The intervention is an online lesson to promote farmers’ market FV purchases and consumption. Women in the control group complete any of seven existing online lessons (lessons are available on breastfeeding, being active, fruits and vegetables, calcium, cholesterol, oral health and iron). The lesson comprises three modules, each consisting of 1) behavior change content presented through a video segment featuring WIC participants, and 2) an interactive activity to build targeted knowledge, attitudes, and skills. Existing online lessons consist of an introductory segment presented with online text and graphics. After reading this material, women have the option to complete one of four lesson activities. The activities provide opportunities for women to read further on the topic and are designed to reinforce key points of the lesson.
Formative research with state and local WIC agency representatives (N = 5) and participants served by the collaborating WIC agency (N = 54) identified influences on farmers’ market FV purchases and consumption that are the focus of the lesson, i.e., lack of knowledge of the FMNP and WIC-authorized farmers’ markets (locations, hours of operation and transportation options to markets), negative attitudes towards farmers’ market FVs (including farmers’ market FV food safety concerns), limited awareness of locally grown, seasonal FVs (items that are in season and selection, storage and parts eaten of the items), limited asking skills (extent of asking farmers about their participation in the FMNP and their produce), limited FV food safety and farmers’ market FV preparation skills, and low positive outcome expectations for consuming locally grown FVs.
Given the advantages Internet interventions afford (e.g., scalability or accessibility by many people, simultaneously and repeatedly; uniformity of delivery; the ability to provide customized, engaging and interactive content and low delivery costs) [28, 29] and interest in designing a resource that would fit well within WIC clinic operations, the lesson is delivered online. In 2009, the New Jersey WIC program launched an interactive website offering nutrition and health lessons . In 2011, computer kiosks were placed in all local agency administrative offices to enable WIC participants to complete lessons while in the WIC clinic . The expansion of topics is one component of the state’s plan for enhancing revisit rates to the website .
Women from the collaborating WIC agency were primarily African American (45 %) and Hispanic (44 %), with 63 % of Hispanics reporting Spanish as the primary language spoken at home. In light of the high proportion of participants speaking Spanish as their primary language, English- and Spanish-language versions of the lesson were developed. Women had less formal education for their age group than the U.S. population average (67 % had earned a high school diploma; 19 % had not completed high school) ; as such, they are likely to experience low health literacy . Content delivery is therefore through short video segments and audio output to maximize accessibility for learners with low literacy and numeracy skills . Because lesson activities require user input, to maximize comprehension, audio instruction on the completion of activities is provided, questions and responses are read aloud as they appear on the screen and viewers can hear them again by clicking on audio icons adjacent to on-screen text. To enhance community ownership of and the credibility and relevance of messages, WIC participants were engaged as full partners in the development of the lesson (as described below) and the delivery of content (videos feature participants). Owing to the strong oral tradition of the primarily African American and Hispanic women for whom the lesson is developed, key messages are conveyed through participant experiential narratives or firsthand stories [34–36].
Six versions of the lesson were planned (one for each month of the farmers’ market season featuring locally grown, seasonal items). This will allow women presenting for nutrition education in a particular month to receive information on FVs they are likely to find at farmers’ markets at that time of year. For the current study, the lesson for the month of July was developed.
In a serial format or single narrative spanning the three modules, WIC participants featured in the lesson discuss challenges that prevented them from purchasing and consuming farmers’ market FVs (exposition and beginning of conflict), events serving as a catalyst for change (complication and climax), ways they overcame the challenges and strategies (education and skills training) for helping viewers do the same (resolution) . Lesson modules, goals (targeted outcome changes), content and activities are summarized below.
Goal: To increase knowledge of the FMNP and WIC-authorized farmers’ markets and promote favorable attitudes towards farmers’ market FVs. Content: Viewers are introduced to three WIC participants shown waiting for a nutrition class at the collaborating WIC agency. Each shares “her story” (reasons she did not shop at farmers’ markets, something that happened to change her mind about this, and lessons learned that she will share with viewers). Thereafter, the first participant discusses common misperceptions about farmers’ market FVs and provides corrective feedback based on her experiences. Activity: Viewers enter their zip code into a Farmers’ Market Locator Tool and are provided with information on the three WIC-authorized markets nearest to their home. They have the option to search in another zip code area and can email themselves information about the different markets.
Goal: To promote positive outcome expectations for consuming locally grown FVs and improve farmers’ market FV knowledge and asking skills. Content: Featuring the second WIC participant filmed at a local farmers’ market, the participant discusses FVs that are in season, reasons to choose locally grown seasonal items, and featured items (selection, storage and parts eaten of three locally grown, seasonal items). Throughout the discussion, the participant emphasizes the importance of getting to know local farmers. A local farmer also is featured. The farmer discusses his motivation for selling his produce at the market (to educate consumers about different FVs that are locally available), his appreciation for having customers ask about the FVs he grows, and his hope that WIC participants will take advantage of the fresh FVs local farmers have to offer, noting that he and other farmers are glad to work with WIC participants and realize how important it is for women and children to have high quality FVs. Activity: Market Smarts Quiz (viewers respond to true/false items about featured items).
In the kitchen
Goal: To improve FV food safety and preparation skills. Content: Featuring the third WIC participant shown in a kitchen, the participant demonstrates the safe handling of FVs, including tips specific to the three items featured in Market Smarts. Video demonstrations of recipes for preparing the items also are shown. Only recipes that can be made with other items in the WIC food package are presented. Activity: Viewers rehearse steps for making a FV recipe by placing, in the correct order, picture cards depicting the different recipe steps.
Immediately after completing the lesson, women receive a packet of handouts highlighting information presented in the lesson. One, two, and three months after the lesson they receive a follow-up email that links to a video featuring WIC participants shown in the lesson. Participants “check-in” with viewers to ask if they have been to a farmers’ market, reinforce key messages presented in the lesson and demonstrate a new recipe for preparing a featured FV item.
The lesson is grounded in formative research on knowledge, attitudes and skills influencing farmers’ market FV purchases and consumption and theoretical understanding of approaches to modifying them. Social Cognitive Theory supports leveraging environmental FV resources (i.e., promoting FMNP voucher and CVV redemption); use of credible and relatable role models (i.e., WIC participants) and feedback to foster observational learning and outcome expectations; and education, experiential activities (e.g., web-mediated skills-building exercises) and delivery modalities that engage and sustain learners’ interest to build knowledge and skills [37–40]. Culture-centric approaches to health promotion program development emphasize the use of narratives or stories to facilitate behavior change [36, 37]. Narratives among audience members about experiences with a focal issue may ground programs so that they are more culturally meaningful . Moreover, behavior change can be enhanced when stories address behavior-change issues and the lesson communicated in the story promotes health behavior change . The Transportation-Imagery Model attributes narrative persuasion to “transportation,” becoming highly absorbed with a story. Transportation makes the story seem more like actual experience; transported individuals are less likely to counter-argue and to believe story propositions and to adopt the beliefs of characters with whom they identify [41, 42].
Lesson development was informed by a community-based participatory research approach actively involving key stakeholders, i.e., state and local WIC agency representatives and WIC participants, as full partners in all phases of the work [43, 44].
WIC advisory board
At the start of the project, an advisory board of WIC agency administrators and staff was convened to meet monthly throughout the project. The board assisted with the development of the lesson and is providing ongoing direction on the coordination of study activities. Monthly meetings document materials reviewed by the board and corresponding recommendations; problems, if any, encountered and strategies to avoid or resolve them; and timeline adherence.
Four focus groups per module (two in English and two in Spanish) were planned to gather feedback specific to the different modules. In total, 14 groups were held, as additional feedback was sought regarding one of the modules. Trained bilingual (English-Spanish) RAs moderated the groups. Groups consisted of 4–5 participants each (N = 56 participants total) and were 50 min in length. To obtain in-depth information from participants, three questions per group were asked (the questions differed for each module), affording each participant 3 min to respond to each question. Groups of questions (and corresponding modules) were focused on attributes of WIC participants featured in the lesson, settings in which shown, and events serving as catalysts for behavior change (Farm Fresh); perceived rewards of eating FVs, reasons to eat locally grown seasonal items, and focal FVs [participants sorted picture cards depicting locally grown seasonal items into piles of familiar and unfamiliar items and selected one item in each pile that they would like to learn more about] (Market Smarts); and reactions to a 4-min FV food safety video (what was liked and what, if anything, might be difficult to do)  and FV recipes (In the Kitchen). The groups were tape-recorded. Audiotapes were transcribed and analyzed using tape-based analysis . Findings guided the development of a written curriculum of lesson content.
The written curriculum guided the development of storyboards for each module. Using established guidelines , a second round of 12 focus groups was held (N = 52 participants) to gather feedback on 1) likely effects of the module on targeted knowledge, attitudes and skills, 2) content, if any, that should be eliminated, and 3) content, if any, that should be changed to increase potential effects. Data were analyzed using the same approach as that used to analyze data from the first set of groups . The written curriculum was revised based on the feedback.
Advisory board members identified English- and Spanish-speaking WIC participants to narrate lesson modules based on character attributes identified in the first round of focus groups. Videos were filmed on location in settings identified by the groups. WIC participants agreeing to narrate the modules received $50 gift cards (redeemable at local supermarkets and chain stores) for each half-day of filming. English- and Spanish-language storyboards developed for the second round of focus groups served as video scripts. RAs assisted the narrators in rehearsing the scripts.
Pretesting video segments
Participant reactions to “rough cuts” of video segments developed for the first lesson module were assessed in one-on-one sessions with RAs using the think aloud method . Five participants each viewed English- and Spanish-language versions of the segments. Following a brief rehearsal of the response task, the participant previewed the focal video and verbalized reactions to it. After the video, participants were orally administered brief measures of the extent of identification with the narrator and transportation into and liking of the video (Cronbach alphas ≥ .76) [42, 49, 50]. Probes were used to explore reasons for answers given. The feedback guided edits to and/or re-filming of footage lacking quality, relevance and impact. The revised segments underwent pretesting in a second round of 1:1 sessions. Lessons learned from pretesting the segments served as a guide for filming the remaining lesson modules.
Developed in HTML, the lesson consists of a user interface with audio output, video segments, Adobe Flash activities, content managed pages (with information on local farmers’ markets for WIC administrators to edit as needed), and reporting tools (reports by WIC participant identifier of lesson modules completed). Internal testing by RAs documented the absence, if any, of required features. The documentation served as a guide for the multimedia producer to revise, beta-test, debug and finalize the lesson in preparation for the outcome study.
Although measures of adult FV-related knowledge, attitudes and skills exist, measures of farmers’ market-specific constructs are lacking and were therefore developed. Adapted from existing instruments (Cronbach alphas ≥ .70) [40, 51–59], up to ten items per scale were written (some negatively worded to avoid a response set) . Response options are on a yes-no, true-false, multiple choice or 7-point Likert scale format tailored to the nature of the questions. Cognitive testing with 15 participants gathered data for improving items and response formats that were unclear and/or difficult to understand . The psychometric properties of the measures will be examined using data from the full administration of the instruments at pretest.
At pretest, the following sociodemographic variables are measured: date of birth, age (in years), pregnancy status, due date (pregnant women), breastfeeding status, race, ethnicity, nativity, preferred language, language(s) spoken at home, marital status, educational attainment, educational attainment of spouse or partner, number of children in the household under age 19, number of children in the household between 2 and 5 years of age, number of other adults in the household, employment status, participation in assistance programs and food security status. Because self-reports of FV intake are influenced by social approval and social desirability biases [62, 63], social desirability trait is also assessed at pretest using a validated, short form of the Marlowe-Crowne Social Desirability Scale .
Assessed at pretest, posttest and 3- and 6-month follow-up, outcome measures are knowledge of the FMNP and WIC-authorized farmers’ markets, attitudes towards farmers’ market FVs, awareness of locally grown, seasonal FVs, farmers’ market FV purchases (ever purchased FVs at a farmers’ market, purchased FVs at a farmers’ market in the past 2 weeks), intentions to purchase FVs at a farmers’ market in the next 2 weeks, FV food safety skills, farmers’ market asking and FV preparation skills, and positive outcome expectations for consuming locally grown FVs (measured using instruments developed for the study). The frequency and quantity of FV intake are assessed with validated instruments [65, 66]. Frequency of intake is measured using the FV module of the Behavioral Risk Factor Surveillance System, shown to have moderate validity and reliability [65, 67]. Quantity of intake is measured with a 2-item screener developed by the National Cancer Institute . The screener is valid (as evidenced by moderate correlations with FV intake assessed via multiple 24-h dietary recalls) and reliable (as evidenced by moderate 2–3 week test-retest correlations) . FMNP voucher and CVV farmers’ market FV purchases are assessed using voucher redemption data provided by the state WIC agency.
Process measures include lesson dose (data recorded by RAs and tracked through the website on the number of lesson modules and activities participants completed [a total of three each for those completing the lesson and one each for those completing existing online health education]), distractions, if any, experienced during lesson play (recorded by RAs), participant self-report data on existing online lessons, if any, completed prior to the study, user satisfaction with the lesson received (ratings, on a 7-point scale, of the extent of enjoyment, interest in, and likelihood of recommending the lesson to other WIC participants), measures of new information learned from the lesson and talking to family and friends about new information learned , whether this was the first time completing an online WIC nutrition education lesson and the perceived novelty of the lesson . Among women who receive the lesson, measures of the following also are administered: what was remembered most about the lesson ; what was liked and disliked about the lesson and what, if anything, could be done to improve it; transportation into the video narrative  and identification with the characters ; liking and learning from lesson activities; the activity that was liked the most; the FV the participant chose to learn a recipe about and whether the participant tried the recipe at home; and whether the participant opened follow-up emails sent after the lesson, watched the videos, and tried the recipes shown and the perceived helpfulness of the information provided. Among all participants, RAs collect information on the number screened, determined eligible/ineligible and enrolled/not enrolled (during recruitment), and follow-up calls made/completed, follow-up assessments scheduled/completed and problems, if any, encountered in reaching participants.
All self-report data are collected by RAs in 1:1 interviews with participants (interviews are conducted in person at pretest and by telephone at posttest and follow-up measurements). One month prior to 3- and 6-month follow-up assessments, RAs telephone participants to confirm their contact information and the date/time of their follow-up interview (participants are also mailed the date/time of their appointment 1 month and 2 weeks prior to the appointment). To enhance the quality of measurement, RAs were trained in a full-day session prior to data collection. Daily half-hour debriefing sessions are used to discuss problems, if, any, encountered with data collection, ways to address them and strategies to avoid their recurrence.
To verify the success of randomization, between-arm differences in sociodemographic variables and prior lessons received will be examined using analysis of variance or chi-square tests as appropriate. Characteristics of women who completed the study and those lost to attrition also will be compared. Variables found to differ by arm will be included as covariates in outcome analyses. Outcome variable distributions will be assessed and normalizing transformations applied as needed. If the transformations are unsuccessful, analyses will be implemented as generalized linear models with the appropriate error distributions (e.g., Gamma or Poisson).
Using an intent-to-treat approach (i.e., all women analyzed as randomized regardless of their adherence to the protocol), analysis of covariance (ANCOVA) models will be used to test hypotheses 1–3. The models will relate posttest FV intake to pretest FV intake, study arm, and covariates identified in preliminary analyses (described above). Tests of pair-wise differences in least-square (LS) means specified in hypotheses 1–3 will determine support for each hypothesis. In a parallel fashion, lesson effects on the redemption of CVVs at farmers’ markets and on FMNP voucher redemption (among women who received the vouchers [hypotheses 4–5]) will be examined. Next, hypotheses 1–5 will be tested in a longitudinal context using linear mixed-effects (LME) models with three repeated measures. Tests of pair-wise differences between LS means specified in hypotheses 1–5 will determine whether lesson effects are sustained over time.
The method of Baron and Kenny and MacKinnon [68, 69] will be used to determine evidence for mediation by knowledge, attitudes and skills (hypothesized mediators) of lesson effects on FV intake and voucher redemption (hypothesis 6). Both single- and multiple-mediator models will be tested. Using ANCOVA for posttest measurements and LME for longitudinal follow-up, lesson effects on potential mediators will be examined. Significant lesson effects that are reduced when controlling for potential mediators (evaluated using Sobel tests) [70, 71], will provide evidence that partial mediation has occurred; the percent of variation in outcomes mediated by knowledge, attitudes and skills will be determined . The absence of lesson effects after controlling for the mediators will indicate that complete mediation has occurred.
Positing that receipt of more of the lesson (as measured by the number of completed modules and activities) will foster more positive outcomes, hypothesis 7 tests dose-response relationships. This hypothesis will be examined in a series of ANCOVA and LME models (using data provided by women who received the lesson) with posttest and follow-up measures of FV intake, voucher redemption, knowledge, attitudes and skills as outcomes, lesson dose as the fixed factor, and receipt of FMNP vouchers, socio-demographic variables and pretest measures of each outcome as covariates. The significance of the coefficient for the lesson dose variable will determine support for hypothesis 7. Similarly, for those who received existing online health education, the significance of the variable reflecting which of the lessons was received will be evaluated.
Descriptive statistics will examine satisfaction ratings among women who received the lesson. A mean item score ≥ 5.0 (on the 7-point scale) will be considered evidence of a high degree of satisfaction. In a test of hypothesis 8, mean satisfaction ratings will be compared to determine whether satisfaction was higher among women who received the lesson relative to those who received existing online health education. Responses to open-ended items will be summarized. The feedback will be used to identify areas for further refining the lesson.
Previous work demonstrates that FMNP effects are stronger when vouchers are supplemented with nutrition education . In light of this work, it is anticipated that at posttest and follow-up measurements, the highest levels of intake will be found among women who receive the lesson and FMNP vouchers (relative to women who receive the vouchers and existing online health education and those who do not receive FMNP vouchers and receive the lesson or existing online health education). Regardless of which lesson women receive, at posttest and follow-up measurements, FV intake is expected to be higher among those who receive FMNP vouchers relative to those who do not, owing to research demonstrating that by improving economic access to FVs, the provision of vouchers leads to improvements in FV intake [13–16]. Regardless of FMNP voucher receipt, FV intake and redemption of CVVs at farmers’ markets also are expected to be higher among women who receive the lesson relative to those who receive existing online health education, owing to improvements in targeted knowledge, attitudes and skills. Improvements in these outcomes also are expected to result in higher voucher redemption among voucher recipients who receive the lesson relative to those who receive existing online health education.
In tests of secondary study hypotheses, it is expected that the lesson will promote positive changes in targeted knowledge, attitudes and skills and that these improvements, in turn, will lead to increases in FV intake, FMNP voucher redemption and the redemption of CVVs at farmers’ markets. Further, it is expected that receipt of more of the lesson will be associated with more favorable changes in targeted knowledge, attitudes and skills. Finally, it is anticipated that user satisfaction will be uniformly high among women who receive the lesson and will be higher in this group as compared to women who receive existing online health education owing to features of the lesson intended to capture and sustain users’ interest, i.e., delivery through a highly accessible medium, the narrative lesson format, use of credible and relatable role models to deliver key messages and inclusion of interactive elements and activities.