improves the effectiveness of messaging advocating fruit and vegetable consumption (Resnicow et al., 2009). Targeting is less time and resource intensive than tailoring but is also potentially less effective.

Culture.

Culture is an extremely complex phenomenon with varying conceptualizations both within and across academic fields (Charles et al., 2006; Sastry et al., 2021). Although no clear consensus definition of “culture” exists, it is generally understood as a set of shared values, norms, behaviors, learnings, assumptions, and beliefs (Betancourt & Flynn, 2009; Davis & Resnicow, 2012; Griffith et al., 2023; Steele-Moses et al., 2009; Stellefson et al., 2008; van der Veen et al., 2014). For example, one widely applied definition of culture derives from the work of Harry Triandis on subjective culture, which conceptualizes culture as a group-level shared pattern of beliefs, attitudes, norms, role perceptions, and values (Davis & Resnicow, 2012; Triandis, 2002). Culture provides a basic framework for people’s understanding of themselves and their environment, thereby shaping the ways in which people think, feel, and behave (Huang & Shen, 2016). This is also true in the context of health, where culture has been shown to influence important beliefs, attitudes, and behaviors, such as perceptions of mental illness and practices related to food selection and preparation (Barrera Jr et al., 2013; Caplan, 2019; Swierad et al., 2017). Additionally, beyond directly or indirectly shaping health-related priorities, decisions, and behaviors, culture can also influence the acceptance and adoption of health communication programs and messages (Kreuter & McClure, 2004).

Defining a group of people as a cultural group is not always straightforward. At a minimum, there should be more homogeneity within the identified groups than between, and more heterogeneity between groups than within, on the identified factor(s). But defining which factors are “cultural” is complex, dynamic, and made challenging by contested meanings. Race and ancestry are often used as proxies for culture in CTT research, but this is both an imprecise and limited operationalization of culture (Griffith et al., 2023). Similarly, rural areas are considered to have a unique culture, even though there is significant heterogeneity in and between rural communities (Hart et al., 2005). For the purposes of the current discussion, however, a precise, consensus definition of what constitutes culture is not necessary — all that is required is that the defined group differs on one or more key factors that are relevant to message content and communication strategies for behavior change. This is because, although cultural attributes are a unique messaging feature and ensuring cultural alignment can be particularly effective in enhancing message receptivity, acceptance, salience, and uptake (Kreuter & McClure, 2004), in terms of the process of effective message design, cultural targeting and tailoring is no different from targeting and tailoring to groups which differ on some quality that is not traditionally considered a part of a group’s culture.

What is the Evidence Base for CTT?

It is often taken for granted that interventions that are culturally appropriate will be more effective than those that are not culturally tailored or targeted (Kreuter et al., 2003). However, this is an assumption that requires empirical confirmation. To gain a sense of the evidence base for CTT, we reviewed meta-analyses of studies evaluating culturally tailored or targeted messages (see supplementary materials for additional details). In general, meta-analyses suggest that CTT is effective for both behavioral and cognitive outcomes, though effect sizes are not large, vary in magnitude across studies, and differ depending on the target population and outcome of focus (e.g., change in attitude vs. behavior). The heterogeneity in study results suggests that targeting and tailoring does not always work and that moderating factors can and do come into play. Some of these moderating factors are methodological (e.g., type of comparison group used), while others are related to intervention content (e.g., surface vs. deeply tailored materials). Relatedly, it should be noted that CTT has been operationalized in a number of different ways in the literature (e.g., using visual design or verbal content to appeal to social identity) (Lapinski et al., 2024) – and the specific ways CTT was operationalized in a given study is not always clearly delineated (Heo & Braun, 2014) – which may also explain some of the heterogeneity observed in the findings.

Additionally, looming large in what appears to be a generally positive assessment of CTT in the literature are questions regarding the quality of the evidence base and inferential processes that would allow the conclusion that CTT interventions are more successful than those designed to work across cultural groups. It is these issues that we aim to address in this essay: namely, what evidence is needed to enable the inference that CTT is effective, and how can this evidence be generated? To this end, we pose several questions to interrogate existing empirical CTT research and guide future CTT work:

• Is cultural tailoring or targeting necessary? CTT has generally been shown to improve the effectiveness of interventions, but is CTT always preferrable to more general adaptations? Are there situations in which a CTT intervention may be less effective than one that is not culturally tailored or targeted?

• What kind of formative research is required to inform CTT interventions? Many studies do not describe whether or how the cultural basis of CTT was established. How can the cultural basis of CTT interventions be empirically established and what are the implications of failing to do so?

• How can CTT studies be designed to allow for the strongest inferences? What impact does the choice of control (or comparison) group have on the inferences that can be drawn about CTT? Although many studies employ random assignment to control and intervention groups, the conclusions that can be drawn about the effectiveness of the intervention will depend on the nature of the comparison group(s) used.

• How can the results of CTT messaging research be made generalizable? The collective body of messaging research should inform subsequent interventions, including identification of message features that are successful and unsuccessful. Are researchers providing sufficiently detailed, objective descriptions of their interventions to enable replication and redeployment in other contexts? Can the active ingredients in the intervention be identified and separated? Which specific message features are most effective (e.g., do deep features of content matter more than surface features?)

• What are the mechanisms of effect for CTT interventions? Does the inclusion of explanatory mediators in studies help shed light on why and how CTT produces its effects? Can this information guide the deployment of the intervention in other contexts and inform future intervention development?

Is Cultural Tailoring or Targeting Necessary?

Although many studies, including the meta-analyses noted above, have shown CTT to be a potentially effective approach for changing health behavior or its antecedents, there are situations in which CTT may not be needed, may not be effective, or may even be problematic. Therefore, it is important to approach CTT interventions thoughtfully and deliberately, ensuring they are necessary and appropriate, before investing in these resource-intensive efforts.

First, there may be situations where cultural adaptations are not necessary, such as when cultural factors are not associated with the specific health behavior being addressed, or not perceived as being relevant by the target population. For example, focus groups conducted to inform the development of a smoking cessation intervention for Hispanic/Latino construction workers revealed that workers did not think cultural characteristics and social norms were important for tailoring the intervention (Asfar et al., 2018). They felt it was more important for the intervention to be responsive to their unique life circumstances (e.g., their workday structure) rather than their cultural perspective (Asfar et al., 2018). This suggests that meeting a target population’s needs may not always require incorporating culturally specific elements into an intervention even though other context-specific factors may need to be considered.

It is important to note that “generic” interventions or health messages are not “culturally neutral” or devoid of underlying cultural assumptions or values (for example, as Charles et al. (2006) point out, standard decision aids that have largely been developed in Western contexts tend to reflect the dominant values and beliefs of those cultures by default, such as an emphasis on autonomy and the individual – rather than the group – in decision making). However, sometimes the cultural views and assumptions underlying the generic message or intervention do not present a problem for populations from other cultural groups (e.g., because those values are shared across cultures, or are not particularly central to the health behavior in question) and will perform equally in those groups as well.

In a 2006 commentary, Lau argues that cultural adaptations must be scientifically defensible and undertaken only when there is evidence to suggest that the application of standard treatments would be ineffective or differentially effective in a specific group (Lau, 2006). According to Lau, cultural adaptation is warranted when data suggest substantial differences between groups in either a) contextual processes influencing vulnerability to the specific health problem or b) response to the standard intervention strategy (in terms of outcomes or engagement) (Lau, 2006). If there is evidence to suggest that a health problem emerges within a distinct sociocultural context in a given group, adaptation may be needed to address these culturally specific risk processes. For example, research supporting a culturally-specific model of alcohol abuse among Native American populations highlights the uniquely important roles of discrimination, historical loss, and enculturation in shaping this behavior, which interventions would likely need to address in order to be effective for this population (Lau, 2006). Similarly, interventions need to be adapted if evidence suggests that certain intervention approaches have limited social validity in the target community as this will negatively affect the community’s engagement in the program, and substantial attrition or marginal participation may hinder the delivery of sufficient intervention dose (Lau, 2006). In the context of health communication interventions, Griffith et al. (2023) note that CTT may be justified when there is evidence of communication factor variability across population subgroups or individuals, such as linguistic differences (e.g., variation in vocabulary across different Spanish-speaking countries) or different preferences for content and delivery, such as presentation format or communication channel, since intervention materials that are responsive to a group’s or individual’s communication preferences are more likely to be noticed, processed, and perceived as trustworthy, understandable, and persuasive.

These criteria suggest that culturally specific adaptations may not be needed if research provides little support for community-specific differences in the health problem or perceptions of the intervention, and when an intervention has been tested in a large, diverse sample and no significant differences in treatment effects, participation, or engagement are observed (Lau, 2006). This selective approach to cultural adaptation is meant to focus efforts where they are most needed (i.e., where there is evidence that the intervention will fail to generalize to a certain community), while safeguarding against less defensible adaptations that may sacrifice fidelity (and, potentially, effectiveness) (Lau, 2006). Adaptation requires resources, and the costs of undertaking a cultural adaptation must be justified by meaningfully superior results compared to the standard intervention (Lau, 2006).

This is an important consideration given that some studies have failed to demonstrate that culturally adapted interventions are more effective than those that are not culturally specific. For example, a randomized control trial (RCT) comparing the effects of a culturally targeted video, a generic (but linguistically appropriate) video, and a factsheet control on mammography uptake among Chinese-American immigrant women found that both videos increased mammography utilization to a similar extent and neither video significantly increased mammography utilization compared to the factsheet control overall (though the culturally targeted video was more effective than the control among low-acculturated women) (Wang et al., 2012). The authors conclude that although culturally-targeted approaches may be more effective for immigrants with the lowest levels of acculturation, their results refute the assumption that cultural targeting is always necessary for minority and immigrant populations (Wang et al., 2012). Additionally, although the comparable outcomes between the two videos may have been due to the fact that the generic video (while not addressing culturally-specific issues) was linguistically appropriate and featured a multiethnic cast that included Asian individuals (Wang et al., 2012), another potential explanation is that perceived barriers may have been a more important driver of mammography uptake in this population than culturally-based views. Women in both video groups reported fewer barriers post-intervention than the control group (Wang et al., 2012). If perceived barriers are the main factor dictating mammogram utilization in this population, the fact that a culturally adapted program is more effective at changing additional determinants (such as cultural views), may not ultimately have a significant impact on intervention outcomes. As noted by Sastry and colleagues (2017), cultural beliefs can influence how groups perceive, interpret, and negotiate structural barriers to health; therefore, it might still be important to investigate and understand the influence of culture in this context, but the results suggest that CTT messages may not always be the right solution.

Culturally-adapted interventions that are unnecessary or do not demonstrate greater efficacy than standard treatments may represent a waste of resources. However, there is also the possibility that attempts at CTT will have a negative impact on intervention effectiveness or acceptability. Although studies show that targeting and tailoring are generally perceived positively (potentially due to increased self-relevancy, fluency, or familiarity), they can also be interpreted negatively in certain situations (e.g., if they are perceived as an attempt at manipulation, or seem to be rooted in stereotypes) (Teeny et al., 2021). For example, pretesting of a computer-tailored nutrition and physical activity intervention for Turkish women suggested that participants found references to religious rules offensive and irrelevant, leading the research team to remove those aspects of cultural targeting from the program (Romeike et al., 2016). Additionally, targeted messages that highlight a group’s worse health status can be stigmatizing and may lead to feelings of hopelessness (Griffith et al., 2023). For example, Landrine and Corral (2015) showed that in some cases, targeted messages highlighting disparities might not be effective in motivating behavior change, but might simply elicit negative emotional responses. In designing CTT health interventions, it is important to ensure that messages do not blame or pathologize the practices of a specific culture (e.g., tying African American food traditions to obesity) (Swierad et al., 2017). Swierad et al. (2017) note that rather than taking a deficit approach, culturally-targeted efforts that highlight the positive, health-enhancing aspects of a culture can empower the target population and validate their experiences and values.

What Kind of Formative Research is Required to Inform CTT Interventions?

Once it is determined that a CTT intervention is needed and appropriate, a strong empirical base for cultural meanings and practices must be established in order to enable inferentially strong message development. Importantly, assuming intervention fit with a target group’s cultural context without confirming the empirical basis for that match should be avoided. Culture is a complex, multi-faceted, multi-level phenomenon that varies across social contexts, and evolves over time (Singer et al., 2016). Formative work is therefore needed to ensure that communities of interest “identify with the culture represented in the message” (Yzer et al., 2018, p. 105). Ideally, these communities are engaged in formative research and included in the co-creation of messages to ensure that themes, content, and style truly reflect community perceptions and preferences (Malla et al., 2024).

However, message development in CTT interventions is often guided by assumptions, prior work, comparison with a dominant social group, or researchers’ own experiences. Sastry et al. write critically of what they label “culturally sensitive” approaches to health communication, where the role of culture is acknowledged, but the exact parameters of culture are defined by researchers rather than the target community (Sastry et al., 2021). They argue that approaches that seek to create targeted or tailored interventions based on cultural characteristics identified as being most relevant by researchers; treat culture as a static set of beliefs, practices, and values; and emphasize external empowerment by outside experts, risk erasing the cultural voices of marginalized communities (Sastry et al., 2021). Centering a cultural group’s experiences allows for different conceptual models of health and health behaviors to be identified, which may help inform which factors need to be addressed in the intervention (Griffith et al., 2023). The literature on co-construction emphasizes that cultural practices and meanings should emerge collaboratively through dialog between community members and scholars (Dutta & Pal, 2010; Sastry et al., 2021). Approaches that entrust the target population with decisions regarding which cultural variables are relevant for a given health issue could help ensure that interventions are authentic, locally relevant, and meaningful to the communities involved.

Across the CTT literature, there is wide variability in formative research methods as well as level of participant engagement used to ascertain the specific cultural elements that resonate with the group and enhance the effectiveness of targeted or tailored messages. For example, evidence may be drawn from epidemiological data (Van Der Veen et al., 2012) or prior research conducted with the group (Chu et al., 2021; Singelis et al., 2018), which entail no direct involvement of the target population. Other strategies for obtaining cultural perspectives from the target community include incorporating the experiences of research team members who share the target population’s cultural background (McFarlane & Morgan, 2021; Singelis et al., 2018) and soliciting insights from community partners and advisory boards (Permuth-Wey et al., 2010; Steele-Moses et al., 2009; Wang et al., 2008). More commonly, qualitative methods such as key informant interviews (Hurtado-de-Mendoza et al., 2020; Robbins et al., 2019) or focus groups (Chu et al., 2021; Wang et al., 2008), and quantitative methods such as surveys (Davis et al., 2010; Wang et al., 2008), are used to directly solicit insights from the target population on key cultural elements foundational to message tailoring. Thoughtful, inclusive, dynamic, participatory formative research that ensures cultural authenticity is critical; CTT interventions must be based on valid, representative empirical work rather than limited perceptions of community needs (Lau, 2006).

In addition to identifying salient cultural meanings and practices, formative research plays a central role in message development and pilot testing. Many health communication campaigns pretest concepts and materials with community members (Robbins et al., 2019; Van Der Veen et al., 2012) to ensure the proposed messages resonate with the target population before their full-fledged deployment. Message development and pretesting work should be iterative, with input from the community of focus guiding and informing message creation and revision; multi-stage, to allow content and format changes to be integrated and tested; and characterized by multiple or mixed methods, in order to capture a range of outcomes and feedback. Additionally, message testing should strive for the highest possible level of community involvement whenever possible, including by engaging members of the target community directly in message creation from the outset. This type of co-creation has several benefits, including engaging and articulating the experience of those most seriously affected by the health or social issue being addressed, producing communication interventions derived by and reflective of target communities, and fostering trust and encouraging compliance with health messages and interventions among target communities (Malla et al., 2024)

Rigorous formative research can enhance the empirical robustness of conclusions drawn about the cultural group’s core characteristics. It is therefore important that the formative research methodologies used ensure representativeness and timeliness of the sample of cultural informants and communities with whom the messages are being tested, recognizing that culture is heterogeneous and fluid (Chu et al., 2021; Singelis et al., 2018), and that multiple cultural identities may intersect in a way that affects response to messaging (Griffith et al., 2023) – such as receptiveness to different message sources. In addition, formative research should adhere to rigorous scientific principles related to the collection, analysis, and interpretation of evidence and data, which are documented and replicable — all toward the goal of understanding how the basis of cultural fit was discovered, incorporated into the message design, and ultimately proven effective or ineffective. Given the importance of both ensuring CTT interventions incorporate cultural elements that resonate with the target group and that proposed messages align with the preferences of the target community, it is critical that researchers not only conduct rigorous formative research, but also report the details of this work in published studies. Evaluating the quality of CTT evidence must include an assessment of the empirical basis from formative research. Having this formative evidence (including message testing results) explicitly included in the research protocol allows other scholars to assess the empirical basis for CTT.

How Can CTT Studies be Designed to Allow for the Strongest Inferences?

Although having a comparison group is the sine qua non enabling researchers to draw strong conclusions from CTT studies, not all comparison groups are created equal. To infer that an intervention is effective, some comparator is necessary, but to infer that the intervention is effective because of CTT requires specific kinds of comparison groups, which are infrequently deployed. To enable strong inferences to be drawn, CTT intervention testing studies should be designed in a way that enables the following core questions to be answered: 1) does the message adapted for the target group work better for that group than a message that is not designed for that group (e.g., a generic message or a message targeted to a different group)? and 2) does the targeted message work uniquely well for the target group or does it work just as well for other groups or for the general population? Without affirmative answers to both questions, it is not possible to declare that research evidence supports a claim of CTT effectiveness.

Considerations for designing a study that would allow for appropriate comparisons are outlined in Figure 1. Suppose a researcher is working with a cultural group (the “red group”), and the researcher develops a CTT message based on formative work with this group (the “red message”). As illustrated in Panel A, the minimal test of a successful CTT intervention requires that the message designed for the target group (the “red message”) be more effective than a generic (or culturally mismatched) message (the “blue message”) in that group. If this comparison is not statistically different, then the messaging intervention fails. Such a failure suggests either that formative research and message design need be revisited or that cultural factors are not primary drivers of the target behavior in this group, making CTT unnecessary as it is unlikely to confer benefits beyond those offered by messages that are not culturally-specific.

Figure 1. RCT design to ensure valid inference of CTT effectiveness.

On the other hand, if the criteria in Panel A are met, this suggests that the culturally adapted message is effective, but that evidence is insufficient to conclude that the message is uniquely effective for the target cultural group. It could still be possible that the “red message” is just as effective in other groups due to similarities in beliefs, values, and moral foundations or format preferences between groups. For example, researchers may test a generic educational intervention for colorectal cancer screening against a CTT educational intervention in a sample of African Americans and observe that the CTT intervention is more effective than the generic intervention in this group, but without testing the intervention in any other cultural group, this study design would not allow the research team to conclude that the intervention is uniquely effective for African Americans. It is possible that the CTT message would outperform a “generic education” condition in any population just due to being, for example, more engaging or due to its emphasis on family support, which may also be important in other cultures.

Therefore, in addition to Panel A, a complete test of CTT requires Panel B, where the “red message” is shown to be ineffective, or at least less effective, in a comparison group (the “blue group”). This empirical comparison makes the uniqueness argument for CTT, showing that messages targeting the unique cultural characteristics of the “red group” work significantly better for that group than for other groups who do not share those characteristics or differ in the extent to which they express, endorse, weigh, or value them.

A study by Baezconde-Garbanati et al. (2014) provides an example of a study that meets the criteria established in Panel B. In this study, researchers compared responses to “Tamale Lesson”, a narrative-based video intervention to promote cervical cancer screening and prevention, between Mexican-American women (the primary audience the intervention was targeted to) and non-Hispanic White women (the comparison group). Although participants in both groups responded positively to the video, there were statistically significant differences between the groups, with Mexican-American women enjoying the film more than non-Hispanic White women, watching the video more often, and being more likely to show the video to others. More importantly, the narrative video reduced the cervical cancer screening disparity observed between the groups: at baseline, non-Hispanic White women were significantly more likely to have had a Pap test in the previous six months than Mexican-American women, but at follow-up, Mexican-American women were more likely to have had a Pap test than non-Hispanic White women (Baezconde-Garbanati et al., 2014). These findings demonstrate that although the video was positively received by both groups, the intervention was particularly effective in the target group, which suggests the culturally resonant elements were essential to the effectiveness of the video among Mexican American women.

A study conducted by An et al. (2021), provides an example of a design that meets the criteria outlined in both Panel A and Panel B. In their study of advertising appeals to promote breast cancer screening in Qatar, the researchers developed two versions of an ad, one featuring surface-level cultural elements designed to appeal to Arab women and the other featuring cultural elements designed for Filipino women, and tested both ads in both groups of women. The two ads were evaluated using Facebook’s A/B testing feature, with the outcome of interest being click through rates (CTR). The culturally resonant ad for each group performed significantly better than the non-culturally resonant ad: among Arab women, the CTR for the advertisement featuring an Arab model was nearly twice the CTR of the advertisement with the Filipino model (2.78% vs. 1.32%), similarly, in the Filipino group, the culturally congruent ad had a substantially higher CTR rate compared to the ad designed for the other group (2.92% vs. 1.88%). Although click through rates do not necessarily translate into screening intention or behavior, they are nonetheless an indicator of audience interest and message fit.

Lastly, Panel C illustrates a flawed design seeking to short cut the steps in Panels A and B by directly comparing the target group (red) receiving its CTT message (red), to the comparison group (blue) receiving a comparison message (e.g., a generic message). The problem with this design is that two factors (the target group and the message) are different between the intervention and the comparison conditions so that differences cannot be uniquely attributed to the message-group fit. This shortcut is tempting because it appears to address the uniqueness issue but cannot do so unequivocally.

In sum, to determine whether a CTT approach is effective for a given community or population, it is critical to have sufficient grounds for comparison. The choice of comparison group is therefore critical. Figure 1 aims to make clear that a CTT intervention must not only outperform a generic (or mismatched) message in its target group but should also perform less well in a non-target group. A CTT intervention that works for different groups, or even all groups, undercuts the need for unique CTT interventions by cultural group. The intervention might still be effective, but it would not be uniquely effective due to the culturally specific elements.

We acknowledge that the criteria outlined above impose substantial demands on researchers evaluating CTT interventions. Most CTT communication research aims to address real-world health communication gaps (e.g., lack of culturally sensitive or linguistically appropriate messages available for an underserved group), with the goal of achieving strong causal inference usually being secondary. The ability to generate strong causal inference needs to be balanced against feasibility of implementation, external validity, and practical considerations in efforts seeking to address disparities and improve the public’s health. At the same time, the costs of creating CTT interventions are high, so being able to conclude more definitively that the culturally adapted message is, in fact, effective, may be worth the investment.

How Can the Results of CTT Messaging Research be Made Generalizable?

Although the literature suggests that CTT interventions generally succeed in improving health-related outcomes, our knowledge of the messaging features and formats that predict the effectiveness of CTT interventions remains limited. More robust conclusions regarding the types of messages or message elements that are successful (or not) are needed to inform application of accumulated CTT knowledge to different contexts. In this section, we examine three barriers to generalization in CTT: surface versus deep message elements; hidden “active ingredients”; and objective versus subjective message features.

Surface and deep elements.

CTT intervention messages consist of both surface and deeper structural features (Resnicow et al., 1999). Tailoring or targeting on surface structure refers to the incorporation of a group’s observable characteristics into an intervention to increase the relevance of the materials to their preferences, interests, and identities (Griffith et al., 2023). These content features include elements such as language, symbols, food, setting, music, and clothing, among others (Resnicow et al., 2000). Tailoring or targeting on deep structures involves incorporating a group’s core values, cultural norms, beliefs, culturally unique experiences, etc. that are important to how individuals in that group perceive themselves and/or the target health behavior (Griffith et al., 2023). Examples of deeper content features include constructs such as religion/spirituality, familismo (importance of family), and communalism (Resnicow et al., 2000).

Individual research studies often do not describe messaging in sufficient detail to allow for a complete understanding of the exact surface-level and deeper elements being incorporated. Several meta-analyses have attempted to distinguish between the impact of surface-level elements and deeper-level components of messaging when the data allow. These meta-analyses indicate that more effective interventions tend to target or tailor based on deeper components of culture (Brevik et al., 2020; Copeland et al., 2018; Huang & Shen, 2016; Noar et al., 2007; Sohl & Moyer, 2007). This is not to say that surface-level features are not important aspects of CTT or that the distinction between deep and surface-level cultural elements is always clear; obviously, certain surface features must be present for comprehension (e.g., language), identification (icons, symbols, dress, etc.), and at least minimal levels of engagement to occur. But if culturally targeted messages are based primarily on surface-level features, they may miss core differences among cultural groups in underlying values and moral considerations and in the process be less effective than if more central cultural elements were manipulated. For example, a study by Yzer et al. (2021) compared messages on the importance of tobacco in Native American cultural traditions (targeting deep cultural elements) to standard health-focused tobacco messages among Native American youth. The results showed that appealing to the cultural value of tobacco to discourage the non-traditional use of tobacco in cigarette smoking was more effective in changing smoking-related attitudes and intentions than standard health appeals among Native American smokers (Yzer et al., 2021) when other messaging elements were kept consistent, including surface cultural features (e.g., use of Native American models).

However, it should be noted that although research shows that targeting and tailoring on deeper cultural elements is effective in the groups they are designed for, there is limited work exploring whether these interventions are uniquely effective for those groups, or whether they would also be effective in other groups. This is an important consideration, as some extant work suggests there is substantial similarity in values across cultural groups. For example, in a summary of research into core values across 82 countries, Schwartz concludes that there is consistency in hierarchies of values, with benevolence, universalism, and self-direction values being the most important, and power and stimulation values being the least important (on average), across national boundaries (Schwartz, 2012). Although tailoring on basic message features is necessary for the most rudimentary forms of successful information transmission, core values may not be as diverse across cultural groups as is generally assumed, allowing messaging elements based on value and moral considerations to be more consistent across groups. This is a testable proposition and, if priorities in value and moral hierarchies are common, then incorporating “deep” cultural features in messaging can be accomplished efficiently.

Active ingredients.

CTT message interventions often have multiple features and studies often fail to describe these features in sufficient detail or to employ study designs that enable the impact of these different features to be teased apart. As noted by Griffith et al. (2023), in many studies, the CTT intervention being tested often differs from the “comparator” intervention across multiple features, some of which are cultural elements and some of which are not (e.g., dose, intensity, format, channel), making it difficult to determine whether the observed effects of the intervention are attributable to the CTT components, or some other aspect of the intervention. In these situations, even if an intervention is shown to be successful for the specific public health problem it was designed to address, without being able to identify the “active ingredient(s)” of the intervention, use and generalization of findings to other contexts and applications is stymied. Rigorous designs that enable the cultural components of CTT interventions to be isolated are needed to advance the field. The most straightforward way of achieving this might be to test one CTT component (or a limited set of CTT components) at a time, while holding other intervention features constant – though this may be somewhat inefficient if researchers are interested in several different messaging elements. Factorial designs can also be used to disentangle active ingredients, but feasibility is an issue, as the inclusion of even a relatively small number of binary factors can quickly escalate into an impractical design (e.g., testing six binary message features would translate factorially into 64 experimental conditions). Alternatives such as fractional designs, which as are less costly to complete than factorial trials (Nair et al., 2008), can be considered, though they are more complex and may be less familiar to researchers.

Again, we recognize that much of the work on CTT is motivated by a need to improve the health of underserved populations using every tool at researchers’ disposal. Therefore, the criterion of advancing the CTT knowledge base by designing studies in a way that allows “active ingredients” to be isolated will at times take a back seat to addressing an immediate public health concern in the most straightforward and economical way possible. However, being able to identify the most important elements of a CTT intervention is not just theoretically useful, but can generate efficiencies in resource allocation and better serve public health in the long term by making it possible to re-deploy, adapt, and scale up tested interventions instead of starting from scratch for each new context or target group.

Objective vs. subjective message features.

Most descriptions of message interventions in CTT studies are too broad and diffuse to allow generalization to other contexts. For example, describing an intervention as consisting of “culturally sensitive education” may make intuitive sense, but does little to help the next research team looking to build on the prior intervention. Specifying what exactly was done to create a “culturally sensitive education intervention” would allow subsequent researchers to replicate accurately all or some of the successful intervention or avoid operationalizations that might have been unsuccessful. Therefore, to advance the field, CTT research needs to employ messages with “intrinsic features” (i.e., objective features) and those features need to be reported fully in published studies, for the benefit of meta-analysts as well as applied researchers. This will help ensure that evidence-based interventions are able to be designed across applications efficiently (O’Keefe, 2003) rather than from scratch for each new undertaking. For example, argument strength has been shown to be key to message effectiveness, but advice to use “strong arguments” is too subjective to be useful for message designers (Cappella & Li, 2023). On the other hand, specifying objective content features that increase perceptions of argument strength (e.g., presence of support and warrants, evidence in the form of explicit premises and quantitative specificity) can help guide message development (Cappella & Li, 2023). Features of the message that are more subjective or intuitive in nature (i.e., those, like perceived argument strength, that can only be assessed by evaluating the cognitive response of the target audience to the message) cannot be as easily manipulated or designed into a message in a principled way (Cappella & Li, 2023).

Discovering message elements that are effective and can be re-deployed for other contexts, other behaviors, and other cultural groups is necessary to advance CTT messaging science. However, the cumulation of knowledge about messaging effects cannot happen when intervention descriptions are subjective, when the operative elements are mixed in a “kitchen sink” intervention, or when details of messaging elements are omitted or given short shrift. At a minimum, researchers should include detailed descriptions of intervention features in all published studies, which could do much to advance the science of CTT while asking relatively little of research teams in terms of resources or effort. When possible, researchers should also use study designs that allow active message ingredients to be isolated in order to increase the ultimate usefulness of their work to others in the field.

What Are the Mechanisms of Effect for CTT Interventions?

Although extant research suggests that CTT is often effective, more work is needed to understand why and how CTT produces its effects. By identifying the mechanisms through which CTT operates, mediation is an empirical precursor to explanations for CTT that can enable findings to generalize beyond the specific application for which they are created. Such information helps advance the design of effective interventions that enhance the impact of known mediators which, in turn, affect behavior. Knowing how to activate the processes of CTT most effectively would enable researchers to optimize intervention outcomes.

Mediation models that offer causal explanations are especially needed, as they suggest key mechanisms that can be leveraged and tested in other contexts. One possible approach to such a model for CTT messaging is an extension of the Elaboration Likelihood Model (ELM) to the issue of “message matching” – for example, in the form of CTT – in the health context (Briñol & Petty, 2006). A possible reason that targeted messages (e.g., those from a culturally congruent source) are more effective than generic messages is that they elicit greater elaboration due to their greater perceived relevance to the recipient (Lustria et al., 2016; Updegraff et al., 2007; Yin et al., 2010). This modest but plausible extension of the ELM can provide guidance for individual researchers and meta-analysts searching to study and explicate the processes explaining successful CTT messaging, thereby enhancing their generalizability. Understanding the mechanism of effect for CTT is also important because it can allow researchers to make theory-driven predictions regarding the conditions under which CTT will be effective. For example, if elaboration is the mechanism of effect for CTT, this suggests that argument strength is a potential moderator of CTT effectiveness: when coupled with strong arguments, matched messages should be more effective, but when coupled with weak arguments, matched messages that stimulate greater elaboration may be less persuasive (Teeny et al., 2021; Updegraff et al., 2007).

Some specific studies offer insight into other potential mediators of CTT interventions, such as trust (Yin et al., 2010), identification with the message (Yzer et al., 2018), and perceived self-relevance (Singelis et al., 2018). For example, the “Tamale Lesson” video intervention described earlier, which was designed to increase cervical cancer screening among Mexican American women, showed that the narrative was not only more effective for these women in comparison to non-Hispanic White women, but that Mexican American women were more transported by the narrative, had stronger emotional reactions, and identified more with the characters than non-Hispanic White women (Baezconde-Garbanati et al., 2014). This suggests that CTT in this context may work by increasing the relevance and saliency of the health messages. This study suggests that engagement is a likely route through which messaging interventions generate changes in key outcomes, and also provides guidance about message features that activate engagement (e.g., use of narrative, demographic matching). Additional studies that pursue the dual goals of conducting sound applied CTT research in service of the public’s health while simultaneously advancing knowledge of explanatory mechanisms are crucial to advancing CTT application and theory.

Conclusion

Our goal in this essay is to provide a roadmap for advancing the science of CTT, to ensure that CTT research yields results that are both robust and practically useful, and to inform the development of interventions that are maximally effective but also efficient. Meta-analyses generally support the claim that CTT interventions are effective; however, a closer look at the empirical findings to date suggests that the current evidence base does not allow conclusions to be drawn regarding the necessity of CTT messages nor the unique impact of CTT on message effectiveness. There are too many alternative explanations of the empirical data to allow a strong inference about CTT to be made. To address these gaps in the CTT literature, we have proposed a set of criteria that researchers can employ to interrogate extant research on CTT messaging and improve future efforts.

The first key consideration is whether CTT is necessary to address a given health issue in the target population. CTT interventions are resource intensive and should not be undertaken unless: 1) cultural factors are driving (or at least associated with) the health behavior or outcome, 2) there is evidence of substantial difference in either the context of the health problem or response to the standard intervention in the target population, and 3) the risk of negative consequences (e.g., stigmatization) from a culturally specific appeal is minimal. It should not be assumed that culturally based appeals will outperform more general appeals; the resource-intensive work of culturally adapting messaging must be scientifically justified.

Once a decision is made to move forward with a CTT intervention, researchers must then ensure that the cultural basis for the CTT intervention is derived from a robust, reliable, and representative empirical base. The cultural basis for communication interventions cannot be assumed or imputed without serious empirical grounding (from both formative research and message pre-testing with the target group). Evaluating studies in terms of the empirical base for the fit of the intervention to the focal cultural group is a sine qua non of CTT research. Future studies should more fully report the formative and message-pretesting work that provide the evidence for the selection of cultural elements employed in message design and demonstrate that the messages were effective and perceived as culturally relevant by the target group. How the cultural basis for the CTT intervention was established should also be explored more carefully in future meta-analyses of CTT research.

After establishing a strong cultural basis for their CTT intervention, researchers should consider study designs that 1) enable strong inferences to be made about the unique effects of CTT, 2) allow knowledge of message effects to be accumulated across studies, and 3) contribute to the field’s understanding of mechanisms of effect in CTT interventions. First, to conclude that an intervention is effective because of CTT requires researchers to establish that the CTT message works better for the target group than a non-CTT message and that the CTT message is uniquely effective for the target group (i.e., the CTT message does not perform equally well in other groups). This requires the use of both a comparator message and a comparator cultural group in the study design. Reporting detailed information regarding the comparison conditions and groups used in CTT studies would enable the validity of conclusions regarding the effectiveness of CTT to be evaluated and would allow meta-analysts to assess the impact of different types of control groups and study designs on the heterogeneity of effect sizes. Second, to advance knowledge of message effects, studies should carefully delineate message features (e.g., distinguishing surface and deep structure cultural components), employ designs that allow active ingredients to be identified and assessed, and employ more objective message elements where possible. These steps would allow researchers not just to demonstrate that their specific intervention is effective, but to shed light on the necessary conditions for CTT interventions to be effective more generally and to inform message design in future work. Finally, if CTT research is to have longer term, broader impact on the field, more attention needs to be paid to mechanisms of effect. Identifying key mediators could help researchers understand how an intervention can be deployed in other contexts and design more effective interventions by focusing on key levers of change; assessing potential mediators in CTT studies could also help build the empirical base for subsequent theory development. Of course, mediation is only an initial, correlational step in establishing true causal sequence; less frequently used designs called parallel encouragement designs can help further establish (or disconfirm) the causal nature of mediators (Kim & Cappella, 2023).

The criteria outlined above can help guide the assessment and interpretation of existing CTT research, and ensure that future CTT research is valid, replicable, rigorous, and robust. Although we are keenly aware that these recommendations may add considerable burden to researchers conducting CTT studies, it is our hope that the considerations discussed in this essay will help advance future CTT research and enable CTT interventions to ultimately be more effective.