We conducted a pre-post, quasi-experimental study in one elementary school in Cheongju, South Korea, where we obtained permission from the principal and teachers to implement the program after explaining the purpose and procedures. Children in two second-grade classes participated in the study. One class (35 children) was designated as the intervention group, and the other class (36 children) was designated as the control group. The teachers gave the children a package with the study information, a consent form, and a checklist to screen for those with allergies to vegetables used in the program to take home. All parents returned written affirmative consents and the checklists. The Institutional Review Board of Chungbuk National University approved the study protocol (CBNU-201405-BMSB-054-01).

Children in the intervention group participated in a four-session cooking program. They were asked to complete questionnaires before and after the program for outcome evaluation as well as a questionnaire for process evaluation after each session. Parents of the children in the intervention group were also asked to complete a questionnaire after the intervention about activities at home and children's behavioral changes. Children in the control group completed questionnaires before and after the program in which they did not participate.

We implemented the Veggiecation program for four weeks in June 2014. The intervention consisted of a weekly 40-min session including classroom lecture and hands-on cooking activities. Two vegetables were introduced in each session. Table 1 shows two main vegetables, the name of the dish, cooking activities, and key nutrients in each session. For the first 10 minutes, a nutrition teacher introduced the vegetables and their key nutrients. For the next 30 minutes, the children prepared recipes, with the help of teacher assistants, and tasted the vegetables they prepared. For the cooking activities, we developed eight new recipes including vegetables familiar to Korean children: bell pepper, daikon, zucchini, tomato, sweet potato, spinach, cucumber, and oyster mushroom. To create a learning and interactive environment, Veggiecation posters of program vegetables were translated into Korean and displayed in the intervention classroom [15]. The title song of the original Veggiecation was played in the background during the cooking activities. All preparation and cooking procedures followed the Hazard Analysis and Critical Control Point system guidelines.

We developed a self-report questionnaire to measure vegetable consumption behavior, intention, and possible influencing factors based on previous studies [78916]. We categorized possible factors as personal, social-environmental, and physicalenvironmental factors. Demographic questions included the children's gender, the mother's work status, and who prepared the meals at home.

Table 2 presents an overview of the study variables. Vegetable consumption behavior was measured with four statements: I eat vegetables at breakfast, I eat vegetables at school lunch, I eat vegetables at dinner, and I eat more vegetables than my friends. Each statement measured the behavior frequency on a four-point scale (often, sometimes, rarely, and never). The behavior was scored by the mean of the points in the four items. The Cronbach alpha was 0.696.

Intention was assessed using three statements: vegetables are food that I want to eat; I plan to increase vegetable intake; and I plan to eat more vegetables in the next six months. Each statement measured the likelihood of the statement on a four-point scale (very likely, likely, unlikely, and very unlikely). The intention was scored as the mean of the points in the three items. The Cronbach alpha was 0.619.

Personal factors included attitude, preference and self-efficacy. The attitude toward eating vegetables was assessed as the mean of the points of three bipolar statements using a five-point scale: eating vegetables is hard/easy, not enjoyable/enjoyable, and not good/good for health. The Cronbach alpha for the attitude measurement was 0.668. Preference was measured by the mean of the scores relative to like or dislike of the eight vegetables used in the intervention on a five-point scale. The Cronbach alpha for the preference was 0.777. Self-efficacy was assessed using the statement “I can eat vegetables any time I want” on a four-point scale (very likely, likely, unlikely, very unlikely). We only included neutral responses for statements assessing attitude and preference in our questionnaire, making it clear that children could express a neutral opinion. The rest of the questions did not include neutral responses to avoid their use as an “I do not know” option for children [17].

Social-environmental factors included modeling and encouragement. Three significant others were listed for the questions to measure modeling and encouragement: mother, father, friends for modeling; and mother, father, a nutrition teacher for encouragement. The Cronbach alphas for modeling and encouragement were 0.639 and 0.775, respectively.

Physical-environmental factors included previous exposure, availability at home, and availability at school. Previous exposure was measured by scoring the experience of eating eight vegetables used in the program. The Cronbach alpha for the previous exposure was 0.508. Availability was assessed by scoring the statement “I always have vegetables available at home/school” on a four-point scale.

Cronbach alpha, a measure of variable reliability, is influenced by the study sample size and the target population's characteristics [18]. The reliability coefficients of mediating variables used in 265 nutrition intervention studies were reported mostly around 0.6–0.7 [18]. In addition, in studies assessing reliabilities of attitude scales, Cronbach alphas ranged between 0.5 and 0.8 among school-aged children, and between 0.4 and 0.7 among preschool children [18]. As the current study participants were second graders, a coefficient above 0.6 should be acceptable [19]. In cases of previous exposure to any of the vegetables used, the Cronbach alpha was not an appropriate reliability measure.

After each session, the children were given another questionnaire for process evaluation to measure the level of comprehension of the lecture, the degree of their satisfaction with cooking, and the taste preference for the vegetables they prepared. A questionnaire for parents after the intervention asked whether the children had brought the program content home, whether the parents saw the recipe used at school and prepared it at home, and whether the children had any changes in their vegetable consumption behavior.

Frequency distributions or means and standard deviations for each item were calculated. Pearson correlations were computed between vegetable consumption behavior, intention, and possible factors. Our regression analyses identified the relative influence of theoretical constructs on the intention to eat vegetables and also on the behavior. We entered mean scores of the construct measures into the regression model. To measure the intervention's impact, we compared the mean scores of behavior, intention, and the personal factors within each group before and after the intervention using the paired t-test. We also used an analysis of covariance to determine intervention effects between the two groups after the intervention. The variables' values before the intervention served as covariates in these analyses. P-values < 0.05 were considered as statistically significant. All statistical analyses were performed using the statistical analysis system software (SAS, version 9.4, SAS Institute Inc., Cary, NC, USA).

Seventy-one children participated in this study, with a 100% survey response. Our participants included 36 boys and 35 girls; 62% of children had working mothers, and 81.7% came from homes in which the mother prepared the meals. The demographic characteristics were similar between the intervention and control groups (P > 0.05). The most frequent response to the question about reasons for not enjoying vegetables was taste, followed by texture, lack of exposure to the vegetable, and smell. The response patterns were similar between the two groups (P > 0.05) (data not shown).

Table 3 shows the correlations between vegetable consumption behavior, intention, and the study variables. Behavior was significantly correlated with all the variables; attitude (r = 0.733), preference (r = 0.473), availability at home (r = 0.413), modeling (r = 0.367), previous exposure (r = 0.362), availability at school (r = 0.334), and self-efficacy (r = 0.275). However, encouragement of parents and a nutrition teacher was negatively correlated with behavior (r = −0.322).

Children's intention was positively associated with preference (r = 0.510), attitude (r = 0.432), self-efficacy (r = 0.343), and modeling (r = 0.299). Attitude was associated with all variables. Preference was positively associated with attitude (r = 0.613), modeling (r = 0.491), exposure (r = 0.336), and availability at home (r = 0.278), and negatively associated with encouragement (r = −0.339). Self-efficacy was not correlated with any environmental variables.

The regression analysis results for intention and behavior are presented in Table 4. We deleted encouragement from the determinant list of intention and behavior before running the regression analyses because encouragement of parents and a nutrition teacher to eat vegetables was negatively associated with the vegetable consumption behavior in children, indicating that children who consumed fewer vegetables were encouraged more as a result

The first model for intention, which included personal factors alone, explained 30.2% of the variance (F = 11.08, P < 0.001). Preference and self-efficacy were identified as influencing factors for intention; and preference was the most determining factor. We added environmental factors to the second model for intention. The variance explained by both personal and environmental factors was also significant with 29.5% (F = 5.18, P < 0.001), yet lower than the variance accounted by the personal factors alone. Our results suggest that intention was mainly dictated by personal factors. In this second model, preference was the most influential factor determining intention.

The first model for behavior explained 54.5% of the variance (F = 21.94, P < 0.001). The second model for behavior with additional environmental factors accounted for 56.3% of the behavior variance (F = 12.26, P < 0.001), an increase of 1.8% from the first model with personal factors alone. In both models, attitude and intention were associated with behavior, and attitude had the greatest influence.

Means and standard deviations of variables in the intervention and control groups before and after the intervention are shown in Table 5. We detected no differences between the groups at baseline. We found a significant improvement in vegetable consumption behavior among children in the intervention group (P < 0.05). Similarly, we found significant improvements in intention, attitude, and preference among children in the intervention group. On the other hand, children in the control group showed no changes in vegetable consumption or any variables during the period between the two assessments. After the intervention, controlling for pre-intervention scores, we found the intervention group had significantly higher scores in all the variables, including self-efficacy, than the control group.

After each weekly cooking activity, children completed a short questionnaire about the intervention process. The children expressed high satisfaction with the cooking: all except one child in the second week indicated the cooking activity was fun or extremely fun. The Veggiecation curricula were well understood by the children: over 85% of them expressed full understanding of the lectures. In addition, more than 85% of children rated all program dishes they had prepared as tasty or extremely tasty (Table 6). More than 70% of children answered that they would surely eat the dish if it were provided at school lunch.

The parents in the intervention group completed a post-intervention survey to evaluate the process (85.7% response), and the results are shown in Table 7. The Veggiecation program was well received by the families: 96.7% of parents were aware of the program recipes that children brought home. However, only 43.3% of parents tried the recipes at home. Most parents (83.3%) reported that the children in the intervention group tried new vegetables after the intervention. The vegetables children tried to eat included bell pepper (46.7%), oyster mushroom (23.3%), daikon (20.0%), tomato (16.7%), and zucchini (16.7%) in that order. All parents had a positive attitude about continuing the intervention at school.