What’s Hoppin’ in Hoptown?: Examining Fruit and Vegetable Consumption Among Hopkinsville’s African American Community

Abstract

Cardiovascular disease is a concern nationwide and disproportionately impacts African Americans residing in the American South. However, this condition can be assuaged by consuming a diet of fruits and vegetables. Utilizing the social ecological model, this study explored the community, interpersonal, and intrapersonal factors that predict fruit and vegetable consumption in Hopkinsville, Kentucky, a small rural city which possesses one of the largest populations of African Americans in the state. Using data from social transect walks and a formative research survey (N=174), this study also sought to discover the best methods to communicate with Hopkinsvillians to improve their diets. Results show that despite some barriers (i.e., cost of quality foods, perceived lack of access to fresh food stores), Hopkinsvillians tend to view fruit and vegetable consumption positively. In addition, family and friends provide interpersonal support to those wanting to eat healthier. The study reveals that communicators must consider all levels of the social ecological model to produce effective health messages.

Obesity is associated with a greater incidence and prevalence of a host of chronic health conditions, such as cardiovascular disease (see Go et al., 2014 for a comprehensive review). Health conditions like obesity and cardiovascular disease are often exacerbated by geographic factors. For example, higher poverty levels and overall lack of resources have been linked to adverse health outcomes in rural areas (Eberhardt & Pamuk, 2004; Hartley, 2004). The National Rural Health Association (2007–14) explains that the health impacts of rural living are even more extreme for people of color. Within rural communities, African Americans are significantly more likely to suffer from heart disease and other chronic health conditions than White Americans within the same communities, as well as other African Americans in urban and non-Southern rural areas (Eberhardt & Pamuk, 2004; Kulshreshtha et al., 2014; Lower Mississippi Delta Nutrition Intervention Research Consortium, 2004; Probst et al., 2004; Taylor et al., 2002). Kentucky follows the national trend with regard to diseases of the heart. Specifically, the crude death rate in the Pennyrile Area Development District, which is a predominately rural area of Kentucky, is 23.63 million, versus 19.74 million in the Kentuckiana Region, which includes the Louisville Metropolitan area (Kentucky Cabinet for Health and Family Services, 2005).

Previous research has highlighted the importance of good dietary practices in reducing the likelihood of adverse health outcomes (Boeing et al., 2012; Estruch et al., 2013; Taylor et al., 2011). Consuming high levels of fruit and vegetables is linked to significant reductions in conditions such as obesity and heart disease (Dubowitz et al., 2007; Ettienne-Gittens et al., 2013). Although reports show that most U.S. adults do not consume an adequate amount of fruit and vegetables (Gary et al., 2004), African Americans and groups of lower socioeconomic status are less likely to consume a fruit and vegetable-rich diet (Casagrande et al., 2007, Ettienne-Gittens et al., 2013). The current minimum recommended daily allowances for adults range from 2.5 to 3 cups of vegetables and 1.5 to 2 cups of fruit, depending on age, sex and physical activity (United States Department of Agriculture, 2010). Rural residents also may be more susceptible to consuming fruit and vegetable-deprived diets, as access to an affordable variety of fruit and vegetables may be an issue (Krebs-Smith & Kantor, 2001).

There has been a recent call in Kentucky for a greater focus on initiatives and programs aimed at preventing heart disease in the state (Rugg, Bailey, & Browning, 2008); these plans often focus on modifying the diets of African American Kentuckians to include more fruit and vegetables. Health research has focused on the African American population of urban centers in Kentucky (e.g., Mayor’s Healthy Hometown Movement, Food in Neighborhoods Committee, 2010), but there is little information about how geographic isolation, historical and cultural differences, and community interactions influence African American diet in smaller Kentucky towns. As such, this study was conducted in Hopkinsville, a small rural city in Christian County, Kentucky that is part of the Pennyrile Area Development District. The specific goals of the study were to: 1) understand the factors that enabled or prevented members of Hopkinsville’s African American community to engage in preventive health behavior such as consuming fruits and vegetables and 2) discover the best ways to communicate with members of the community to improve their diets. To explore these issues the social ecological model was utilized to guide the investigation.

Theory: Social Ecological Model

The social ecological model is an appropriate theoretical framework for this study because it posits that a systems-level perspective can help identify ways in which the environment enables or inhibits healthy practices (Glanz, Rimer, Viswanath, 2008). This model requires health communicators to explore and understand the unique communicative dynamics of particular environments. Identifying these dynamics, including intrapersonal and interpersonal factors, as well as institutional and community factors, helps communicators understand the levels at which individuals communicate within themselves and with others regarding fruit and vegetable consumption. Also of interest to health communication researchers is understanding the media and channels with which community members prefer to interact when receiving messages about diet. For instance, Fleury and Lee (2006) indicate that communication targeting communities must respect communities’ social norms and include “culturally relevant social support” (p. 133) at the interpersonal level. Given the lack of research focused on African Americans living in more rural areas of Kentucky, the researchers were concerned with identifying message components that would be culturally relevant for an African American audience in Hopkinsville, Kentucky.

Ecological models that guide health researchers suggest that, beyond attention to individual factors, health professionals should give greater attention to environmental factors when attempting to help individuals modify unhealthy behavior. According to Stokols (1996), “the congruence between people and their surroundings [can be] an important predictor of wellbeing” (p.286). The main premise behind ecological models is that a more inclusive environmental approach to health behavior change can reveal behavioral antecedents unique to specific cultures, geographies, and settings. An ecological perspective dictates that researchers assess behavioral antecedents at several different levels of the social system. The first to formally state the constructs of the social ecological model were McLeroy, Bibeau, Steckler, and Glanz (1988), who explain that it has five factors: individual characteristics, social networks, including those of friends/family and work associates, institutional factors, community factors within specific boundaries, and public policies on the national, state, and local level.

Since its inception, the model has been utilized to understand the structure of various contexts and develop plans tailored for these particular contexts. Gregson et al. (2001) applied the social ecological model to organize and evaluate a nutrition education program for those of lower socio-economic means. In another instance, Fleury and Lee (2006) used the model to analyze the factors regarding the relationship between African American women and physical activity. They found that intrapersonal factors, such as the ability to undertake a program and one’s perception of her health, as well as interpersonal factors, such as having support and encouragement from social networks, including the church, impacted African American women’s physical activity. In addition, community and environmental resources (e.g., including community members in developing community campaigns and programs, as well as ensuring the safety and security of community members so they will want to participate in these initiatives), in addition to organizational resources (e.g., work place policies that encourage exercise) all impacted the participants’ attempts to be more physically active.

Hopkinsville, Kentucky: An Introduction

Due to the dearth of research on African American populations outside of urban centers in Kentucky, particularly in this area of focus, this study was conducted in Hopkinsville, Kentucky. Hopkinsville possesses the second largest population of African Americans in the state, but differs greatly from Louisville, Kentucky’s major urban center and the city with the largest African American population. As such, it is important to assess the community- and personal-factors among African American Hopkinsville residents that might influence message design concerning fruit and vegetable consumption. At the community-level, Hopkinsville is a small rural city, with almost 33,000 residents, surrounded by farmland. Additionally, nearly one-quarter of Hopkinsvillians (25%) live below the poverty level. (See Table 1 for demographic information for the city and Kentucky as a whole.) This geographic isolation and financial disadvantage is of import to health communicators because these factors may pose unique barriers to African American residents attempting to access and consume fruits and vegetables in Hopkinsville.

Table 1.

Hopkinsville and Kentucky Demographic Information

	Hopkinsville	Kentucky
2012 Population Estimate	32,966	4,379,730
Population of White Community	62.6%	87.8%
Population of African American Community	31.9%	7.8%
High school graduate or higher, percent of persons age 25+, 2008–2012	83.3%	82.4%
Bachelor’s degree or higher, percent of persons age 25+, 2008–2012	16.3%	21.0%
Households, 2008–2012	13,043	1,691,716
Persons per household, 2008–2012	2.32	2.49
Median household income, 2008–2012	$34,664	$42,610
Persons below poverty level, percent, 2008–2012	25.0%	18.6%

Source: United States Census Bureau (2014)

At the community- and intrapersonal-levels, Hopkinsville’s residents’ are less likely to graduate from college than the average Kentuckian (12.5% graduation rate in Christian County versus 17.1% state-wide) (Kentucky Council on Post-Secondary Education, 2006), which likely stifles opportunities for general health knowledge acquisition. This limited access to general knowledge might also hinder residents from recognizing instances in which they do have personal agency in changing and maintaining healthy diets. Given these realities, it is imperative for health communicators to understand how community-level factors interact with intrapersonal factors to enable or inhibit dietary changes so that health communication messaging can be tailored to residents’ real and perceived supports and challenges for increasing fruit and vegetable consumption.

Finally, at the interpersonal level, the social structure of Hopkinsville may influence interactions among residents of the same race and class and between residents of varied race and class. For instance, Hopkinsville, as a remnant of its segregated past, still possesses neighborhood “pockets” where African Americans reside in homes once occupied by their ancestors. Although these same neighborhoods were once home to African American schools and more than 200 successful African American businesses (Alleyne, 2013), today they are some of the most impoverished neighborhoods in Hopkinsville. Nevertheless, as the city has grown westward, newer, more culturally diverse neighborhoods have developed where African Americans live, shop, and play alongside Hopkinsville residents of other races and ethnicities. Not only are these newer neighborhoods (e.g., the western areas of Indian Hills and Millbrooke) home to people of different cultural backgrounds, but they are also predominantly middle class. As such, there may be differences in relational communication processes across various areas and neighborhoods of Hopkinsville that need to be addressed or considered during the development of behavioral modification messaging.

Given the potential for interplay between different levels of the social ecological framework in the study context, the team focused on the intrapersonal (knowledge, attitudes, skills, self-efficacy), interpersonal (social support, family influence, peer influence), and group/community (neighborhood organizations, institutional affiliations) levels of assessment:

RQ1: What community, interpersonal and intrapersonal factors influence fruit and vegetable consumption for African Americans in Hopkinsville?

RQ2: What community, interpersonal and intrapersonal factors should health communicators consider in message design and media selection?

Method

Data Collection

This study was part of a larger multidisciplinary project in which mixed methods were used to gather information about intrapersonal, interpersonal, and environmental factors influencing fruit and vegetable consumption in both West Louisville and Hopkinsville, Kentucky. For the analyses discussed here, however, the team focused exclusively on Hopkinsville in an effort to gain additional insight into the culture and lifestyle of a smaller population in Kentucky. In Hopkinsville, the team conducted a sequence of qualitative research (i.e., social transect walks), followed by quantitative research (a formative paper and pencil survey).

Social Transect Walks

Specifically, data collection began with informal meetings with various members of the Hopkinsville community. These community members were friends or relatives of individuals in the Department of Communication and/or the Department of Pan-African Studies at a metropolitan, research intensive university located in Kentucky. Department contacts made the initial introduction between community and research team members, helping to establish a level of comfort and rapport that would have been difficult to replicate otherwise. These community members initially provided a verbal description of Hopkinsville and answered informal questions from the team. These answers helped frame the role of community and food-related organizations in Hopkinsville residents’ fruit and vegetable purchase and consumption behavior.

As these informal meetings were taking place, members of the research team also engaged in social transect walks of Hopkinsville. Social transect walks are a highly participatory observational tool that facilitate direct researcher experience in a community. They are often used in sociological research to enhance one’s knowledge of a locale. They have the added advantage of being easy to facilitate in low-literacy communities because they are primarily observational and interactional (i.e., no reading and writing are required by participants). In essence, these walks are a time during which researchers pay special attention to elements of and interactions among the environment, people, resources, and community activities and organizations (Barton, Borrini-Feyerabend, de Sherbinin, & Warren, 1997).

The team conducted two different transect walks in Hopkinsville in September and November 2012. One was guided by a local resident; the other was conducted solely by research team members. The local resident provided feedback and answered questions as the researchers observed and interacted with each area. These walks and community discussions were focused on assessing purchase outlets for fruit and vegetables in each area. However, they also provided background on the general history and economics of the African American populations living in the area.

During each walk, members of the research team audio recorded their thoughts and observations. In total, the team collected a total of 2 hours and 57 minutes of observational data. The team also took digital pictures of important elements in the environment that might significantly influence fruit and vegetable consumption. For example, pictures were taken of fruit and vegetable displays at farmers’ markets and grocery stores. Finally, each team member who participated in a walk independently typed up her observations and shared these notes with others on the team. These notes and observations were instrumental in developing some of the items for the formative research survey, as discussed below.

Formative Research Survey

Information from the transect walks and informal meetings with Hopkinsville community members helped guide the development of a formative research survey. Formative research can include qualitative and quantitative research designed to help health communicators create effective message strategies and select appropriate communication media. For this study, the researchers referred to the survey as a formative research survey because it was designed to identify prominent determinants of fruit and vegetable consumption that should be leveraged during the message creation process and to assess communication channel preferences among African American Hopkinsville residents. Data from this survey were collected from October 2012 through November 2012.

Procedures.

The team began data collection using the extended social networks of contacts in Hopkinsville. Although this snowball sampling approach yielded a nonprobability sample of local residents, it also helped the research team reach individuals who may have been reticent to talk with outsiders had the team used some other probability-type sampling approach (e.g., a random digit dial within study zip codes). Using this sampling methodology the team collected data at local churches and barbershops. Before administering the survey, the team screened for individuals who:

• Self- identified as African American

• Were overweight or obese according to the National Institutes of Health body mass (BMI) index chart (National Heart, Lung and Blood Institute, n.d.)

• Self-identified as being over 18 years of age

• Were able to speak and read English

• Did not reside on a military base

• Lived in a Hopkinsville or Pennyrile zip code

In total the team collected 174 data points from Hopkinsville residents. Participants were offered a $20 incentive for their participation.

Sample.

Two-thirds of the data were collected from women (68.8%) and one third from men (31.2%). The mean age was 40.6 years. On average, there were between one and two children living at home. The sample was fairly evenly split across individuals who were married (29.5%), divorced (22%) and single/never married (35.3%). The remainder of the sample either reported that they were widowed, separated, or a member of an unmarried couple living together. Only two out of ten respondents reported completing an undergraduate degree (19.1%). Thirty-seven percent completed only high school and another 30.1% said that they had completed between one and three years of college or technical school. Based on this educational distribution, it should not be surprising that the average income was between $15,000 and $25,000. However, the team did note that income was bi-modal in this sample. That is, 38.2% of the sample reported making less than $15,000 annually, while 30.9% reported making $40,000 or more. Thirty percent reported receiving some sort of food assistance (e.g., food stamps; Women, Infants and Children [WIC] support). Twenty-nine percent reported that they did not have health insurance, while 20.1% and 5.2% reported receiving insurance benefits through Medicaid and Medicare, respectively.

Survey design and measures.

Regarding RQ1, survey questions were designed to elicit a greater understanding of the antecedents for fruit and vegetable consumption behavior at various levels of the environment, as suggested by the social ecological framework. Specifically, the team queried respondents about factors that might serve as barriers or enablers for fruit and vegetable consumption at the community level (4 items about physical access). These questions were greatly influenced by the social transect walks, which asked respondents to rate the following items on a 5-point Likert scale (strongly disagree to strongly agree): “It is hard for me to eat more fruit and vegetables because they are too expensive.”; “It is hard for me to find good quality fruit and vegetables in my neighborhood.”; “When I eat out, it is easy for me to eat fruit and vegetables.”; and “I would eat more fruit and vegetables if I could afford it.”

At the interpersonal and intrapersonal levels, the survey included 7 items that measured social support and 4 items that measured social sabotage, as well as 6 items that measured self-efficacy, 5 that measured perceived psychosocial access/barriers, and 10 that measured decisional balance. All scales were assessed using 5-point Likert response options. Table 2 below provides additional detail regarding the scales employed in this survey. In general, internal reliability was appropriate (i.e, α= 0.75 or higher) for conducting additional analyses.

Table 2.

Measurement Information

Level of Measurement	Scale	Internal Reliability
Community	Physical access (Erinosho, Moser, Oh, Nebeling & Yaroch, 2012)	n/a
	Communication media preference	n/a
	Communication source preference	n/a
Interpersonal	Social support (Baranowski et al., 2006)	α = .903
	Social sabotage (Sallis, Grossman, Pinski, Patterson & Nader, 1997)	α = .756
Interpersonal	Self-efficacy (Erinosho, Moser, Oh, Nebeling & Yaroch, 2012)	α = .880
	Perceived psychosocial access (Erinosho, Moser, Oh, Nebeling & Yaroch, 2012)	α = .874
	Decisional balance - pros (Ling & Horwath, 2001)	α = .778
	Decisional balance - cons (Ling & Horwath, 2001)	α = .791

Finally, the outcome variable was fruit and vegetable consumption. The team measured this variable using Wolf and colleagues’ (2008) three self-report items. These items asked about the number of servings of fruit, vegetables and potatoes consumed yesterday (e.g., “Think about yesterday. How many servings of fruit did you eat? An example of a serving of fruit is a medium apple, a small glass of fruit juice, or a handful of cut fruit.”). Although self-report measures are not as robust as daily food diaries, they are frequently used for quantitative data collection because of their time efficiency. From these questions, the team calculated a total 24-hour recall measure of fruit and vegetable consumption by computing a composite sum of participants’ responses to these three questions. The team also asked a series of questions designed by Block and associates (2004) to assess motivational readiness to increase fruit and vegetable consumption.

Communication campaign strategies measures.

Because the team was interested in both determinants of fruit and vegetables consumption and ways in which the team might influence Hopkinsville residents to eat more of these foods, the team designed several questions to measure communication preferences. Specifically, the team asked a series of questions about where and how respondents would prefer to receive communication promoting fruit and vegetable consumption (e.g., “Where are you most likely to learn about new foods or new ways to prepare meals”; “Do you have access to the Internet on your phone”; “Who do you trust to tell you the truth about how easy or hard a food is to prepare and cook?”). These questions were designed specifically for this survey and had not previously been validated. Still, they lend insight into the kinds of communication message strategies that might prove effective for affecting positive behavior change in Hopkinsville.

Analyses.

To address RQ1, the team first assessed descriptive statistics for the measures employed in the study. Then, using the total fruit and vegetable consumption measure, the team computed a multiple regression analysis to identify significant predictors of consumption behavior. Fruit and vegetable consumption data were first cleaned for outliers (i.e., individuals were deleted who reported eating extremely high amounts of fruit and vegetables) and checked for normality. Once the outliers had been deleted, the data resembled the normal distribution. The team first ran a bivariate correlation matrix with the outcome variable and demographic variables that the team suspected might covary with the antecedent variables of interest (e.g., age, gender, income). Only gender and the number of adults in the household were significantly correlated with fruit and vegetable intake. A bivariate correlation matrix was constructed using the consumption determinants of interest (i.e., social support, barriers to access). Psychosocial access/barriers and the items “It is hard for me to eat more fruit and vegetables because they are too expensive” and “It is hard for me to find good quality fruit and vegetables in my neighborhood” were significantly related to fruit and vegetable intake (see Table 3).

Table 3.

Correlation Matrix

	Total F&V	Gender	# of adults in HH	It is hard for me to eat more F&Vs because they are too expensive	It is hard for me to find good quality F&Vs in my neighborhood	PsychoSocial Barriers
Total F&V	1
Gender	−.176* (p = .028)	1
# of adults in HH	−.200* (p = .012)	−.083* (p = .278)	1
It is hard for me to eat more F&Vs because they are too expensive	−.262** (p= .001)	−.014* (p = .854)	−.094* (p = .224)	1
It is hard for me to find good quality F&Vs in my neighborhood	−.165* (p = .040)	−.050* (p = .514)	−.025* (p = .747)	−.681* (p = .000)	1
PsychoSocial Barriers	−.213** (p = .008)	−.127* (p = .096)	−.143* (p = .060)	−.602* (p = .000)	−.598* (p = .000)	1

^*.Correlation is significant at the 0.05 level (2-tailed).

^**.Correlation is significant at the 0.01 level (2-tailed).

Next, the enter method in IBM SPSS Statistics 21 was utilized to perform a multiple regression on the fruit and vegetable consumption variable. This method allows for a particular block of independent variables to be entered together into a regression model, all in one step. In the first block, potential covariates were entered into the analysis (i.e., gender and number of adults in the home). In the second block, the consumption determinants of interest were entered (i.e., psychosocial barriers and the items stating that quality and expense are an issue).

Finally, to address RQ2, the team reviewed participants’ responses regarding communication preferences. The team looked at descriptive frequencies and measures of central tendency for communication design measures. From these descriptive statistics the team provides an overview of media and message options for future health communication campaigns to increase fruit and vegetable consumption among African American residents of Hopkinsville, KY.

Results

Social Transect Walks Results

Hopkinsville boasted numerous venues and access points for purchasing fruit and vegetables. Within the city limits there were nine grocery stores. Additionally, the researchers noted a designated location for a farmers’ market in the middle of downtown Hopkinsville. For the most part, the fruit and vegetables available looked to be ripe and fresh. Additionally, the prices in Hopkinsville were comparable to grocery stores in larger metropolitan areas (e.g., Louisville). It was noted, however, that the two grocery stores located closer (less than a mile) to the historically African American section of town had smaller produce sections with less appealing variety than some of the other grocery stores in town. Kroger was the nicest and newest store, but it was two miles from the economically challenged, historically African American area of Hopkinsville. Hopkinsville is not serviced by public transportation, so access to a vehicle is required to shop at stores that are not within walking distance. Nevertheless, the downtown farmers’market is located less than a mile from the aforementioned neighborhood and could be accessed on foot. During the transect walk the research team also learned that local residents frequent Mennonite stands by the side of the road, and that a second farmers’ market operates in the mall parking lot during the summer months.

Based on observations of Hopkinsville, it was concluded that several environmental enablers for fruit and vegetable consumption existed in the town. Also, there appeared to be a culture of buying local produce in Hopkinsville because local residents described frequenting roadside stands. The researchers attributed this local buying culture to the fact that Hopkinsville is surrounded by farmland, and many of its residents remember when they or their family members lived on farms.

Formative Research Survey Descriptive Results

Although the environmental scan indicated that physical access to fruit and vegetables was relatively abundant, it was found that African American residents in Hopkinsville reported consuming only 1.28 servings of fruit, 1.08 servings of potatoes, and 1.24 servings of vegetables in the past 24 hours. The median number of total fruit and vegetables servings was 3 (M = 3.57). This statistic is in line with other national data on fruit and vegetable consumption. That is, the Centers for Disease Control and Prevention’s Behavioral Risk Factor Surveillance System (BRFSS; 2014) indicates that Americans most frequently reported consuming between 3 and 4 servings of fruit and vegetables a day (41.6%) in 2009. In Kentucky, the largest portion of responses on the BRFSS question about fruit and vegetable consumption indicated that participants most frequently consumed 3 servings per day (Centers for Disease Control and Prevention, 2014). This number is below the recommended daily allowance, which calls for consumption of anywhere between 4 to 5 cups of fruit and vegetables a day, which roughly translates to between 5 and 11 servings a day, depending on one’s caloric intake needs (United States Department of Agriculture, 2010).

In general, study participants indicated that they were very interested in trying to eat more fruit and vegetables in the future. In fact, two-thirds of participants (68.6%) indicated that they were actively trying to eat more fruit and vegetables, which is promising for health communicators in Hopkinsville who might be trying to affect behavior change. Only 15.1% of respondents indicated that they were not even thinking about trying to increase their fruit and vegetable consumption behavior. Interestingly, when asked how many servings of fruit and vegetables a person should eat each day, the median response was 3. Comparing this perception with self-reported consumption, it was noted that participants were eating the number of servings they thought they should eat.

Looking more closely at predictor variables, participants reported physical barriers to access were at or below the midpoint of a 5-point Likert Scale. Averages ranged from 2.5 to 3.34. For the most part, respondents indicated that they would eat more fruit and vegetables if they could afford to do so (see Table 4).

Table 4.

Physical Barriers to Access

Item	Mean	Median
I would eat more F&Vs if I could afford it	3.13	3.00
It is hard for me to eat more F&Vs because they are too expensive	2.60	2.00
It is hard for me to find good quality F&Vs in my neighborhood	2.50	2.00
When I eat out, it is easy for me to eat F&Vs	3.34	4.00

At the interpersonal level, participants indicated that family and friends supported them in trying to eat more fruit and vegetables. The average score for familial social support was 3.34 and the average score for social sabotage was 2.11, indicating that participants did not perceive others as trying to derail their healthy eating attempts.

On the individual level, respondents indicated fairly strong confidence (M = 3.69) that they could consume fruit and vegetables across a variety of different tempting situations (e.g., when they were really hungry or when others were eating high-fat foods in front of them). In keeping with this confidence, respondents reported a positive decisional balance (M pros = 3.90 versus mean cons = 2.52) and fairly low levels of psychosocial barriers to consumption (M = 2.39).

Formative Research Survey Inferential Results

When the five potential predictors from the initial correlational analyses were entered into a multiple regression model, the model was significant and explained 12% of the variation in fruit and vegetable consumption (Adjusted R² = 9%). Interestingly, when entered together in the model, and after accounting for multicolliearity between predictors, only one of the five variables entered remained in the model. Gender was the only significant predictor of fruit and vegetable consumption at the p = .05 level (t = −2.215, p = .028). On average, men reported eating 3.8 servings of fruit and vegetables in the past 24 hours while women only reported eating 3.0 servings). The item “It is hard for me to eat more fruit and vegetables because they are too expensive” was directionally significant (t = −1.769, p = .079). Table 5 presents the final regression output and standardized regression coefficients for the variables entered.

Table 5.

Regression Analyses

Model	Unstandardized Coefficients		Standardized Coefficients	t	p
	B	Std. Error	ß
(Constant)	5.485	.992		5.527	.000
Gender	−.828	.374	−.180	−2.215	.028
It is hard for me to eat more F&Vs because they are too expensive	−.376	.212	−.209	−1.769	.079
PsychoSocial Barriers	−.328	.281	−.133	−1.167	.245
# of adults in HH	.282	.200	.114	1.408	.161
It is hard for me to find good quality F&Vs in my neighborhood	.149	.231	.077	.644	.521

Communication Campaign Strategies Results

Concerning RQ2, when asked about mediated communication preferences for food-related information, the vast majority of participants indicated that they preferred a written form of communication (67.4% requested information delivered via the postal system and 21.7% requested it delivered via email). Overall, participants indicated that they were most likely to learn about new foods or new ways to prepare meals from television (34.3%), a personal friend (27.4%), or a family member (26.3%). And, participants communicated that they would be most persuaded to buy or prepare a new food if someone they knew cooked it for them (46.9%). When asked directly about whether they use the internet to look for food-related information, less than half of the Hopkinsville participants indicated that they had ever done so (40.7%). And although seven of ten participants (69.8%) indicated that they had access to the internet via a home computer, tablet or e-reader and nearly two of every three participants (66.9%) indicated that they could access the internet from their cell phones, just over half of those with home internet access (54.9%) and less than half of those with mobile internet access (42.5%) had used their devices to obtain food-related information.

In thinking about sources of information, different types of individuals were designated as trustworthy depending on the type of knowledge being sought by the participant. Overall, however, one’s mother or grandmother appeared to be the first place participants would turn for advice. The overwhelming majority of participants said that they would consult their mother or grandmothers for an accurate assessment of how challenging a food is to prepare or cook (60.3%). A close personal friend (14.4%) and one’s sister (13.8%) were distant seconds to one’s mother or grandmother when it came to information about food preparation. Similarly, one’s mother or grandmother were also thought to be the best person to consult for the truth about how a food tastes (53.4%), and to a lesser extent a close personal friend was viewed as someone who might be a reliable source of opinion about how a food tastes (19.5%). In slight contrast, however, participants indicated they would seek information about nutrition not only from their mothers and grandmothers (38.9%), but also from their personal physicians (18.3%).

Discussion

The findings from the transect walks and the descriptive data from the formative survey revealed that even though there are some structural barriers in Hopkinsville that prevent residents from eating more fruits and vegetables, such as the cost and accessibility to fresh food outlets, residents are knowledgeable of various locations to purchase healthy foods. In addition, the participants generally regarded fruit and vegetable consumption positively, although their consumption levels were below recommended health standards. However, the participants were under the impression that they were getting the number of servings of fruits and vegetables they were supposed to be getting in their diets, and they did possess a desire to consume more of these foods. McLeroy et al.’s (1988) model recognizes that this general positive attitude would lend itself to campaigns that would encourage more behaviors in this vein. Not only did participants view eating fruit and vegetables positively, but they also reported that their goals to eat healthier were supported by those close to them. This support suggests that health communication campaigns should not only focus on the individual, but his/her social networks as well. This finding echoes Fleury and Lee’s (2006) argument, which stated that intrapersonal desire for positive, personal change is strongly impacted by interpersonal support. Focusing on Hopkinsville through the lens of the social ecological model suggests that an individual’s desire to eat fruits and vegetables is bolstered by interpersonal and community factors in this city.

An analysis of the inferential data from the formative survey indicates that gender is the only statistically significant factor when determining the likelihood of participants’ fruit and vegetable consumption. Past research has shown that African American women consume, on average, a higher daily intake of fruit and vegetables than African American men (Gary et al., 2004; McClelland, Demark-Wahnefried, Mustian, Cowan, & Campbell, 1988; Subar et al., 1995). Women are often viewed as “in charge” of household shopping, which can possibly translate into a more conscientious consumer of fruit and vegetables. However, having to manage the dietary (and other) responsibilities of a household can prove daunting, thus leading to the lower consumption of fruit and vegetables evident in this sample. For instance, in one study of fruit and vegetable consumption of toddlers, African American mothers with toddlers were significantly less likely to consume vegetables on a regular basis (Horodynski et al., 2010). A closer examination of relationships and responsibilities of women within households should be considered for future research.

To provide the possible perspective of the male participants, fruits and vegetables often require little preparation time. It is likely that men would turn to these foods as time efficient and simple alternatives, especially because men are not usually the household member responsible for meal preparation. This time efficiency might also explain why male participants consumed significantly more fruits and vegetables than their female counterparts. Therefore, although it seems obvious, it is worth stating that the development of any health communication campaign must consider how demographic factors such as gender impact one’s behaviors.

The formative survey data presents both interpersonal- and community-level sources that impact an individual’s willingness to accept new information about fruits and vegetables. On a macro-level, arguably larger than that of the community, the media have impact. Many of the participants did not have mobile internet access. Further, many of those who did have mobile internet access did not utilize their devices to gather food-related information, such as recipes. However, many of them found television to be a fruitful information source. Concerning interpersonal-level factors, personal friends and family members who could vouch for the tastiness of foods were invaluable. In particular, the participants trusted their mothers and grandmothers’ culinary repertoires and palates. Again, these findings, with their emphasis on the importance on interpersonal and community factors, lend further support to the model and echo findings presented by Fleury and Lee (2006) and Gregson and colleagues (2001).

Conclusion

One limitation of the study concerns its reliance on snowball sampling methods that produced a bi-modal sample with regard to income. More specifically, the researchers attempted to make contact with the African American community through members who had many contacts in their own personal networks. Because of basic human nature to socialize with others who are like themselves, people tended to refer those to the study who were similar to them in demographic characteristics, such as socioeconomic status and food assistance dependence (or lack thereof). As a result, the study’s findings might not be generalizable to all African Americans who live in South Central Kentucky, or even all that live in Hopkinsville. However, the mixed method study design included the triangulation of methods, resulting in both quantitative and qualitative data that supported each other. Further, this purposive sampling method remained faithful to the cultural element of this project by interacting with participants in areas that are important to them, including barbershops and churches. This method also served to maximize the comfort of the participants, who often expressed initial skepticism in participating in the study even when interacting with them face-to-face in comfortable local environments.

Another limitation is that, with the exception of the transect walks, the study produced data that were self-reported by the participants. Self-reported data can be a concern because there is often the tendency for this type of data to produce socially desirable responses. Despite this concern, the data produced by all four phases of the study did bolster each other.

Although the study certainly confirms the validity of the tenets of the social ecological model (McLeroy et al., 1988), it also speaks to practical applications for communication researchers concerned with designing campaigns for communities. Concerning the Hopkinsville, Kentucky-based data, the model suggests:

• Hopkinsville grocers should make a concerted effort to provide quality produce in areas accessible to all community members, including those that are within walking distance of the historically African American sections of the city. Recall that the social transect walks and the descriptive results of the formative survey indicated that access to quality fruits and vegetables is a barrier for participants. The research team recognizes that this is no easy task, but one that would likely produce healthy outcomes for large numbers of community members.

• In addition to grocers making an effort to put their stores in locations that would be best for all Hopkinsvillians, it would also be helpful if grocery stores provided healthy foods at low cost. As discussed above, the item “It is hard for me to eat more fruit and vegetables because they are too expensive” was directionally significant (t = −1.769, p = .079) in the regression model. Making fruits and vegetables more affordable for the participants, rather than junk foods, would likely have a positive impact on the community.

• Campaigns to increase fruit and vegetable consumption should enlist family members to help one another eat more fruit. In this study, 68.6% or 2/3 of the participants said they are actively trying to eat more vegetables and fruit, and that they were supported by family members and friends in seeking this healthy change in their diets.

• Those charged with the task of developing initiatives targeted toward the African American community in Hopkinsville must fully consider the best media outlets for providing information about fruits and vegetables. Participants indicated that they would prefer to see messages on television and receive written communications through the mail. Of course, the cost effectiveness of these messages must be considered, as well as the time of the day and the network on which the information would be broadcast on TV, for example.

• Also of concern for health communication practitioners is determining exactly what elements the informational messages should contain. The data here suggested that participants would respond more favorably to messages that explain that fruit and vegetables are easy to prepare and inexpensive to purchase. Exploring these possibilities further, the research team is currently entering the last phase of the study, analyzing data produced by a marketing trade-off survey, to determine the best course of action to encourage African American Hopkinsvillians to eat more fruits and vegetables.

In conclusion, the study, which sought to determine community, interpersonal and intrapersonal factors that influence fruit and vegetable consumption of African American community members in Hopkinsville, provides data that reify the structure of the social ecological model. A strength of the research lies in the fact that the research team heeded the recommendations of McLeroy and associates (1988) and Fleury and Lee (2006), who posited the importance of including and consulting members of the community in determining ecological strategies. The research team’s community contacts’ willingness to engage and interact with the study, through all of its phases, has helped the researchers to develop communication strategies that the team hopes to implement in the near future. The research team’s ultimate goal is to encourage African American Hopkinsville residents to make lifestyle choices that better their health.