Unpacking elderberry demand: how consumer segments and socioeconomic factors shape purchasing behavior in the United States Open Access

Over the past 2 decades, elderberry (Sambucus spp.) has gradually moved from being a niche crop to gaining broader attention in North American agriculture, attracting interest from consumers and farmers alike. Known for its antioxidant-rich properties and wide range of uses—from immune-boosting remedies to alcoholic beverages—elderberry’s potential continues to grow (Chuang et al., 2014; Curtis et al., 2024; Osman et al., 2023). To build on this momentum, it is important to understand consumer behavior and the factors influencing demand for elderberry products.

In particular, significant questions remain regarding consumer attitudes, market segmentation and the drivers of consumption decisions. To address these issues, this paper asks how consumer segments—defined based on consumer attitudes—along with other socioeconomic characteristics, influence the likelihood of purchasing elderberry products and the frequency of consumption. Specifically, this study uses a 2022 US consumer survey conducted by Cai et al. (2024), which reports on a series of consumer attitudes toward elderberry products, to identify consumer segments through factor and cluster analysis. Then, it investigates how segment membership and other socioeconomic factors influence the demand for elderberry products using a double-hurdle model, which treats the purchasing decision and the frequency of consumption as independent processes (Cragg, 1971; Skevas et al., 2014).

Forecasts predict a 5.8% annual growth rate of the elderberry market between 2023 and 2028 (Technavio, 2024). Yet the literature on this emerging market remains sparse and fragmented. Early work focused predominantly on supply-side challenges, from horticultural practices (Charlebois et al., 2010) to market entry barriers like scale economies and credit constraints identified through stakeholder interviews (Cernusca et al., 2011, 2012; Cernusca and Gold, 2013). On the demand side, foundational studies by Mohebalian et al. (2012, 2013) provided initial segmentation into health-conscious consumers and analyzed preferences for specific products like juice and jelly, finding value in health claims and local production. A recent replication and extension by Cai et al. (2024) confirms a significant rise in consumption and identifies consumer willingness-to-pay for attributes like organic and American-grown labels. However, this existing demand-side research has two key limitations: first, it often focuses on specific product types; and second, it primarily analyzes stated preferences (e.g., via conjoint or discrete-choice experiments), falling short of explaining how these preferences translate into actual purchasing decisions and consumption frequency.

A robust, parallel stream of literature in food economics demonstrates the power of attitudinal segmentation to decode complex consumer markets, moving beyond simple demographics and looking into the underlying psychological drivers. For instance, Sgroi et al. (2024) employ K-means clustering on socioeconomic characteristics of Sicilian consumers (e.g., age, gender and city size) and identify three actionable typologies: small-town young consumers, middle-aged males from medium-sized cities and older females from large cities. In a sustainability framework, Schäufele-Elbers and Janssen (2023) utilize K-means to segment consumers based on their expenditures for organic food (environmental proxy), meat (climate proxy) and sweet snacks (health proxy), revealing that pro-environmental behavior in one area does not guarantee it in another. Similarly, segmentation of food waste behaviors identifies groups like “conscientious conservers” and “guilty carb wasters” (Li and Roe, 2024). Other research used principal component analysis to reduce numerous behavioral traits into three factors—“healthy and sustainable”, “hedonic” and “price-conscious” food choices—and then conducted a hierarchical cluster analysis on these factors to identify consumer segments (Hempel, 2024). These studies underscore that global attitudes such as “health consciousness” are insufficient and that nuanced segmentation is crucial. However, while this literature excels at profiling who consumers are and what they value, it stops short of linking these detailed attitudinal segments directly to observed purchasing and consumption behavior for emerging food products.

To analyze how attitudinally-defined segments might influence purchasing decisions and consumption patterns—whether in terms of frequency or quantity—the double-hurdle model (Cragg, 1971) can be used, as it explicitly separates the initial market participation decision from subsequent consumption intensity. In food demand analysis, this model has been applied to show that different factors can govern each stage. For instance, a product’s visual appearance may drive the decision to purchase, while its taste determines consumption quantity (Ray et al., 2024). Similarly, socioeconomic factors can have divergent effects: while parents’ educational level may increase their willingness to participate in a cost-shared school food program, having more children may decrease the amount they are willing to pay (Gupta et al., 2023). The double-hurdle model’s flexibility to accommodate such complex, two-tiered decision-making processes has been validated across diverse contexts, from tea consumption (Chen et al., 2020) to willingness-to-pay for credence attributes like winemaker gender (Gallais and Livat, 2024). Notably, focusing solely on the purchase decision—or combining purchasers and non-purchasers—may fail to capture the full complexity of consumption behavior.

This paper contributes to the literature in several important ways. First, it advances the elderberry literature by moving beyond stated preferences for specific products (e.g., elderberry juice and jelly) to analyze actual purchasing behavior across the entire product category and by developing a more nuanced, attitudinally-based consumer segmentation that captures dimensions beyond basic health consciousness. Second, it contributes to the consumer segmentation literature by directly linking attitudinally-based segments to observed market behavior, thereby addressing the common attitude–behavior gap—i.e., the disconnect between an individual’s attitudes or intentions and their actual behaviors (Vermeir and Verbeke, 2006; Szeląg-Sikora et al., 2025)—and providing a framework that moves from profiling who consumers are to predicting how they behave. By integrating detailed consumer segments into the analysis of both market participation (purchase decision) and consumption intensity (frequency of use), this study offers a multidimensional perspective on market heterogeneity that can inform targeted marketing and agricultural development strategies for the growing elderberry sector.

The remainder of this paper is structured as follows. Section 2 describes the data from the 2022 consumer survey. Section 3 presents the methodology. Section 4 reports the results, including purchasing and consumption frequency, consumer attitudes and perceptions, segmentation analysis and the determinants of both purchasing decisions and consumption frequency. Section 5 discusses the key findings, linking them to the literature and highlighting their implications for stakeholders. Finally, Section 6 concludes with a summary of contributions, limitations and avenues for future research.

This study uses data from a non-probability, online survey of US consumers conducted by Cai et al. (2024) in 2022. The survey was pretested with a small group of 17 consumers in October before full deployment in November using the Qualtrics online survey platform. Participants were recruited through Qualtrics Research Services, which maintains a diverse panel of over 4 million respondents and provides compensation such as e-gift cards, retail shopping points and air miles. Invitations and instructions were sent via email with daily reminders until the predetermined quota of 1,000 responses was met. In total, 1,036 surveys were collected, of which 597 provided complete data on all the variables of interest and were included in our analysis.

The survey was divided into four sections: the first section inquired about consumers’ familiarity with elderberry products, their past purchasing behaviors and their consumption frequency. Specifically, the survey collected consumption patterns for eleven elderberry products, including juice, fresh berries, wine, concentrate, tea, syrup, jelly, pie, vitamins, drinks and others. Respondents were asked whether they had purchased any of these products—a binary response variable—and their overall frequency of consumption, captured through a six-point Likert scale with the following levels: “less than once a year”, “1–2 times a year”, “3–10 times a year”, “once a month”, “multiple times per month” and “once a week or more”. Importantly, the frequency question was asked in relation to elderberry products in general rather than for specific product types.

The second section gauged consumers’ attitudes and perceptions toward elderberry products. Specifically, the consumers were asked about their degree of health consciousness, knowledge of food-related health benefits, awareness of elderberry’s nutritional and antioxidant value, and elderberry’s richness in vitamins and minerals. Responses were recorded using a five-point Likert scale ranging from “strongly disagree” to “strongly agree”. Moreover, respondents were asked about the relative importance of several product attributes, including nutrition, taste, price, ingredients, packaging, brand, environmental benefits and origin (US produced vs imported), all measured on a five-point Likert scale with the options “not at all important”, “slightly important”, “moderately important”, “very important” and “extremely important”.

The third section conducted a discrete-choice experiment to elicit consumer preferences for selected attributes of a 355 mL bottle of elderberry juice. However, this survey component was not used in the present study, since it had already been exploited by Cai et al. (2024). Finally, the fourth section collected socioeconomic characteristics of the respondents, including gender, education, household income, age, number of children, marital status, race and distance to the nearest urban center. Gender was reported in four categories (male, female, non-binary and other). For parsimony, we consolidated them into two groups: male and female/other gender ^[1]. Educational attainment was collected across eight levels and grouped into four categories based on the typical years of education required to complete each degree: high school, associate’s degree, bachelor’s degree and postgraduate. Household income was originally reported in seven levels and was consolidated into three categories: under $35,000; $35,000 to $74,999; and $75,000 or more. Age was originally reported in seven groups and consolidated into three categories: under 35, 35 to 54 and over 54. The number of children was treated as a discrete numeric variable. Marital status was recorded as a binary variable: married or single. Race included multiple categories but was condensed into three groups—White, Black and other race—due to minimal responses in the remaining categories. Finally, distance to the nearest urban center was categorized into three groups based on proximity to the respondents’ residence: less than 5 miles, 5 to 29 miles and more than 29 miles. Summary statistics of the socioeconomic variables are presented in Table 1.

As shown in Table 1, the sample was relatively balanced by gender (47.1% male, 52.9% female/other) and age, with the largest cohort being middle-aged (35–54 years, 40.7%). Educational attainment was diverse: 37.2% had an associate’s degree, 24% a bachelor’s and 23.6% a high school diploma. Income was fairly evenly distributed across the three categories. The sample was predominantly White (76.4%), with Black (11.4%) and other racial groups (12.2%) also represented. Most respondents lived in urban or suburban areas, with over half (50.8%) within 5 miles of an urban center. Approximately half of the respondents (46.9%) were married, and the average household had fewer than one child (0.76).

This paper proceeded in four methodological steps. First, the twelve attitudinal variables captured by the survey were consolidated into fewer categories using factor analysis. This technique shrinks the variable space by combining two or more correlated variables into single factors, which are linear combinations of the original variables, allowing the detection of structure in the relationships among them (Skevas et al., 2014). Prior to extraction, the suitability of the data for factor analysis was assessed using the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity (Watkins, 2018). Principal component factoring (PCF) was used as the extraction method, retaining factors with eigenvalues greater than 1. The retained factors were then subjected to an orthogonal varimax (Crawford–Ferguson) rotation to enhance interpretability. Finally, the internal consistency of items loading highly (≥0.6) on each factor was assessed using Cronbach’s alpha (Cronbach, 1951).

Second, after consolidating the attitudinal variables, cluster analysis was conducted (Kaufman and Rousseeuw, 1990) to divide heterogeneous markets into clearly defined groups or segments of consumers based on the factors identified in the previous step (Welk et al., 2025). The optimal number of clusters was determined using the Calinski–Harabasz pseudo-F index, with the solution that maximized this index selected as optimal.

Third, to examine how socioeconomic characteristics varied across clusters, differences in cluster means were assessed using analysis of variance (ANOVA) for the continuous variables (i.e., number of children) and chi-squared tests for all binary variables.

Fourth, a double-hurdle model (Cragg, 1971) was implemented to explore the determinants of both purchasing and consumption frequency. The double-hurdle framework was preferred over alternative approaches such as Tobit or Heckman-type models because it explicitly separates the decision to participate in the market from the decision regarding intensity of consumption. This distinction was crucial in our context, as some consumers did not purchase elderberry products at all, while others varied in how frequently they consumed them. Specifically, the model assumes a two-stage data-generating process: in the first stage or “hurdle”, consumers decide whether to purchase any elderberry product. The first hurdle is considered “crossed” if the respondent reported purchasing any positive quantity. Conditional on a reported purchase, the second stage models overall consumption frequency at the aggregate product level, as the survey did not collect frequency data for individual product types. Thus, the double-hurdle model can be represented by two sets of equations and corresponding censoring rules, as follows:

where $di*$ is a latent variable defining the purchasing decision and $di$ is the observed binary outcome; $yi*$ is a latent variable defining the frequency of consumption and $yi$ is the observed ordinal frequency; $xi$ is a vector of explanatory variables for both purchasing and frequency; $α$ and $β$ are the vectors of parameters to be estimated; and $ε1,i$ and $ε2,i$ are independently and identically normally distributed random errors with zero means and constant variances (Engel and Moffatt, 2014). Because the estimated parameters $α$ and $β$ do not have a direct economic interpretation, it is common to compute and report average marginal effects.

For the first hurdle, in particular, a binomial probability model—specifically, a logit model—was employed, which is appropriate for a binary dependent variable (⁠$di$ in Eq. (1)) equal to 1 if a respondent purchased any elderberry product and 0 otherwise. For the second hurdle, an ordered logistic regression model was used, suitable for a dependent variable with discrete, ordered categories (⁠$yi$ in Eq. (2)) representing the frequency of consumption. In this case, the ordered categories were “less than once a year,” “1–2 times a year,” “3–10 times a year,” “once a month,” “multiple times per month,” and “once a week or more”.

The proportional odds assumption of the ordered logistic model was assessed using the Brant test (Brant, 1990) to confirm the appropriateness of this specification. The results derived from these methodological procedures are presented in the subsequent section.

Figure 1 shows that 57% of respondents in our sample purchased elderberry products, while 43% did not. As noted earlier, a total of eleven product types were considered, including juice, fresh berries, wine, concentrate, tea, syrup, jelly, pie, vitamins, drinks and others. Among the 341 purchasers, the most commonly bought products were elderberry-based vitamin supplements (49%), juice (36%) and wine (26%), as shown in Figure 2—totals exceed 100% because respondents often purchased more than one product type. This distribution reveals a strong preference for vitamin supplements, suggesting a significant appeal of elderberries’ wellness and immune-boosting attributes in medicine-style formats. At the same time, substantial purchases of traditional food and beverage products such as juice, wine and tea indicate that consumers also value incorporating functional benefits into daily consumption.

With regard to consumption frequency, six levels were captured by a Likert scale (see Section 2). Figure 3 shows the distribution of consumption frequency among the 341 purchasers. Nearly 40% of respondents consumed elderberry products once or more than once a month, while over 60% consumed them three to ten times per year or less. This suggests that while many consumers engage with elderberry products on a regular basis, a substantial share does so only occasionally.

As described in the data section, the survey includes twelve variables capturing respondents’ attitudes toward elderberry products and perceptions of specific product attributes. Figure 4 shows the distribution of responses for each variable on a 5-point Likert scale. Nutrition and antioxidants were considered very or extremely important, with 87% of respondents rating them at levels 4 or 5. Taste was the next most valued attribute, with 82% of respondents rating it very or extremely important. More generally, consumers tended to agree or strongly agree (or to rate as very or extremely important) with statements related to knowledge of food-related health benefits, awareness of elderberry’s vitamin and mineral content, overall health consciousness and nutrition, highlighting a strongly health-motivated consumer base. By contrast, extrinsic attributes such as environmental impact, origin, brand and packaging were rated relatively less important, with fewer respondents assigning them a 4 or 5 level on the Likert scale.

Given the multiplicity of attitudes and perceptions, the first step was to consolidate them using factor analysis. As established in the Methodology, the suitability of the data for factor analysis was confirmed: the KMO measure of sampling adequacy was 0.80, classifying the data as “meritorious” according to conventional standards (Watkins, 2018). Bartlett’s test of sphericity was significant (χ²[66] = 2212.36, p = 0.000), indicating that the variables are sufficiently intercorrelated and suitable for factoring. The PCF extraction method and eigenvalue >1 retention rule identified four factors, which were then subjected to varimax rotation to enhance interpretability. Together, these four factors accounted for 66.95% of the common variance. Factor loading scores are presented in Table 2.

Considering loadings above 0.6, four factors were identified: (1) “sustainability and identity orientation”, with high loadings on statements related to branding, packaging, origin and environment; (2) “health awareness”, with high loadings on statements related to overall health; (3) “nutritional awareness”, with high loadings on statements related to nutrition and vitamins; and (4) “taste-price sensitivity”, with high loadings on statements related to taste and price.

The internal consistency of items loading highly (≥0.6) on each factor was assessed using Cronbach’s alpha (Cronbach, 1951). The first three factors demonstrated good internal consistency with alpha values ranging from 0.78 to 0.79, all above the conventional threshold of 0.6. The fourth factor (“taste-price sensitivity”) had a lower but still acceptable alpha of 0.48, common for factors comprised of only two items (Taber, 2018; Ekolu and Quainoo, 2019). The overall reliability for the entire 12-item scale was 0.80.

After consolidating the attitudinal variables into four factors, a non-hierarchical K-means cluster analysis was conducted to identify consumer groups. The Calinski-Harabasz pseudo-F index was used to identify the optimal number of clusters (Table 3).

The pseudo-F index increased with the addition of more clusters but peaked at five before declining. Hence, five clusters or consumer segments were selected (see Table 4), comprising 16.8%, 21.8%, 19.1%, 25.6% and 16.8% of respondents, respectively.

To characterize each segment, we compared the factor mean scores across the five clusters. These scores reflect the relative standing of each cluster on the identified factors, as shown in Table 4.

From the information in Table 4, we classify consumers as follows:

1. Cluster 1: Respondents show the lowest concern for health awareness among all clusters, with slightly below-average or near-average scores on other factors. We label this cluster “health-indifferent consumers.”

2. Cluster 2: Respondents show the highest concern for sustainability and identity orientation, along with above-average interest in health, nutrition and taste-price factors. We name this cluster “socially conscious consumers.”

3. Cluster 3: Respondents are primarily motivated by health awareness. They score the highest on health awareness while showing relatively low concern for taste and price sensitivity, as well as low interest in sustainability and nutritional content. We label this group “health-motivated consumers.”

4. Cluster 4: Respondents show the lowest concern for sustainability and identity while displaying moderate concern for health, nutrition and taste-price factors. They prioritize personal benefits (health, taste and price) over social or environmental issues. We name this cluster “conventional consumers.”

5. Cluster 5: Respondents show the lowest concern for nutritional awareness, low concern for sustainability and identity, moderate concern for health benefits, and the highest concern for taste and price among all segments. We name this cluster “taste- and value-seekers.”

To understand how respondents’ socioeconomic characteristics vary by segment, statistical tests were conducted (ANOVA for the continuous variable and chi-squared tests for categorical variables) to compare cluster means and assess significant differences across clusters. The results show that 9 of the 17 socioeconomic traits differ significantly across the five clusters (at varying levels of significance), as shown in Table 5, implying that the clusters exhibit distinct profiles.

These differences are reflected in socioeconomic patterns across the segments. Socially conscious and health-motivated consumers are the most engaged with the elderberry market, exhibiting the highest purchase rates and consumption frequency, as well as the largest shares of high income and postgraduate respondents. In contrast, health-indifferent consumers are the least likely to purchase elderberry products and present the lowest consumption frequency, as well as the lowest shares of postgraduates, high income-earners and older adults (over 54). Taste- and value-seekers stand out for having the highest share of older consumers (over 54) and the lowest share of males. Demographic differences also appear in racial composition: conventional consumers are predominantly White, while the socially conscious cluster has the largest share of Black respondents.

We used a double-hurdle model to analyze the purchasing decision and the frequency of consumption of elderberry products. As explained in the methodology section, the first hurdle involves the estimation of a logistic regression, where the dependent variable is a binary indicator of whether the respondent purchased any elderberry product or not. Independent variables include the socioeconomic variables from Table 1, plus dummy variables for four of the five clusters, leaving cluster 1—health-indifferent consumers—as the reference category.

The results of the logistic regression are presented in Table 6. The model shows good fit (p < 0.001; Pseudo R² = 0.098). In interpreting the results of the logistic regression, we focus on the column reporting average marginal effects, as these provide a more intuitive measure of the change in probability of purchasing elderberry products for a typical respondent.

Education emerges as a key factor: respondents with an associate’s degree are 16.7 percentage points more likely to purchase elderberry products than those with only a high school degree. Those with a bachelor’s degree are 24.4 percentage points more likely, and respondents with a postgraduate degree are 23.3 percentage points more likely. This suggests that more educated consumers may have greater health awareness or familiarity with functional foods. Age also plays a significant role. Compared to respondents under 35, those aged 35–54 are 18.0 percentage points less likely to purchase elderberry products and those over 54 are 15.8 percentage points less likely. This result is in line with Mohebalian et al. (2013), who found a similar pattern, and may reflect established consumption habits among older consumers that reduce their willingness to try new products.

Household composition matters as well: having children increases the probability of purchase by 5.1 percentage points, suggesting that families might be more inclined to seek out products perceived as healthy or immune-boosting. Finally, consumer attitudes captured by the segments are strongly associated with the purchase likelihood. Membership in the socially conscious cluster increases purchase probability by 18.6 percentage points, and being in the health-motivated cluster increases it by 12.6 percentage points, compared to the health-indifferent segment. Other variables, including gender, income, marital status, race and distance, were not statistically significant predictors. Overall, these results highlight that education, age, household composition and consumer attitudes are the primary drivers of the likelihood of purchasing elderberry products.

The second hurdle examines the frequency of elderberry consumption. The Brant test indicates that the proportional odds assumption is not violated (⁠$χ2$ = 85.55, p = 0.131), confirming that the ordered logistic regression is appropriate. The model shows good overall fit (p < 0.001; Pseudo R² = 0.051), and the results are presented in Table 7.

The results reveal clear drivers of elderberry consumption frequency. All consumer segments show significantly higher odds of frequent consumption compared to the health-indifferent reference group. The socially conscious and health-motivated segments are particularly strong predictors, with odds ratios of 4.07 and 3.83, respectively, indicating that these consumers are about four times more likely to report higher consumption frequency. The conventional and taste- and value-seeker segments also show significantly higher odds, with odds ratios of 2.48 and 2.39, respectively.

Among socioeconomic characteristics, Black consumers are about three times more likely than the “other race” baseline to be frequent consumers. Higher income consistently predicts more frequent consumption: consumers earning $35,000–$75,000 and those earning over $75,000 are about 2.5 times more likely to be high-frequency consumers compared to low-income respondents. In contrast, postgraduate education is associated with significantly lower odds of frequent consumption (odds ratio = 0.37), indicating that these individuals are roughly 63% less likely than those with a high school education to report higher consumption frequency. Other variables, including gender, age, number of children, marital status and distance to the nearest urban center, were not significantly associated with consumption frequency.

A more intuitive interpretation is obtained from the average marginal effects in Table 8, which illustrate how each factor shifts probability mass across the consumption spectrum. For instance, socially conscious consumers are 15.6 percentage points less likely to fall in the lowest frequency category and 19.5 percentage points more likely to be in the highest—representing the largest shift among all segments and underscoring their role as the core consumer base. Similarly, Black consumers exhibit a strong tendency toward regular use, being 11.7 percentage points less likely to be low-frequency consumers and 17.4 percentage points more likely to be high-frequency users. This pattern may reflect enduring cultural traditions of using natural or plant-based remedies within African diasporic communities (Gallego, 2019).

The relationship with education is more complex. Postgraduate consumers are 15.5 percentage points more likely to belong to the lowest frequency group and 9.5 percentage points less likely to be in the highest, suggesting a gap between purchase intention and sustained use. This contrast with the first hurdle results—where higher education predicted purchase—may indicate that, while highly educated consumers are open to trying elderberry products, their greater skepticism or critical evaluation may hinder regular use, particularly if their consumption experience does not align with expectations about taste or efficacy (Grunert, 2010).

Finally, income effects show a roughly symmetric shift across the consumption spectrum: middle- and high-income consumers are about 11–12 percentage points less likely to be in the lowest frequency category and approximately the same amount more likely to be in the highest frequency level. Meanwhile, the conventional and taste- and value-seeking clusters show positive but more moderate shifts toward regular consumption, while health-inclined, taste-price indifferent consumers remain among the more engaged segments overall.

This study set out to answer how consumer segments, defined by their attitudes, along with socioeconomic characteristics, influence the likelihood of purchasing elderberry products and the frequency of their consumption. By integrating attitudinal segmentation with a double-hurdle econometric framework, the analysis provides a multi-layered understanding of consumer behavior, confirming and extending the existing literature on elderberry and functional food consumption.

Regarding consumer segments, the findings advance the elderberry-specific literature. While Mohebalian et al. (2012) established a basic dichotomy of health-conscious consumers, factor and cluster analysis reveals a more complex, five-segment typology. This nuanced segmentation addresses limitations of prior research that relied on broad categorizations and product-specific analyses. The emergence of distinct segments such as socially conscious consumers (valuing sustainability and identity) and taste- and value-seekers demonstrates that motivations in the modern elderberry market extend beyond health considerations alone. This aligns with the broader consumer segmentation literature (e.g., Schäufele-Elbers and Janssen, 2023; Hempel, 2024), which highlights the multi-faceted nature of consumer attitudes.

The double-hurdle model not only links attitudinal profiles to actual behavior but also highlights differences between market entry and consumption intensity. For the initial purchase decision, only the socially conscious and health-motivated segments had a significant effect. This is broadly consistent with Cai et al. (2024), who found that premiums for attributes such as “organic” and “American-grown” influence purchasing behavior. The results here suggest that the socially conscious segment—identified as valuing sustainability and identity—may be particularly responsive to such attributes, indicating convergence between their stated preferences and the motivations captured by attitudinal segmentation. For consumption frequency, all non-baseline segments contributed significantly, indicating that while specific values may be necessary to induce trial, a broader range of motivations supports habitual use. These results help bridge the “attitude-behavior gap” noted in segmentation literature (e.g., Li and Roe, 2024), showing that the influence of attitudes varies across different stages of consumption.

Socioeconomic characteristics reveal additional heterogeneity. Younger consumers are more likely to make initial purchases, consistent with early adoption patterns for emerging functional foods (Mohebalian et al., 2012; Baker et al., 2022). The strong, consistent effect of higher income on both purchase and consumption frequency underscores the role of economic access, highlighting that financial resources facilitate participation in this emerging market (Verbeke, 2005). The most striking demographic result is the consistently high consumption frequency among Black consumers, which may be contextualized by cultural traditions of using natural remedies (Gallego, 2019). This suggests that elderberry producers may find a particularly loyal consumer base among this demographic.

Education presents a dual character. Specifically, postgraduate consumers are more likely to purchase, yet more likely to remain low-frequency consumers, suggesting a gap between trial and sustained consumption. This aligns with Baker et al. (2022), whose literature review on determinants of consumer acceptance toward functional foods notes that education can have mixed effects on consumption, with some studies finding higher education positively associated with use and others negatively. The two-stage model provides insight into this pattern: while higher education may increase health awareness and drive initial purchases, the critical evaluation skills associated with postgraduate education may lead to more discerning and less frequent consumption, consistent with theories that perceived quality evolves with experience (Grunert, 2010).

These findings offer a strategic roadmap for stakeholders in the elderberry industry, highlighting differences between factors influencing customer acquisition and those affecting retention. For acquisition, younger consumers (under 35) and families with children appear particularly responsive, with ethical production resonating with socially conscious consumers and immune-boosting benefits appealing to health-motivated consumers. Patterns of regular use suggest that addressing a range of consumer preferences may be beneficial—for example, premium brand-aligned products for socially conscious consumers, vitamin supplements for health-inclined consumers, and affordable, flavorful options for taste- and value-seekers. Communication that provides clear, science-backed information may help mitigate skepticism among highly educated individuals.

Although the study is based on US survey data, the methodological approach of integrating segmentation with a double-hurdle model and the observed distinction between drivers of purchase and consumption offer a useful framework for research on emerging functional foods in other markets. Context-specific studies would be needed to determine whether these patterns hold in other regions such as the UK or Europe.

This study examined how consumer segments defined by attitudes, along with socioeconomic characteristics, influence both the likelihood of purchasing elderberry products and the frequency of their consumption. By integrating attitudinal segmentation with a double-hurdle econometric framework, the findings provide a multi-layered understanding of these behaviors, offering both confirmation and meaningful extensions to the literature on elderberry and functional food consumer behavior.

Five consumer segments emerged: health-indifferent consumers, socially conscious consumers, health-motivated consumers, conventional consumers, and taste- and value-seekers. Socially conscious and health-motivated consumers exhibited the highest likelihood of purchase, while all non-baseline segments showed higher consumption frequency compared to health-indifferent consumers. Across segments, age, income, education, race and family composition also influenced purchasing behavior, suggesting demographic factors that may shape market outreach for elderberry products. These results highlight the importance of distinguishing between drivers of initial purchase and those that sustain regular consumption, providing a framework for dual marketing strategies—targeting socially and health-inclined consumers for market entry, and addressing a broader set of motivations and product formats for market retention.

Several limitations should be noted. First, the analysis relied on a non-probability sample without demographic or geographic quotas or post-stratification weights. Therefore, while the sample is national in scope, it is not statistically representative of the entire US population. Moreover, participants in online surveys may systematically differ from the general public in terms of online literacy or interest in functional foods. Second, all data—including purchasing behavior and consumption frequency—were self-reported, which may introduce recall bias or social desirability bias (e.g., by over-reporting consumption of products perceived as healthy or environmentally responsible). Third, the purchasing decision and frequency of consumption were captured at an aggregate level (elderberry products overall) rather than for specific product types. Although this approach allows for assessment of overall demand, it limits the ability to detect nuanced patterns for individual products such as wines, juices, or vitamin supplements.

Future research could explore the underlying motivations behind consumer preferences, particularly regarding emerging trends such as sustainability and environmental values. Longitudinal studies would be valuable for tracking how consumer attitudes and consumption patterns evolve over time, especially as the elderberry market grows. Additionally, research could assess whether similar consumer segment patterns emerge in other markets or cultural contexts, such as Europe or the UK. Such insights may support more effective marketing strategies, product development and policy decisions, helping businesses and policymakers understand shifting consumer dynamics and plan for future market opportunities.