Study Design
This study was part of an independent evaluation of the FNV Campaign conducted by the research team between September 2015 and December 2017. In 2015, the Robert Wood Johnson Foundation (RWJF) commissioned an independent evaluation of the FNV Campaign around the time of the PHA’s launch in the two pilot markets. Therefore, the research team was unable to use a randomized- or matched-control study design to compare the FNV Campaign implemented in the two pilot markets. Due to resource limitations and a lack of baseline (pre-intervention) data, alternative designs that included comparison groups (e.g., alternative markets) or pre- post- tests were not feasible, so a cross-sectional study design was used. As the FNV Campaign was a new intervention, no specific hypothesis was established to test or to conduct sample size calculations.
Data for this study were collected from a cross-sectional online survey conducted among the target audiences from the two pilot markets between February and July 2017. The study aims, eligibility criteria, and measures used in the survey were developed in conjunction with the RWJF with input from the PHA staff. The Virginia Tech Institutional Review Board approved the study protocol that involved human subjects in December 2016 (IRB #15-1110).
The evaluation approach used to assess associations between awareness of the FNV Campaign and fruit- and vegetable-related outcomes in this cross-sectional study was similar to that used for the Fruits and Veggies–More Matters Campaign [31]. The Fruits and Veggies–More Matters Campaign was implemented nationally to encourage adults to consume fruits and vegetables, so the survey sample was recruited from a national panel of US adults. As the FNV Campaign was aimed at encouraging fruit and vegetable purchases and consumption among mom and teen audiences in two CA and VA markets, the recruitment efforts and eligibility criteria corresponded to these target population criteria.
The survey was pilot-tested to ensure that the procedure from participants initiating the survey to completion and compensation the participant initiation of the survey through completion and compensation went as planned, and to ask for participants’ input on survey clarity and ease of use. A small sample of mothers (n = 3) and teens (n = 2) who lived outside of the pilot markets were recruited to take the survey and provide any feedback on the clarity and process for initiating and completing the survey. The survey procedure was executed as planned and participants did not report any issues or suggestions to improve the survey, so the team then began recruitment efforts in the two pilot Fresno, CA and Hampton Roads, VA markets.
Recruitment Sites
Participants were recruited by members of the research team to complete an online survey through a non-probability convenience sampling strategy that was intended to reach members of the target audiences in Fresno, CA and Hampton Roads, VA. Recruitment in the two markets included community outreach with assistance of local organizations (e.g., daycare and youth activity centers, faith-based and social support non-profits) involved with the target populations, and distribution of print and digital flyers to local organizations that described the study, eligibility criteria, and provided a link to access the survey. Organizations were asked to post and/or email recruitment flyers to potentially eligible contacts. The main organizations involved with recruitment in Virginia included Head Start, Virginia Cooperative Extension, and the Five Points Community Farmers Market; main organizations involved with recruitment in Fresno included Head Start, Fresno Parks and Recreation, and the non-profits Cultiva La Salud and Centro La Familia.
Participants
Adolescents and young adults (referred to as teens) aged 14 – 20 years and mothers aged 21 – 36 years who were residents of Fresno, CA or Hampton Roads, VA were eligible to participate in the study. The target audience criteria for this survey were identified in coordination with the PHA and the grant funders based on the target audience demographic characteristics of the FNV Campaign at the time. Participants were instructed to use the survey link that corresponded to their demographic criteria so that those ages 14 – 20 years could participate in the adolescent/young adult survey and mothers aged 21 – 36 years could choose the mom survey.
Participants recruited from Fresno, CA and Hampton Roads, VA were screened prior to taking the survey to ensure that their demographic characteristics and residence met the eligibility criteria, by asking respondents to indicate their age range and place of residence. Participants received information about the research prior to participating and provided implied informed consent or assent by beginning the survey. As there was minimal to no risk involved in participating in the survey, a request to waive parental permission for teens under the age of 18 was approved by the Virginia Tech Institutional Review Board. Participants could take the online survey in English or Spanish one time either onsite with a researcher using an iPad tablet, or remotely at participants’ convenience. Respondents that did not meet the eligibility criteria (e.g., outside of the target audience age range or market location) were restricted from participating in the survey. Participants received a $10 gift card in-person or through email after completing the survey.
Survey Measures
To assess familiarity with, or awareness of the FNV Campaign, the research team developed the following questions with input from the PHA and RWJF: (1) “Do you know what this brand or logo represents?” (featured in an image of the FNV logo along with the text of this question), (2) “Have you seen any versions of the FNV brand or logo around town or in your community?”, and (3) “Have you heard of the FNV Campaign?”; response categories were “yes”, “no”, or “unsure”. Respondents who selected “yes” to one or more of the three questions were coded as aware of the FNV Campaign.
Three items were adapted from the Food Attitudes and Behaviors Survey, which was developed based on a comprehensive literature review and expert content validity review of psychosocial constructs correlated with fruit and vegetable intake [32, 33]. Participants were asked to indicate their agreement with each statement on a five-point Likert scale ranging from “strongly disagree” to “strongly agree”, which were recoded as values 1 through 5, respectively. Attitudes toward consuming new fruits and vegetables (i.e., neophobia) were assessed through the statement: “I enjoy trying new fruits and vegetables”. Cognitive beliefs around perceived barriers to fruit and vegetable consumption were evaluated through respondents’ agreement with the statement “I just do not think of fruits and vegetables when I am looking for something to eat”; for the purposes of this evaluation, this statement was also used to indicate salience of fruits and vegetables. Encouragement was assessed through respondents’ agreement with the statement “I encourage my friends and family to eat fruits and vegetables”. Behavioral intentions related to fruit and vegetable purchases and intake were evaluated using two internally developed items that asked participants how likely they were to buy, and eat, “a fruit or vegetable over the next week”, as these were targeted behavioral outcomes of the FNV Campaign. The response categories for the two behavioral intention items were collapsed into unlikely (“unsure” or “unlikely”), and likely (“likely” or “very likely”) to dichotomize the responses to intending or not intending to buy and eat fruits and vegetables over the next week.
Behavioral outcomes were assessed for new fruit and vegetable intake and frequency of fruit and vegetable intake. An open-ended, internally developed question asked respondents to list any new fruits and/or vegetables they had tried; “What new fruits or vegetables have you tasted over the past 3 to 6 months that you have never eaten previously?” Written responses were transformed into a dichotomous response variable to indicate whether respondents reported trying any new fruits or vegetables. Fruit and vegetable intake frequency was assessed through items adapted from the validated Behavioral Risk Factor Surveillance System (BRFSS) fruit and vegetable screener [34]. Participants were asked through six items to indicate how many times they ate or drank 100% fruit juice, fruit, vegetable juice; and dark green, orange, and other vegetables during the past month, from “never” to “2 or more times per day”. At the time that this study was developed, legumes and beans were not widely promoted in the FNV Campaign and were not included in the vegetable intake assessment. Fruit and vegetable intake responses were converted into frequencies per day, which was summed into total daily fruit and vegetable intake frequency for analysis. The demographic characteristics of survey respondents included sex, age, race/ethnicity, highest level of education, and geographic residence.
Data analysis
All statistical analyses were performed separately for the mom and teen target audience respondents and by location using SPSS statistical software version 25 for Windows. Data analysis methods were selected and implemented with consultation from the Virginia Tech Statistical Applications and Innovations Group based on study objectives and data.
Descriptive statistics were calculated for respondent demographic characteristics. Chi-squared tests were used to assess sample representativeness based on proportions of race/ethnicity [20] and educational attainment [21] as compared to proportions in the Fresno and Hampton Roads market locations. Race/ethnicity, sex (adolescents only), education (mothers only), age, and geographic location were included as covariates in all analyses because of potential biasing effects on outcomes related to fruit and vegetable consumption [35, 36].
Multivariate analysis of covariance (MANCOVA) was conducted to assess whether fruit- and vegetable-related attitude, belief, and encouragement outcomes (dependent variables) differed between respondents aware of the FNV Campaign (independent variable). The use of MANCOVA to analyze several related variables is recommended over conducting separate univariate analyses to reduce the risk of family-wise error [37]. Associations between awareness and dichotomous variables for behavioral intentions and trying new fruits and vegetables were assessed using binary logistic regression. Analysis of covariance (ANCOVA) was used to determine whether total frequencies of daily fruit and vegetable intake differed between aware and unaware respondents, to assess potential influence of the FNV Campaign on targeted intake behaviors.