Test–retest reliability and comparison of children's reports with parents’ reports of young children's fruit and vegetable preferences

Measurement is the foundation of research and theory on children's eating behaviours: Current issues and directions for progress

Eating behaviours affect food intakes and are involved in the aetiology of obesity. There has been impetus to translate findings about children's eating behaviour into intervention and policy programs. However, measurement limitations have hindered our capacity to understand and influence children's eating behaviours. In the present paper we provide an overview of some of the key methodological and measurement issues facing the field of children's eating behaviours and highlight implications for research and health promotion. Drawing on insight from parallel issues that occur in the measurement of early social and emotional development, we examine two overlapping themes in children's (aged 0-∼12 years) eating behaviours (Somaraki et al., 2021) measurement issues related to validity and reliability, and (Steinsbekk & Wichstrøm, 2015) associated methodological challenges, such as contextual influences and the importance of designing studies that use multiple informants and multiple methods. We then suggest insights and strategies aimed at advancing approaches to measurement of children's eating behaviours. To progress our understanding of children's eating behaviours, we conclude that a range of psychometrically sound, fit-for-purpose measurement instruments and procedures are needed for use in multi-trait, multi-method, multi-informant studies in a range of populations and contexts.

Understanding Preschoolers' Anticipation of Trying A New Food and Past Food Experiences

OBJECTIVE

The primary objectives of the study were to explore (1) preschoolers' anticipation when trying a new food and (2) preschoolers' past experiences with the food they like and do not like.

METHODS

Two semistructured interviews were conducted 3-7 days apart with each participant. Through thematic analysis, themes and subthemes related to children's food-related perspectives were identified.

RESULTS

A total of 34 preschoolers (4-5 years old) from low-income families in North Carolina participated in the study. Three major themes were found: preschoolers' avoidance techniques, predictors behind preschoolers' food-related emotions, and preschoolers' description of their food preferences.

CONCLUSIONS AND IMPLICATIONS

Preschoolers were able to discuss both their positive and negative food-related emotions and perspectives. Thus, strategies allowing children to discuss their perspectives of food could provide new insights into children's food preferences and improve dietary quality.

Comparison of Parent and Child Ratings of Fruit and Vegetable Liking to Assess Parent Accuracy as Proxy Reporters

This study examined the accuracy of parent-report child fruit/vegetable (FV) liking. Child/parent dyads (n = 24) were recruited from six Head Start preschools in North Carolina. Liking for 10 FVs was assessed using a validated pictorial tool for children; a similar scale was used for parents. Negative relationships were observed between parent/child for one fruit (grapes) and one vegetable (broccoli). Positive relationships were observed among oranges, grapes, and overall fruit rankings. Parents tended to rank children's liking of fruits higher than their children, while children ranked liking vegetables higher. Findings suggest parents may not be accurate respondents for preschool children's FV liking.

Study of the reasons for the consumption of each type of vegetable within a population of school-aged children

BACKGROUND

Several studies have evaluated the existence of factors that influence the consumption of vegetables in children, such as family environment, daily exposure to one or several vegetables, parents' consumption and consumption pattern and the way in which vegetables are prepared in the household, among others. The objective of this study was to investigate the reasons for consumption associated with each vegetable in school-aged children through a survey designed to be answered by the parents.

METHODS

A preliminary study with 162 parents was carried out on the consumption of vegetables in children aged 6 to 12 years. Based on the information obtained, a survey was designed with 14 phrases to investigate the reasons for the low consumption of each type of vegetable among school-aged children, which was answered online by 419 parents.

RESULTS

The results obtained allowed us to categorise the vegetables into 6 groups. Group A consisted of tomatoes, corn, pumpkin and carrots as the vegetables that children like to eat most. Group B contains the vegetables that are consumed mostly camouflaged in other preparations, such as onions and red peppers. Group C contains only cauliflower, which was negatively associated with senses, such as colour, smell and taste. This vegetable was never offered to children by a high percentage of parents. Group D consists of green vegetables: zucchini, spinach, chard and peas. Vegetables of this group are added to other foods and the child usually ingests them camouflaged or obliged. Group E consists of beetroot, lettuce and broccoli. Beetroot and lettuce were the vegetables parents reported were most often rejected by their children. This rejection, they stated, was due to sensory aspects, such as colour, texture and taste.

CONCLUSIONS

The reasons for consumption among school-aged children depend on each type of vegetable and cannot be generalized. The sensory characteristics of the vegetable (mainly colour and flavour) and the habits of consumption in the family environment play a major role in children's acceptance or rejection of vegetables.

Mid-childhood fruit and vegetable consumption: The roles of early liking, early consumption, and maternal consumption

Previous studies have shown that early liking, early consumption, and maternal consumption of fruits and vegetables (F&V) each predict children's F&V consumption, but no one has examined the independent contributions of these three correlated factors. We aim to examine the extent to which each of these 3 factors is associated with F&V consumption in mid-childhood after accounting for the other 2 in the analysis. We analyzed data from 901 mother-child dyads from Project Viva, a prospective pre-birth cohort study. Mothers reported their child's early liking and consumption of F&V at age 2 years and later consumption at mid-childhood (median age 7.7 years). They also reported their own consumption of F&V at 6 months postpartum. We used multivariable linear regression models to examine the independent associations of these 3 factors with mid-childhood consumption, adjusting for socio-demographic, pregnancy, and child confounders. At 2 years, 53% of the mothers strongly agreed that their child liked fruit and 25% strongly agreed that their child liked vegetables. F&V consumption was 2.5 (1.3) and 1.8 (1.1) times/d at age 2 y and 1.5 (1.1) and 1.3 (0.8) times/d in mid-childhood. Maternal F&V consumption was 1.4 (1.1) and 1.5 (1.0) times/d, respectively. Children's early consumption played the most predominant role. For every 1 time/d increment in children's early consumption of F&V, mid-childhood consumption was higher by 0.25 (95% confidence interval [CI]: 0.19, 0.30) times/d for fruits and 0.21 (95% CI: 0.16, 0.26) times/d for vegetables, adjusted for confounders plus the other 2 exposures. In conclusion, children's early F&V consumption has the most significant influence on children's later consumption.

Early determinants of food liking among 5y-old children: a longitudinal study from the EDEN mother-child cohort

Background

Identifying the determinants of child’s liking for different foods may help to prevent future choices of unhealthy food.

Objective

To study early-life food-related characteristics associated with child’s liking for different foods at 5y with a longitudinal study.

Design

1142 5y- old children completed a liking test for “fruit and vegetables”, “meat, fish and eggs”, desserts and cheese. Data related to maternal food intake before pregnancy, infant feeding during the first year of life, maternal feeding practices at 2y, child’s food intake at 3y, and child’s food neophobia from 1 to 4y were collected prospectively from the mother. The associations between these factors and child‘s liking for each category of foods were analyzed using structural equation modelling.

Results

High food neophobia at 4 y was related to lower child’s liking for all food groups. Maternal feeding practices at 2y were associated with liking for dessert: negatively for the practices allowing child to control his/her own food intake, positively for restriction of child’s food intake for weight reasons. Moreover, child’s food intake at 3y was positively associated with child’s liking for “fruit and vegetables” as well as for cheese. Finally, adherence to the infant feeding pattern “long breastfeeding, later introduction of main meal components and use of home-made products” was positively associated with child’s liking for meat/fish/eggs.

Conclusions

For all food groups, food neophobia was a common determinant of child’s liking for food at 5y, whereas other factors were associated with food liking for specific food groups.

Electronic supplementary material

The online version of this article (doi:10.1186/s12966-016-0342-5) contains supplementary material, which is available to authorized users.

Impact of sensory-based food education in kindergarten on willingness to eat vegetables and berries

Background

Children use all of their senses when exploring new foods, and sensory-based food education provides new possibilities for promoting healthy dietary habits.

Objective

To evaluate the effect of sensory-based food education activities on children's willingness to eat test samples of selected vegetables and berries.

Design

Two kindergartens in Hanko, Finland, participated in the study and the subjects were children aged 3–6 years, divided in the intervention (n=44) and control (n=24) kindergarten. In the intervention kindergarten, five sensory-based food education sessions focusing on vegetables and berries were implemented, once per week for 5 weeks. A tasting protocol was performed with the children at baseline and after the intervention. The willingness to eat (5 different vegetables and 3 Finnish berries) was categorised. Parents also filled in a questionnaire on the children's food preferences at home.

Results

In the intervention kindergarten, the willingness to eat the samples increased significantly (p≤0.001, Wilcoxon and Friedman), while in the control kindergarten, no significant change was observed when all of the test samples were taken into account. The parental report of their children's preferences and children's actual eating of the test samples corresponded relatively weakly.

Conclusions

Sensory-based food education activities may promote a willingness to eat vegetables and berries. Child-centred test methods are important for evaluating the effects of dietary interventions among children.

Agreement between parent and child report on parental practices regarding dietary, physical activity and sedentary behaviours: the ENERGY cross-sectional survey

Background

Parents and their parenting practices play an important role in shaping their children’s environment and energy-balance related behaviours (EBRBs). Measurement of parenting practices can be parent- or child-informed, however not much is known about agreement between parent and child perspectives. This study aimed to assess agreement between parent and child reports on parental practices regarding EBRBs across different countries in Europe and to identify correlates of agreement.

Methods

Within the ENERGY-project, a cross-sectional survey was conducted among 10–12 year old children and their parents in eight European countries. Both children and parents filled in a questionnaire on 14 parental practices regarding five different EBRBs (i.e. soft drink, fruit juice and breakfast consumption, sports activity and watching TV) and socio-demographic characteristics. Children’s anthropometric measurements were taken at school. We calculated percentages of agreement between children and their parents and weighted kappa statistics (for ordinal variables) per practice and country and assessed factors associated with agreement using multilevel linear regression.

Results

Reports of 6425 children and their parents were available for analysis. Overall mean agreement between parent and child reports was 43% and varied little among countries. The lowest agreement was found for questions assessing joint parent–child activities, such as sports (27%; Kappa (κ) = 0.14) or watching TV (30%;κ = 0.17), and for parental allowance of the child to have soft drinks (32%;κ = 0.24) or fruit juices (32%;κ = 0.19), or to watch TV (27%;κ = 0.17). Having breakfast products available at home or having a TV in the child’s bedroom were the only practices with moderate to good agreement (>60%;κ = 0.06 and 0.77, respectively). In general, agreement was lower for boys, younger children, younger parents, parents with less than 14 years of education, single parents, parents with a higher self-reported body mass index and parents who perceived their child to be underweight.

Conclusions

Parents and children perceive parental practices regarding dietary, physical activity and sedentary behaviours differently in all parts of Europe, with considerable variation across specific practices and countries. Therefore, future studies should assess both, parents and children’s view on parental practices.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2458-14-918) contains supplementary material, which is available to authorized users.

Evaluation of a pictorial method to assess liking of familiar fruits and vegetables among preschool children

Research suggests fruit and vegetable (FV) preferences greatly influence on FV intake. Methods for assessing children's FV preference in a reliable and valid manner are needed. The purpose of this study was to develop a practical, reliable, and valid method for evaluating FV liking among preschool-aged children using photographs. Authors formatively assessed a series of digital FV photographs and hedonic scales to develop content for the liking measure. The measure for assessment included 20 high-quality, digital photographs presenting 13 FVs. A non-gendered 5-point face scale (super yummy to super yucky) was chosen to determine level of liking. We used this measure to establish reliability (i.e., test re-test) and concurrent validity (i.e., photograph versus tasting experience) of the pictorial method. Data were analyzed using Spearman's Rho Correlation Coefficients and Wilcoxon signed-rank tests. The measure demonstrated varying levels of reliability/validity for individual FV items and the fruit scale; however, the vegetable scale and collapsed FV scale were determined to be valid measures. Authors recommended the removal of one weak pictorial fruit item (halved peach) from the fruit and FV scales to improve validity. The final recommended measure included 19 high-quality, digital photographs presenting 12 FVs. The pictorial FV measure and subscales may be useful for assessing FV liking among groups of preschool-aged children within the studied population. Additional research is needed to further validate the use of the pictorial FV measures in a larger, more generalizable sample.

Test-retest reliability and agreement between children's and parents' reports of a computerized food preferences tool

OBJECTIVE

To investigate test-retest reliability of primary-school children's reports of food preferences and to investigate agreement with parental reports.

DESIGN

Children completed an online test and retest, one to two weeks later, during school hours; parents completed a paper-and-pencil or an online questionnaire at home. The children's preferences questionnaire contained 148 food items, reduced to twelve scales; the parental questionnaire contained seventy-eight items reduced to nine scales.

SETTING

Children of fourteen primary schools in Belgium-Flanders.

SUBJECTS

In total 572 children participated; test-retest data were available for 354 children, children's tests could be matched to 362 parental reports.

RESULTS

Test-retest intraclass correlations were on average 0.73, ranging between 0.62 and 0.86; correlations between children's and parents' reports were on average 0.50, ranging between 0.32 and 0.62. Retest preferences were significantly higher for more than half of the scales. Children reported higher preferences than their parents for milk & milk products, fruit and soft drinks, while parents reported higher preferences for bread & breakfast cereals, meat, snacks and sauces.

CONCLUSIONS

The results indicate that the test-retest stability was good; however, agreement between parents and children was rather low to moderate.

Permissive parental feeding behavior is associated with an increase in intake of low-nutrient-dense foods among American children living in rural communities

Parents play an important role in shaping children's eating habits. Few studies have evaluated the influence of both parenting style and parenting practices on child outcomes such as dietary intake. During spring 2007, 99 parent-child dyads from four rural US areas participated in this cross-sectional study. Child food intake was reported during two interviewer-administered, parent-assisted 24-hour recalls. Diet quality was defined as the average number of low-nutrient-dense (LND) foods consumed. Validated questionnaires were used to assess parental feeding practices and feeding style. Pearson correlations identified relationships among child food intake, parental feeding style typologies, and covariates. Regression analyses were used to predict child diet quality. Sixty percent of children and 76% of parents were overweight or obese. A permissive feeding style, which is highly responsive to a child's requests and sets few demands on him or her, was the most common (n=37) parental feeding style. This feeding style was associated with child intake of LND foods (r=0.3; P<0.001) and moderated the relationship between parental feeding practices and child intake of LND foods. In the presence of a permissive feeding style, higher levels of monitoring were associated with child intake of LND foods (β=.69; P<0.05). Parental feeding style may alter the effectiveness of parental feeding practices on children's food intake. More research is needed to understand the parent-child feeding relationship in the context of parental feeding styles and practices.