Abstract
Background
Early childhood is a critical period for the development of dietary behaviors, and parents and caregivers play an important role in shaping children’s food environments, feeding routines, and eating practices. Evidence from the West Bank remains limited regarding the combined relationships among parental nutrition knowledge, multidimensional feeding practices, and preschool children’s dietary behaviors. This study assessed parental nutrition knowledge and feeding practices and examined their associations with selected dietary behaviors among preschool children in the West Bank, Palestine.
Methods
This analytical cross-sectional study was conducted between March and May 2026 among 350 parents and primary caregivers of children aged 2–5 years residing in selected governorates of the northern West Bank. Participants completed a structured self-administered questionnaire assessing sociodemographic and child-related characteristics, parental nutrition knowledge, nine Feeding Practices and Structure Questionnaire-28 (FPSQ-28) domains, and five child dietary behaviors. Descriptive statistics, non-parametric bivariate analyses, multiple linear regression, and ordinal logistic regression were performed. Benjamini–Hochberg false discovery rate (FDR) correction was applied to the principal substantive exposure–outcome comparisons. Sensitivity analyses addressed the bounded parental knowledge score, ordinal coding of screen time, and violations of the proportional-odds assumption.
Results
Mean parental nutrition knowledge was 5.09 ± 0.83 out of 6, indicating generally high knowledge. Only 49.7% correctly identified that natural juice is not a healthy substitute for whole fruit. The FPSQ-28 demonstrated high internal consistency (Cronbach’s α = 0.918). In adjusted linear regression, higher child screen time was associated with a lower parental knowledge score (B = − 0.116, 95% CI: −0.210 to − 0.023, p = .015), and this association remained consistent in sensitivity analyses. After FDR correction in the dietary-behavior models, higher screen time remained associated with less frequent breakfast consumption and fruit and vegetable intake and with more frequent unhealthy snack and sweetened drink consumption. Higher healthy food availability and positive reinforcement were associated with more frequent fruit and vegetable intake, whereas pressure to eat was associated with less frequent fruit and vegetable intake. Food as reward was associated with more frequent sweetened drink consumption, restriction with less frequent unhealthy snack consumption, and variety encouragement with more frequent unhealthy snack consumption and lower reported food refusal. Associations between parental nutrition knowledge and the dietary outcomes did not remain statistically supported after FDR correction.
Conclusions
Parental nutrition knowledge was generally high, but feeding practices and child dietary behaviors varied. Child screen time was the most consistent correlate of dietary behaviors, while selected feeding-practice domains showed behavior-specific associations. The findings suggest that family-centered nutrition approaches may benefit from combining nutrition information with practical feeding guidance, supportive home food environments, structured eating routines, attention to screen-related behaviors, and involvement of caregivers in early childhood nutrition. Because of the cross-sectional design, causal or directional relationships cannot be inferred.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-29092-4.
Keywords: Preschool children, Parental nutrition knowledge, Feeding practices, Dietary behaviors, Screen time, West Bank, Palestine
Introduction
Early childhood, particularly between two and five years of age, is a critical period for establishing dietary preferences, eating routines, and food-related behaviors that may persist into later childhood. During this period, children become increasingly independent in their food choices while remaining strongly influenced by their parents, caregivers, and home food environment. Adequate nutrition is important for physical growth, cognitive development, immune function, and the prevention of both undernutrition and excessive weight gain [1, 2].
Parents and primary caregivers may influence preschool children’s dietary behaviors through the foods made available at home, meal timing and structure, exposure to unfamiliar foods, responses to food refusal, and the feeding strategies used during meals. However, children’s dietary behaviors may also be related to household resources, food availability, childcare arrangements, child preferences, screen exposure, and established family routines. Consequently, parental nutrition knowledge alone may not consistently correspond to favorable child dietary behaviors [1, 3].
Parental feeding practices represent another aspect of the family food environment that may be associated with children’s dietary behaviors. These practices should not be classified uniformly as “healthy” or “unhealthy,” because their interpretation depends on the specific feeding domain and dietary behavior being assessed. Parent–child feeding relationships may also be bidirectional: caregivers may influence children’s eating behaviors, while children’s existing preferences, food refusal, or eating patterns may influence the feeding strategies adopted by caregivers [3]. In this study, feeding practices were therefore examined as separate domains using the Feeding Practices and Structure Questionnaire-28 rather than combined into a single overall feeding-practice score [4].
In this study, feeding practices were not classified collectively as “healthy” or “unhealthy.” Instead, the nine FPSQ-28 domains—pressure to eat, food as reward, structured meal routine, healthy food availability, distraction-free eating, responsive feeding, restriction, variety encouragement, and positive reinforcement—were examined and interpreted separately. The terms “more favorable” and “less favorable” were used only to describe the direction of the selected child dietary behaviors. More frequent breakfast consumption and fruit and vegetable intake were considered more favorable behaviors, whereas more frequent unhealthy snack consumption, sweetened drink consumption, and stronger reported refusal to try new foods were considered less favorable behaviors.
The Palestinian context is particularly relevant to the study of early childhood nutrition. Available evidence has identified coexisting concerns related to undernutrition, micronutrient deficiencies, overweight and obesity, limited dietary diversity, and socioeconomic inequalities in access to adequate nutrition. However, the available evidence is geographically uneven and has more commonly focused on nutritional status, breastfeeding, micronutrient deficiencies, or household food insecurity than on the combined relationships among caregiver knowledge, feeding practices, and preschool children’s dietary behaviors [5, 6].
Nutrition and feeding in the West Bank occur within a broader environment shaped by household income, food prices and availability, childcare arrangements, and differences in access to health and nutrition information. Recent Palestinian evidence among children aged 6–59 months has shown that nutritional status is related to interconnected socioeconomic circumstances, dietary intake, early feeding practices, and food availability [7]. These circumstances reinforce the importance of examining parental knowledge and feeding practices within the context in which families make daily food-related decisions.
Despite this context, evidence remains limited regarding how parental nutrition knowledge and multidimensional feeding practices are associated with specific dietary behaviors among preschool children in the West Bank. Examining these constructs together may help distinguish general nutrition awareness from the feeding strategies, family routines, and child-related characteristics associated with everyday dietary behaviors.
In the present study, parental nutrition knowledge and the individual FPSQ-28 feeding-practice domains were examined in relation to five separate child dietary behaviors: breakfast consumption, fruit and vegetable intake, unhealthy snack consumption, sweetened drink consumption, and refusal to try new foods. Sociodemographic and child-related characteristics, childcare setting, and child screen time were also examined as associated factors. Feeding practices were not evaluated as formal mediators or effect modifiers, and the study did not test a causal pathway among these variables.
Accordingly, this study aimed to assess parental nutrition knowledge and feeding practices among parents and primary caregivers of preschool children in selected governorates of the northern West Bank and to examine their associations with selected child dietary behaviors. The study also examined associations between participant and child characteristics, parental nutrition knowledge, and feeding-practice domain scores. Given the cross-sectional design, all findings were interpreted as associations rather than causal relationships.
Specifically, this study addressed the following research questions: [1] What are the levels of parental nutrition knowledge, feeding-practice domain scores, and child dietary behaviors among parents and caregivers of preschool children in selected governorates of the northern West Bank [2]? Which participant and child-related characteristics are independently associated with parental nutrition knowledge scores and FPSQ-28 feeding-practice domain scores [3]? Which participant, child-related, and feeding-practice factors are independently associated with preschool children’s dietary behaviors, including breakfast consumption, fruit and vegetable intake, unhealthy snack consumption, sweetened drink consumption, and refusal to try new foods?
Methods
Study design, setting, and population
This analytical cross-sectional study was conducted between March and May 2026 among parents and primary caregivers residing in selected governorates of the northern West Bank, Palestine, including Nablus, Jenin, Tulkarem, and Tubas. The questionnaire was distributed electronically through social media platforms and WhatsApp groups and in paper format through community settings in the included governorates. Convenience sampling was used, and the sample was not intended to be geographically representative of all parents and caregivers in the West Bank.A primary caregiver was defined as an adult with regular responsibility for the child’s feeding, food provision, or meal preparation. Eligible participants were aged 18 years or older, resided in one of the included northern West Bank governorates, and cared for at least one child aged 2–5 years at the time of data collection. Caregivers were excluded if their child had a medical or developmental condition expected to substantially alter usual dietary requirements or feeding patterns. These conditions included metabolic disorders requiring a therapeutic diet, severe food allergies requiring a medically restricted diet, dependence on enteral tube feeding, or severe neurodevelopmental conditions substantially affecting oral feeding. Eligibility and exclusion criteria were assessed through screening questions completed by the parent or caregiver. Reported conditions were not independently verified using medical records. The study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations for cross-sectional studies, and the completed checklist is provided as Supplementary File 1 [8].
Sample size
The initial recruitment target was estimated using a single-proportion precision approach, assuming a 95% confidence level, 5% absolute precision, and a conservative expected proportion of 50%. This calculation was used pragmatically to guide recruitment and was not intended as a formal power calculation for the multivariable regression analyses. Because convenience sampling was used, the margin-of-error calculation should not be interpreted as indicating population representativeness, and no finite population correction was applied. A total of 350 eligible participants were included in the descriptive analyses. No formal regression-based sample-size calculation was conducted before data collection. To evaluate the adequacy of the achieved sample, model complexity, complete-case sample sizes, outcome-category frequencies, confidence-interval widths, and model diagnostics were considered. Given that no formal regression-based sample-size calculation was conducted before data collection, estimates from multivariable models involving small outcome categories or wide confidence intervals were interpreted cautiously.
Sampling, recruitment, and data collection
A convenience sampling approach was used. Data were collected using a structured self-administered questionnaire distributed through two parallel methods: an online version shared through social media platforms and WhatsApp groups and a paper-based version distributed in community settings in the included governorates. The online and paper versions contained the same questions, wording, order, response options, eligibility criteria, consent information, and participant instructions. Participants completing the paper questionnaire were provided with a private environment in which to complete it. Using both approaches was intended to improve access to participants from different geographic and community settings.
Potential participants were invited to take part voluntarily after receiving information about the study objectives, eligibility criteria, confidentiality of responses, and estimated time required to complete the questionnaire. Participation was anonymous, and no identifying personal information was collected. For parents or caregivers with more than one child aged 2–5 years, participants were instructed to answer the child-related questions in relation to one eligible child only.
Each caregiver completed one questionnaire concerning one eligible child only. When a caregiver had more than one child within the eligible age range, questionnaire responses referred to the youngest eligible child. Because no household identifier was collected, participation by more than one caregiver from the same household was possible, and potential clustering of responses within households could not be assessed.
Data collection was conducted between March and May 2026. Submitted questionnaires were screened for eligibility before analysis. One response was excluded because the child’s age was outside the prespecified eligible range. No questionnaires were excluded because of missing data on the principal study outcomes. For multivariable analyses, complete-case analysis was used when data were missing for one or more variables included in a particular model, and the analytical sample size is reported in the corresponding table.
Questionnaire and study measures
Data were collected using a structured self-administered questionnaire that included sociodemographic and child-related items, six nutrition-knowledge items, selected child dietary-behavior items, and the established Feeding Practices and Structure Questionnaire-28 (FPSQ-28) [4, 9].
The questionnaire included four sections; The first section collected sociodemographic and child-related information, including parent/caregiver age, sex, educational level, occupation, number of children, monthly household income, child age, childcare setting, and daily child screen time. Child age was categorized into three groups: 2–3 years, 3–4 years, and 4–5 years.
The second section assessed parental nutrition knowledge using six factual items related to healthy nutrition for preschool children. Items addressed daily fruit and vegetable intake, fruit juice versus whole fruit, dairy intake, fast-food consumption, breakfast, and reducing salt and added sugar in children’s food. Each correct response was scored as 1, and each incorrect or “I do not know” response was scored as 0. The total knowledge score ranged from 0 to 6, with higher scores indicating greater nutrition knowledge.
The third section assessed parental feeding practices using the Feeding Practices and Structure Questionnaire-28 (FPSQ-28), a 28-item instrument assessing feeding-practice structure and behaviors among parents of young children [4].
The FPSQ-28 includes nine feeding-practice domains: pressure to eat, food as reward, structured meal routine, healthy food availability, distraction-free eating, responsive feeding, restriction, variety encouragement, and positive reinforcement. Items are rated on a five-point Likert scale ranging from 1 (never) to 5 (always). Mean scores were calculated for each domain, with higher scores indicating greater use of the corresponding feeding practice.
The fourth section assessed five selected child dietary behaviors: breakfast consumption, fruit and vegetable intake, unhealthy snack consumption, sweetened drink consumption, and refusal to try new foods. These behaviors were analyzed separately rather than combined into a single score. For interpretation, more frequent breakfast consumption and fruit and vegetable intake represented more favorable dietary behaviors. More frequent unhealthy snack consumption and sweetened drink consumption and stronger agreement that the child refused to try new foods represented less favorable dietary behaviors.
Translation, pilot testing, validity, and reliability
The questionnaire was initially developed in English and subsequently translated into Arabic for administration to the target population. The Arabic version was reviewed by the research team for linguistic clarity, conceptual equivalence, and cultural relevance. The translated version was then reviewed by experts in public health, nutrition, and child health to assess face and content validity.
A pilot study was conducted among 25 parents/caregivers who met the eligibility criteria to evaluate the clarity, readability, and acceptability of the questionnaire. Feedback from the pilot study and expert review was used to refine wording and improve the final version of the questionnaire. Participants included in the pilot study were excluded from the final analysis.
Internal consistency was assessed using Cronbach’s alpha for the full FPSQ-28 and its multi-item domains, with values of 0.70 or higher considered acceptable. In the final sample, the full FPSQ-28 demonstrated excellent internal consistency (Cronbach’s α = 0.918), and the multi-item FPSQ-28 domains showed acceptable to good internal consistency, with alpha values ranging from 0.767 to 0.887. Cronbach’s alpha was not calculated for the single-item positive reinforcement domain. Internal consistency was not assessed for the six-item nutrition knowledge index because it consisted of factual items covering distinct nutrition domains rather than a unidimensional psychometric scale. The child dietary behavior items showed low internal consistency (Cronbach’s α = 0.517) and were therefore analyzed separately rather than as a combined scale.
Ethical approval
Ethical approval was obtained from the Institutional Review Board of An-Najah National University before data collection (Approval No. Med.Nov.2025/174). The study was conducted in accordance with the principles of the Declaration of Helsinki.
All participants received information about the study objectives, voluntary nature of participation, confidentiality of responses, and their right to withdraw before submitting the questionnaire. Electronic or written informed consent was obtained from all participants before participation. No identifying personal information was collected, and all responses were treated confidentially and used only for research purposes.
Children did not complete questionnaires, interact with the research team, or undergo any study-related procedures. Adult respondents provided anonymous proxy reports concerning the dietary behaviors of one eligible child. Child assent was therefore not sought. Electronic or written informed consent was obtained from all participating adult parents and caregivers before completion of the questionnaire, in accordance with the protocol approved by the Institutional Review Board.
Statistical analysis
Data were analyzed using IBM SPSS Statistics. Descriptive statistics were used to summarize participant and child characteristics, parental nutrition knowledge, FPSQ-28 feeding-practice domain scores, child dietary behaviors, and internal consistency of the study measures. Categorical variables were presented as frequencies and percentages, while continuous and scale-based variables were summarized using means and standard deviations or medians and interquartile ranges, as appropriate. Valid denominators were reported for variables with missing data.
Parental nutrition knowledge was assessed using six factual items, with each correct response scored as 1 and each incorrect or “I do not know” response scored as 0, yielding a total score ranging from 0 to 6. FPSQ-28 feeding-practice domains were analyzed as mean scores ranging from 1 to 5. Breakfast consumption, fruit and vegetable intake, unhealthy snack consumption, sweetened drink consumption, and refusal to try new foods were analyzed as separate ordinal outcomes. Internal consistency was assessed using Cronbach’s alpha for the full FPSQ-28 and its multi-item domains, with values ≥ 0.70 considered acceptable. Cronbach’s alpha was not calculated for the single-item positive reinforcement domain or for the six-item nutrition knowledge index, which comprised factual items covering distinct nutrition topics. The five child dietary-behavior items were analyzed separately because they represented distinct behaviors rather than a single underlying construct.
Normality of the total nutrition knowledge score and FPSQ-28 domain scores was assessed using the Shapiro–Wilk test. Because the knowledge score and several study variables were ordinal or non-normally distributed, non-parametric methods were used for bivariate analyses where appropriate. Associations between participant and child characteristics and parental nutrition knowledge were examined using the Mann–Whitney U test for binary variables, the Kruskal–Wallis H test for nominal variables with three or more categories, and Spearman’s rank correlation for ordered variables. Specifically, sex was examined using the Mann–Whitney U test; occupation and childcare setting using Kruskal–Wallis H tests; and parent age group, education level, number of children, household income, child age group, and child screen time using Spearman’s rank correlation. Associations between parental nutrition knowledge and child dietary behaviors were assessed using Kruskal–Wallis H tests. Spearman’s rank correlation was also used to examine associations between parental nutrition knowledge and FPSQ-28 domains and correlations among FPSQ-28 domains. Associations between FPSQ-28 domain scores and each child dietary behavior were assessed using Kruskal–Wallis H tests. These bivariate analyses were considered exploratory.
Multiple linear regression was used to examine factors associated with parental nutrition knowledge. The total knowledge score was the dependent variable, and sex, parent age rank, education rank, occupation, number of children rank, household income rank, child age rank, childcare setting, and child screen time rank were included as independent variables. Female sex, homemaker occupation, and home childcare were reference categories. Ordered predictors were entered as rank variables. Parent age, education level, number of children, monthly household income, and child screen time had been collected using predefined ordered response categories selected during questionnaire development for local relevance and practical interpretability; child age was collected as free text and subsequently categorized for analysis. Unstandardized regression coefficients (B), 95% confidence intervals, and p-values were reported. Because the knowledge score was bounded and concentrated toward the upper end of its range, results from the linear model were interpreted cautiously.
Separate multiple linear regression models were fitted for each of the nine FPSQ-28 domain scores: pressure to eat, food as reward, structured meal routine, healthy food availability, distraction-free eating, responsive feeding, restriction, variety encouragement, and positive reinforcement. Each domain score was modeled as the dependent variable, with sex, parent age rank, education rank, occupation, number of children rank, household income rank, child age rank, childcare setting, child screen time rank, and total parental nutrition knowledge score as independent variables. Female sex, homemaker occupation, and home childcare were reference categories. Unstandardized regression coefficients were presented in the main manuscript, with 95% confidence intervals provided in the supplementary material. Multicollinearity was assessed using variance inflation factors, and model explanatory power was summarized using R².
Separate ordinal logistic regression models were used for breakfast consumption, fruit and vegetable intake, unhealthy snack consumption, sweetened drink consumption, and refusal to try new foods. Independent variables were sex, parent age rank, education rank, household income rank, childcare setting, child screen time rank, parental nutrition knowledge score, and the nine FPSQ-28 domain scores. Female sex and home childcare were reference categories. For breakfast consumption and fruit and vegetable intake, adjusted odds ratios (aORs) greater than 1 indicated higher odds of being in a more frequent favorable-behavior category. For unhealthy snack consumption and sweetened drink consumption, aORs greater than 1 indicated higher odds of being in a more frequent unfavorable-behavior category; for food refusal, aORs greater than 1 indicated higher odds of stronger agreement that the child refused new foods. Model explanatory power was summarized using McFadden’s pseudo-R².
The proportional-odds assumption was assessed for each ordinal logistic regression model. When evidence of violation was identified, multinomial logistic regression using the same predictor set was performed as a sensitivity analysis. For fruit and vegetable intake, the “never” and “rarely” categories were combined in the multinomial model because only 11 participants reported “never,” resulting in sparse estimates when modeled separately. In addition, because child screen time was entered as an ordinal-rank predictor in the primary models, sensitivity analyses re-estimated screen time as a categorical predictor using the lowest screen-time category as the reference group. The direction and pattern of estimates were compared with those from the primary rank-coded models to assess sensitivity to the assumption of a linear trend across ordered categories.
Complete-case analysis was used for multivariable models, with participants excluded from a given model when data were missing for one or more included variables. Model-specific analytical sample sizes are reported in the corresponding tables or table notes. Because missing child-age data reduced the sample available for the primary parental nutrition knowledge regression, that model was refitted without child age, allowing all 350 participants to be included. Because the knowledge score was bounded and showed clustering toward higher values, an additional ordinal logistic regression sensitivity analysis was performed in which scores of 1–4 were combined into one category and scores of 5 and 6 were retained as separate ordered categories. This model used the same covariates as the complete-sample sensitivity model and included all 350 participants.
All statistical tests were two-sided. To address multiplicity, analyses were organized according to their role in the study. The five multivariable dietary-behavior models were treated as the principal multivariable analyses. Benjamini–Hochberg false discovery rate (FDR) correction was applied across 55 substantive exposure–outcome tests comprising child screen time, parental nutrition knowledge, and the nine FPSQ-28 feeding-practice domains across the five dietary-behavior outcomes. Sociodemographic and childcare variables included primarily for covariate adjustment were not included in this multiplicity family. The nine regression models examining correlates of individual FPSQ-28 domain scores were treated as secondary analyses and were reported with effect estimates and 95% confidence intervals without formal multiplicity adjustment. The 45 bivariate FPSQ-28 domain–dietary-behavior comparisons were considered exploratory and were subjected to Benjamini–Hochberg FDR correction as a separate family. For analyses subjected to FDR correction, both nominal p-values and FDR-adjusted q-values were reported. Effect estimates, confidence intervals, consistency across analyses, and findings from sensitivity analyses were considered alongside statistical significance when interpreting the results.
Results
Participant and child-related characteristics
A total of 351 questionnaire responses were assessed for eligibility. One response was excluded because the child’s age was outside the eligible range, leaving 350 participants in the final descriptive sample. The analytical sample was smaller in regression models requiring complete information on all included variables.
Most participants were female (n = 274, 78.3%), and the largest age group was 25–34 years (n = 170, 48.6%). More than half of participants had a bachelor’s degree (n = 190, 54.3%), and nearly half were homemakers (n = 173, 49.4%). The most frequently reported monthly household income category was 1,800–3,000 ILS (n = 115, 32.9%). Most children were cared for either at home (n = 150, 42.9%) or in private childcare settings (n = 144, 41.1%). Daily screen time was most commonly 1–2 h (n = 122, 34.9%). Child age group data were available for 337 participants; percentages for this variable are reported using the valid denominator (Table 1).
Table 1.
Participant and child-related characteristics of the sample (N = 350)
| Variable | Category | N | % |
|---|---|---|---|
| Age | |||
| < 25 | 55 | 15.7 | |
| 25–34 | 170 | 48.6 | |
| 35–44 | 88 | 25.1 | |
| 45+ | 37 | 10.6 | |
| Sex | |||
| Female | 274 | 78.3 | |
| Male | 76 | 21.7 | |
| Education | |||
| Secondary or less | 71 | 20.3 | |
| Diploma | 41 | 11.7 | |
| Bachelor’s degree | 190 | 54.3 | |
| Postgraduate | 48 | 13.7 | |
| Occupation | |||
| Homemaker | 173 | 49.4 | |
| Employee | 120 | 34.3 | |
| Self-employed | 57 | 16.3 | |
| Number of children | |||
| 1 | 66 | 18.9 | |
| 2 | 97 | 27.7 | |
| 3 | 79 | 22.6 | |
| 4 | 58 | 16.6 | |
| 5 | 28 | 8.0 | |
| > 5 | 22 | 6.3 | |
| Monthly income (ILS) | |||
| < 1,800 | 55 | 15.7 | |
| 1,800-3,000 | 115 | 32.9 | |
| 3,000–6,500 | 99 | 28.3 | |
| 6,500 − 12,000 | 34 | 9.7 | |
| > 12,000 | 47 | 13.4 | |
| Child age group | |||
| (valid n = 337) | 2–3 years | 78 | 23.1 |
| 3–4 years | 77 | 22.8 | |
| 4–5 years | 182 | 54.0 | |
| Childcare setting | |||
| Home | 150 | 42.9 | |
| Government | 56 | 16.0 | |
| Private | 144 | 41.1 | |
| Daily screen time | |||
| < 1 h | 66 | 18.9 | |
| 1–2 h | 122 | 34.9 | |
| 2–3 h | 98 | 28.0 | |
| > 3 h | 64 | 18.3 | |
Values are presented as n (%) unless otherwise indicated. Percentages were calculated using the full sample denominator of 350, except for child age group, which was calculated using the valid denominator of 337. ILS, Israeli shekel
Parental nutrition knowledge
Parental nutrition knowledge scores were high overall. The mean total knowledge score was 5.09 ± 0.83 out of 6. The most frequently correct items were the importance of breakfast for child growth and development (n = 346, 98.9%) and daily fruit and vegetable intake (n = 336, 96.0%). The least frequently correct item was the statement that natural juice is a healthy substitute for whole fruit, for which 174 participants (49.7%) gave the correct response (Table 2).
Table 2.
Parental knowledge about healthy child nutrition (N = 350)
| Item | n | Correct response, n (%) |
|---|---|---|
| Children should eat daily servings of fruits and vegetables | 336 | 96.0 |
|
Natural juice is a healthy substitute for whole fruit (Correct answer = False) |
174 | 49.7 |
| Children need daily dairy products for calcium | 292 | 83.4 |
| Fast food negatively affects child growth | 318 | 90.9 |
| Breakfast is important for child growth and development | 346 | 98.9 |
| Salt and added sugar should be reduced in children’s food | 314 | 89.7 |
Correct responses are presented as n (%). The total knowledge score ranged from 0 to 6, with higher scores indicating greater nutrition knowledge. Total knowledge score: mean = 5.09, SD = 0.83. The correct response for the item “Natural juice is a healthy substitute for whole fruit” was “False. “
Feeding-practice dimensions
Mean FPSQ-28 domain scores ranged from 3.09 ± 1.19 for pressure to eat to 4.37 ± 0.95 for positive reinforcement. The highest mean scores were observed for positive reinforcement, healthy food availability (4.34 ± 0.83), and variety encouragement (4.23 ± 0.82). Lower mean scores were observed for food as reward (3.52 ± 1.04) and pressure to eat (3.09 ± 1.19) compared with most other feeding-practice dimensions (Table 3).
Table 3.
FPSQ-28 feeding-practice domain scores (N = 350)
| Domain | Items | M | SD |
|---|---|---|---|
| Pressure to eat | 3 | 3.09 | 1.19 |
| Food as reward | 3 | 3.52 | 1.04 |
| Structured meal routine | 3 | 3.59 | 0.97 |
| Healthy food availability | 3 | 4.34 | 0.83 |
| Distraction-free eating | 3 | 3.98 | 1.01 |
| Responsive feeding | 3 | 4.05 | 0.84 |
| Restriction | 6 | 3.89 | 0.83 |
| Variety encouragement | 3 | 4.23 | 0.82 |
| Positive reinforcement | 1 | 4.37 | 0.95 |
Values are presented as mean and standard deviation. Each FPSQ−28 domain score ranged from 1 to 5, with higher scores indicating greater use of the corresponding feeding practice. FPSQ−28, Feeding Practices and Structure Questionnaire−28
Child dietary behaviors
Breakfast before school or before starting the day was reported as “always” by 153 participants (43.7%), whereas 32 participants (9.1%) reported that their child never ate breakfast. Daily fruit and vegetable intake was reported as “always” by 205 participants (58.6%). Unhealthy snack consumption more than five times per week was reported for 65 children (18.6%), and daily sweetened drink consumption was reported for 40 children (11.4%). Refusal to try new foods was reported as “agree” or “strongly agree” by 112 participants (32.0%) (Table 4).
Table 4.
Child dietary behaviors (N = 350)
| Variable | Category | n | % |
|---|---|---|---|
| Breakfast before school | |||
| Always | 153 | 43.7 | |
| Sometimes | 113 | 32.3 | |
| Rarely | 52 | 14.9 | |
| Never | 32 | 9.1 | |
| Daily fruits and vegetables | |||
| Always | 205 | 58.6 | |
| Sometimes | 113 | 32.3 | |
| Rarely | 21 | 6.0 | |
| Never | 11 | 3.1 | |
| Unhealthy snacks per week | |||
| 0–1 | 60 | 17.1 | |
| 2–3 | 131 | 37.4 | |
| 4–5 | 94 | 26.9 | |
| > 5 | 65 | 18.6 | |
| Sweetened drinks | |||
| Rarely | 126 | 36.0 | |
| 1–2 times/week | 101 | 28.9 | |
| 3–5 times/week | 83 | 23.7 | |
| Daily | 40 | 11.4 | |
| Refuses new foods | |||
| Strongly agree | 31 | 8.9 | |
| Agree | 81 | 23.1 | |
| Neutral | 135 | 38.6 | |
| Disagree | 63 | 18.0 | |
| Strongly disagree | 40 | 11.4 | |
Values are presented as n (%). Breakfast consumption and fruit/vegetable intake were treated as favorable ordinal behaviors, with higher frequency indicating more favorable behavior. Unhealthy snack consumption and sweetened drink consumption were treated as unfavorable ordinal behaviors, with higher frequency indicating less favorable behavior. For refusal to try new foods, stronger agreement indicated greater reported food refusal
Internal consistency of study instruments
The full FPSQ-28 demonstrated high internal consistency in the current sample (Cronbach’s α = 0.918). Internal consistency for the multi-item FPSQ-28 domains ranged from 0.767 for food as reward to 0.887 for healthy food availability. Cronbach’s alpha was not applicable for the single-item positive reinforcement domain. The five child dietary behavior items showed low internal consistency (Cronbach’s α = 0.517), supporting their analysis as separate behaviors rather than as a combined scale (Table 5).
Table 5.
Internal consistency of study instruments
| Scale | Items | α |
|---|---|---|
| Pressure to eat | 3 | 0.804 |
| Food as reward | 3 | 0.767 |
| Structured meal routine | 3 | 0.811 |
| Healthy food availability | 3 | 0.887 |
| Distraction-free eating | 3 | 0.800 |
| Responsive feeding | 3 | 0.783 |
| Restriction | 6 | 0.827 |
| Variety encouragement | 3 | 0.842 |
| Positive reinforcement | 1 | -- |
| Full FPSQ-28 | 28 | 0.918 |
| Child dietary behaviors | 5 | 0.517 |
Cronbach’s alpha was calculated for the full FPSQ−28 and its multi−item domains. Internal consistency was not assessed for the six−item nutrition knowledge index because it consisted of factual items covering distinct nutrition domains rather than a unidimensional psychometric scale. Cronbach’s alpha was not applicable for the single−item positive reinforcement domain
Bivariate associations
In bivariate analyses, parental nutrition knowledge score was associated with education level, household income, childcare setting, and child screen time. Higher education and household income were positively correlated with knowledge score, whereas higher child screen time was negatively correlated with knowledge score. Knowledge score was also associated with fruit and vegetable intake, unhealthy snack consumption, and sweetened drink consumption, but not with breakfast consumption or refusal to try new foods (Tables 6 and 7).
Table 6.
Association between participant characteristics and parental nutrition knowledge score
| Variable | Category | Mdn (IQR) | Statistic | p |
|---|---|---|---|---|
| Sex | U = 11242.0 | 0.246 | ||
| Female | 5.0 (5.0–6.0) | |||
| Male | 5.0 (5.0–5.0) | |||
| Age group | rs = 0.074 | 0.165 | ||
| < 25 | 5.0 (4.0–5.0) | |||
| 25–34 | 5.0 (5.0–6.0) | |||
| 35–44 | 5.0 (5.0–6.0) | |||
| 45+ | 5.0 (4.0–5.0) | |||
| Education | rs = 0.137 | 0.010* | ||
| Secondary or less | 5.0 (5.0–5.0) | |||
| Diploma | 5.0 (5.0–5.0) | |||
| Bachelor’s | 5.0 (5.0–6.0) | |||
| Postgraduate | 5.0 (5.0–6.0) | |||
| Occupation | H = 2.463 | 0.292 | ||
| Homemaker | 5.0 (5.0–6.0) | |||
| Employee | 5.0 (5.0–6.0) | |||
| Self-employed | 5.0 (5.0–5.0) | |||
| Number of children | rs = 0.027 | 0.617 | ||
| 1 | 5.0 (5.0–5.0) | |||
| 2 | 5.0 (5.0–6.0) | |||
| 3 | 5.0 (5.0–6.0) | |||
| 4 | 5.0 (5.0–6.0) | |||
| 5 | 5.0 (5.0–5.0) | |||
| > 5 | 5.0 (5.0-5.8) | |||
| Income (ILS) | rs = 0.179 | < 0.001*** | ||
| < 1,800 | 5.0 (5.0–5.0) | |||
| 1,800-3,000 | 5.0 (4.0–6.0) | |||
| 3,000–6,500 | 5.0 (5.0–6.0) | |||
| 6,500 − 12,000 | 5.0 (5.0–6.0) | |||
| > 12,000 | 5.0 (5.0–6.0) | |||
| Childcare setting | H = 6.256 | 0.044* | ||
| Home | 5.0 (5.0–5.0) | |||
| Government | 5.0 (5.0–6.0) | |||
| Private | 5.0 (5.0–6.0) | |||
| Screen time | rs = -0.144 | 0.007** | ||
| < 1 h | 5.0 (5.0–6.0) | |||
| 1–2 h | 5.0 (5.0–6.0) | |||
| 2–3 h | 5.0 (5.0-5.8) | |||
| > 3 h | 5.0 (4.8-6.0) |
Values are presented as median and interquartile range unless otherwise indicated. Spearman’s rank correlation was used for ordinal variables; Mann–Whitney U test was used for binary variables; and Kruskal–Wallis H test was used for nominal variables with three or more categories. Valid denominators varied according to missing data. Mdn, median; IQR, interquartile range; rs, Spearman correlation coefficient. *p < .05, **p < .01, ***p < .001
Table 7.
Association between parental nutrition knowledge score and child dietary behaviors
| Behavior | Category | Mdn (IQR) | H | p |
|---|---|---|---|---|
| Fruits/vegetables | 14.966 | 0.002** | ||
| Never | 5.0 (4.0–5.0) | |||
| Rarely | 5.0 (5.0–5.0) | |||
| Sometimes | 5.0 (5.0–6.0) | |||
| Always | 5.0 (5.0–6.0) | |||
| Unhealthy snacks | 10.734 | 0.013* | ||
| 0–1 | 5.0 (5.0–6.0) | |||
| 2–3 | 5.0 (5.0–6.0) | |||
| 4–5 | 5.0 (5.0–6.0) | |||
| > 5 | 5.0 (4.0–5.0) | |||
| Sweetened drinks | 8.440 | 0.038* | ||
| Rarely | 5.0 (5.0–6.0) | |||
| 1–2 times/wk | 5.0 (5.0–6.0) | |||
| 3–5 times/wk | 5.0 (5.0–6.0) | |||
| Daily | 5.0 (4.0–5.0) | |||
| Breakfast | 3.931 | 0.269 | ||
| Never | 5.0 (4.8-5.0) | |||
| Rarely | 5.0 (5.0–6.0) | |||
| Sometimes | 5.0 (5.0–6.0) | |||
| Always | 5.0 (5.0–6.0) | |||
| Refuses new foods | 8.326 | 0.080 | ||
| Str. disagree | 5.0 (5.0–6.0) | |||
| Disagree | 5.0 (5.0–6.0) | |||
| Neutral | 5.0 (5.0-5.5) | |||
| Agree | 5.0 (5.0–6.0) | |||
| Str. agree | 5.0 (5.0–6.0) |
Values are presented as median and interquartile range. Kruskal–Wallis H tests were used because child dietary behaviors were analyzed as ordinal outcomes. Mdn, median; IQR, interquartile range. *p < .05, **p < .01, ***p < .001
Parental knowledge score showed weak but statistically significant correlations with several FPSQ-28 domains. Higher knowledge was correlated with lower pressure to eat and lower use of food as reward, and with higher healthy food availability, responsive feeding, restriction, variety encouragement, and positive reinforcement. Correlations among FPSQ-28 domains were generally positive, indicating that several feeding-practice dimensions tended to co-occur. Because these bivariate analyses were exploratory and included multiple comparisons, findings were interpreted cautiously (Table 8).
Table 8.
Spearman correlation matrix for knowledge score and FPSQ-28 feeding-practice domains
| Knowledge | Press. | Reward | Routine | Avail. | Distr. | Resp. | Restr. | Variety | Reinf. | |
|---|---|---|---|---|---|---|---|---|---|---|
| Knowledge | -- | -0.185*** | -0.122* | 0.049 | 0.142** | 0.097 | 0.135* | 0.209*** | 0.153** | 0.227*** |
| Press. | -- | 0.534*** | 0.235*** | 0.169** | 0.054 | 0.150** | 0.178*** | 0.224*** | 0.087 | |
| Reward | -- | 0.257*** | 0.265*** | 0.106* | 0.330*** | 0.224*** | 0.278*** | 0.183*** | ||
| Routine | -- | 0.376*** | 0.346*** | 0.355*** | 0.454*** | 0.284*** | 0.240*** | |||
| Avail. | -- | 0.409*** | 0.526*** | 0.512*** | 0.603*** | 0.444*** | ||||
| Distr. | -- | 0.458*** | 0.474*** | 0.418*** | 0.322*** | |||||
| Resp. | -- | 0.548*** | 0.480*** | 0.423*** | ||||||
| Restr. | -- | 0.506*** | 0.434*** | |||||||
| Variety | -- | 0.599*** | ||||||||
| Reinf. | -- |
Values are Spearman correlation coefficients. Only the upper triangle of the matrix is shown. Press., pressure to eat; Reward, food as reward; Routine, structured meal routine; Avail., healthy food availability; Distr., distraction−free eating; Resp., responsive feeding; Restr., restriction; Variety, variety encouragement; Reinf., positive reinforcement. *p < .05, **p < .01, ***p < .001
Associations between FPSQ-28 domain scores and child dietary behaviors are presented in Supplementary Table S1. Several statistically significant associations were observed across feeding-practice domains and dietary behaviors; however, because of the large number of comparisons, these findings should be considered exploratory.
Regression analyses
In the multiple linear regression model examining factors associated with parental nutrition knowledge score, child screen time was the only variable independently associated with knowledge after adjustment for sociodemographic and child-related variables. Each higher screen-time category was associated with a lower mean parental knowledge score (B = − 0.116, 95% CI: −0.210 to − 0.023, p = .015). Household income showed a positive but borderline association with knowledge score (B = 0.082, 95% CI: −0.004 to 0.168, p = .062). Other variables, including sex, parent age, education, occupation, number of children, child age, and childcare setting, were not independently associated with knowledge score. The model explained a small proportion of variation in knowledge score (Table 9).
Table 9.
Multiple Linear Regression: Factors Associated With Parental Nutrition Knowledge Score
| Predictor | B [95% CI] | p-value |
|---|---|---|
| Sex, male | −0.036 [− 0.308, 0.236] | 0.795 |
| Age rank | 0.034 [− 0.103, 0.171] | 0.623 |
| Education rank | 0.050 [− 0.063, 0.163] | 0.385 |
| Employee | −0.071 [− 0.308, 0.167] | 0.559 |
| Self-employed | −0.217 [− 0.507, 0.072] | 0.141 |
| Number of children rank | 0.029 [− 0.051, 0.108] | 0.478 |
| Income rank | 0.082 [− 0.004, 0.168] | 0.062 |
| Child age rank | −0.068 [− 0.198, 0.062] | 0.305 |
| Government childcare | 0.067 [− 0.216, 0.350] | 0.642 |
| Private childcare | 0.170 [− 0.062, 0.402] | 0.150 |
| Screen time rank | −0.116 [− 0.210, − 0.023] | 0.015 |
The dependent variable was the total parental nutrition knowledge score, ranging from 0 to 6, with higher scores indicating greater nutrition knowledge. Values are unstandardized regression coefficients (B) with 95% confidence intervals. Positive B values indicate higher mean knowledge scores, whereas negative B values indicate lower mean knowledge scores. Reference categories were female sex, homemaker occupation, and home childcare. Parent age, education, number of children, household income, child age, and screen time were entered as ordinal rank variables. Complete-case analysis was used; N = 336. Model R² = 0.066; adjusted R² = 0.035. VIF values were below 2 for all predictors, indicating no evidence of problematic multicollinearity. Because the knowledge score was bounded and clustered toward higher values, findings should be interpreted cautiously. B, unstandardized coefficient; CI, confidence interval; VIF, variance inflation factor
In a sensitivity analysis excluding child age and including all 350 participants, the findings were materially unchanged. Higher screen time remained associated with a lower parental nutrition knowledge score (B = − 0.119, 95% CI: −0.207 to − 0.031, p = .008), while household income remained non-significant (B = 0.078, 95% CI: −0.006 to 0.161, p = .068) (Supplementary Table S4).
The association with child screen time was also robust to alternative modeling of the bounded knowledge score. In an ordinal logistic regression sensitivity analysis, each higher screen-time category was associated with lower odds of being in a higher knowledge-score category (aOR = 0.721, 95% CI: 0.586–0.887, p = .002). Household income showed a positive association in this sensitivity model (aOR = 1.231, 95% CI: 1.015–1.492, p = .034), although this association did not reach statistical significance in the primary linear regression analyses and was therefore interpreted cautiously.
In separate multiple linear regression models for FPSQ-28 feeding-practice domains, male participants reported higher pressure to eat, food as reward, and structured meal routine scores. Higher parental nutrition knowledge score was associated with lower pressure to eat and with higher scores for healthy food availability, distraction-free eating, responsive feeding, restriction, variety encouragement, and positive reinforcement. Higher screen time was associated with higher pressure to eat and with lower distraction-free eating and restriction scores. The explanatory power of these models was low to modest (Table 10).
Table 10.
Multiple linear regression: predictors of FPSQ-28 feeding-practice domain scores
| Predictor | Press. | Reward | Rout. | Avail. | Distr. | Resp. | Restr. | Var. | Reinf. |
|---|---|---|---|---|---|---|---|---|---|
| Sex (male) | 0.997*** | 0.905*** | 0.443** | 0.089 | 0.265 | 0.263 | 0.189 | -0.150 | -0.242 |
| Age rank | -0.114 | -0.121 | -0.062 | 0.010 | 0.017 | 0.007 | 0.081 | 0.053 | 0.075 |
| Education rank | -0.108 | -0.123 | -0.006 | 0.026 | -0.014 | -0.018 | -0.075 | -0.012 | -0.062 |
| Employee | 0.043 | 0.017 | 0.055 | -0.015 | -0.109 | -0.102 | 0.159 | 0.079 | 0.189 |
| Self-employed | -0.313 | -0.323 | -0.100 | 0.125 | -0.034 | -0.172 | 0.118 | 0.126 | 0.089 |
| Children rank | 0.117* | 0.062 | 0.100* | 0.034 | -0.007 | 0.059 | 0.008 | 0.041 | 0.067 |
| Income rank | -0.005 | -0.036 | 0.086 | 0.079 | 0.020 | 0.041 | 0.093* | 0.097* | 0.089 |
| Child age rank | -0.040 | 0.081 | 0.097 | -0.040 | 0.122 | -0.016 | 0.042 | -0.029 | 0.048 |
| Govt childcare | -0.139 | -0.353* | -0.244 | -0.419** | -0.181 | -0.158 | -0.321* | -0.182 | -0.206 |
| Private childcare | -0.114 | -0.143 | -0.113 | -0.032 | -0.017 | -0.055 | -0.185 | -0.078 | -0.081 |
| Screen time | 0.133* | 0.104 | -0.049 | -0.030 | -0.257*** | -0.020 | -0.136** | -0.024 | -0.001 |
| Knowledge score | -0.194** | -0.010 | 0.104 | 0.187*** | 0.172** | 0.178** | 0.192*** | 0.136* | 0.250*** |
Values are unstandardized B coefficients from separate multiple linear regression models for each FPSQ−28 domain mean score. Positive B values indicate higher domain scores. Reference categories were female sex, homemaker occupation, and home childcare. Parent age, education, number of children, household income, child age, and screen time were entered as ordinal rank variables. VIF values were below 2 for all predictors, indicating no evidence of problematic multicollinearity. Model R² values were: pressure to eat = 0.177, food as reward = 0.167, structured meal routine = 0.081,
healthy food availability = 0.112, distraction−free eating = 0.099, responsive feeding = 0.066, restriction = 0.139, variety encouragement = 0.069, and positive reinforcement = 0.100. FPSQ−28, Feeding Practices and Structure Questionnaire−28; VIF, variance inflation factor. Press., pressure to eat; Reward, food as reward; Rout., structured meal routine; Avail., healthy food availability; Distr., distraction−free eating; Resp., responsive feeding; Restr., restriction; Var., variety encouragement; Reinf., positive reinforcement. *p < .05, **p < .01, ***p < .001. Full models with 95% confidence intervals are provided in Supplementary Table S2.
In the ordinal logistic regression models for child dietary behaviors, child screen time showed the most consistent associations across outcomes. After Benjamini–Hochberg FDR correction across the 55 substantive exposure–outcome tests, higher screen time remained associated with lower odds of more frequent breakfast consumption and fruit and vegetable intake and with higher odds of more frequent unhealthy snack and sweetened drink consumption. Among the feeding-practice domains, higher healthy food availability and positive reinforcement remained associated with higher odds of more frequent fruit and vegetable intake, whereas pressure to eat remained associated with lower odds. Food as reward remained associated with higher odds of more frequent sweetened drink consumption. Restriction was associated with lower odds of more frequent unhealthy snack consumption, while variety encouragement was associated with higher odds of more frequent unhealthy snack consumption and lower odds of reported food refusal. Other associations that met the nominal p < .05 threshold in the primary models did not remain statistically supported after FDR correction and were therefore interpreted cautiously (Table 11; Supplementary Tables S3 and S6). In a sensitivity analysis, child screen time was entered as a categorical rather than ordinal-rank predictor. The direction and overall pattern of associations with breakfast consumption, fruit and vegetable intake, unhealthy snack consumption, and sweetened drink consumption were materially unchanged, supporting the robustness of the primary rank-coded specification. Assessment of the proportional-odds assumption indicated evidence of violation for the breakfast consumption, fruit and vegetable intake, and food-refusal models, whereas no clear evidence of violation was observed for the unhealthy-snack or sweetened-drink models. Multinomial logistic regression sensitivity analyses were therefore performed for the three affected outcomes. The principal findings for screen time, fruit and vegetable availability, and variety encouragement were generally consistent with these sensitivity analyses, although some associations observed in the primary ordinal models were not consistently reproduced across multinomial outcome categories (Supplementary Table S5).
Table 11.
Ordinal logistic regression: predictors of child dietary behaviors
| Predictor | Breakfast | Fruits/Veg | Snacks | Drinks | Refuses |
|---|---|---|---|---|---|
| Sex (male) | 0.670 | 0.284 | 1.584 | 1.292 | 1.181 |
| Age rank | 0.801 | 0.887 | 0.868 | 1.025 | 0.815 |
| Education rank | 0.771 | 0.962 | 1.101 | 1.074 | 1.001 |
| Income rank | 0.885 | 1.213 | 0.973 | 1.104 | 0.997 |
| Govt childcare | 0.389 | 0.810 | 1.490 | 1.329 | 0.893 |
| Private childcare | 0.319 | 1.068 | 1.106 | 0.912 | 1.391 |
| Screen time | 0.654 | 0.643 | 1.967 | 2.206 | 0.842 |
| Knowledge score | 1.306 | 1.183 | 0.790 | 0.765 | 1.028 |
| Pressure to eat | 0.935 | 0.616 | 1.064 | 1.126 | 1.193 |
| Food as reward | 0.742 | 1.091 | 1.328 | 1.522 | 0.925 |
| Meal routine | 1.298 | 1.232 | 0.732 | 0.934 | 0.924 |
| Food availability | 1.209 | 2.492 | 1.012 | 0.921 | 0.989 |
| Distraction-free | 0.878 | 0.683 | 0.776 | 1.014 | 0.905 |
| Responsive feeding | 0.934 | 0.817 | 1.541 | 1.120 | 1.090 |
| Restriction | 0.773 | 0.909 | 0.561 | 0.695 | 0.853 |
| Variety | 1.521 | 0.724 | 1.773 | 1.073 | 0.521 |
| Reinforcement | 0.950 | 1.496 | 1.059 | 1.119 | 1.346 |
Values are adjusted odds ratios from separate ordinal logistic regression models. For breakfast consumption and fruit/vegetable intake, adjusted odds ratios greater than 1 indicate higher odds of being in a more frequent favorable-behavior category. For unhealthy snack consumption, sweetened drink consumption, and food refusal, adjusted odds ratios greater than 1 indicate higher odds of being in a more frequent or stronger unfavorable-behavior category. Full models with 95% confidence intervals and nominal p-values are provided in Supplementary Table S3. Benjamini–Hochberg false discovery rate–adjusted q-values for the 55 substantive exposure–outcome tests are provided in Supplementary Table S6. Sociodemographic and childcare covariates were included primarily for adjustment and were not included in the FDR family. McFadden pseudo-R² values were: breakfast = 0.099, fruit/vegetable intake = 0.152, unhealthy snacks = 0.120, sweetened drinks = 0.121, and food refusal = 0.029
The food-refusal model explained relatively little variation (McFadden pseudo-R² = 0.029). The inverse association between variety encouragement and food refusal remained supported after FDR correction and was also consistent with the multinomial logistic regression sensitivity analysis. The nominal association between positive reinforcement and greater food refusal did not remain supported after FDR correction and was not consistently reproduced in the multinomial sensitivity analysis (Supplementary Table S5).
Discussion
This study examined parental nutrition knowledge, multidimensional feeding practices, and selected dietary behaviors among preschool children aged 2–5 years in the West Bank, Palestine. Overall, parental nutrition knowledge was high, whereas reported feeding practices and child dietary behaviors varied substantially. Across the principal multivariable dietary-behavior models, child screen time showed the most consistent associations, while selected FPSQ-28 feeding-practice domains were associated with specific dietary behaviors. These findings support the view that preschool children’s dietary patterns may be related not only to what caregivers know about nutrition but also to the organization of meals, the foods made available at home, screen-related routines, and the feeding strategies used in response to children’s eating behaviors [1, 3].
Parental nutrition knowledge showed only limited evidence of an independent relationship with the dietary behaviors assessed. In the primary ordinal logistic regression models, higher knowledge showed nominal associations with more frequent breakfast consumption and less frequent sweetened drink consumption; however, neither association remained statistically supported after Benjamini–Hochberg FDR correction. Accordingly, these findings should be interpreted cautiously rather than as robust independent associations. At the same time, higher child screen time remained associated with lower parental nutrition knowledge in the primary linear regression model, in the complete-sample sensitivity analysis excluding child age, and when the bounded knowledge score was modeled as an ordinal outcome. This consistency strengthens the observed screen time–knowledge relationship, although the cross-sectional design does not establish its direction. More broadly, the limited associations between knowledge and dietary behaviors are compatible with evidence suggesting that nutrition knowledge alone may be insufficient to translate into healthier child eating patterns when food availability, household routines, child preferences, and caregiver feeding responses also shape behavior [1, 3].
Several feeding-practice domains showed associations with specific dietary behaviors after adjustment for multiplicity. Greater healthy food availability and positive reinforcement remained associated with more frequent fruit and vegetable intake, whereas greater pressure to eat remained associated with less frequent fruit and vegetable intake. Variety encouragement remained associated with lower reported food refusal, although it was also associated with more frequent unhealthy snack consumption, illustrating why the individual FPSQ-28 domains should not be classified uniformly as beneficial or harmful. Restriction was associated with less frequent unhealthy snack consumption, while food as reward remained associated with more frequent sweetened drink consumption. These findings reinforce the importance of interpreting feeding practices in relation to the specific behavior being examined rather than assigning a single favorable or unfavorable meaning to an entire domain [3, 10–12].
The findings involving food as reward and pressure to eat warrant particular attention. After FDR correction, food as reward remained associated with more frequent sweetened drink consumption, while its nominal associations with breakfast and unhealthy snack consumption were no longer statistically supported. Pressure to eat remained associated with less frequent fruit and vegetable intake. These patterns are compatible with literature suggesting that coercive or reward-based feeding strategies may be related to children’s food preferences and eating responses; however, the direction of these relationships cannot be determined in a cross-sectional study. Parents may adopt pressure or reward strategies because a child already demonstrates selective eating, limited intake of particular foods, or strong preferences for less nutritious foods. Bidirectional parent–child influences should therefore remain an important consideration when interpreting these associations [3, 13, 14].
Child screen time was the most consistent correlate of dietary behavior in the principal adjusted models. Higher screen time remained associated after FDR correction with lower odds of more frequent breakfast consumption and fruit and vegetable intake and with higher odds of more frequent unhealthy snack and sweetened drink consumption. Importantly, the direction and overall pattern of these associations were materially unchanged when screen time was entered categorically rather than as an ordinal-rank predictor, reducing concern that the findings were driven by the assumption of a linear trend across screen-time categories. The screen-time findings were also generally supported by the multinomial sensitivity analyses for outcomes in which the proportional-odds assumption was violated. Screen exposure may coincide with distracted eating, greater opportunities for snacking, and disruption of structured meals, although these mechanisms were not directly measured in the present study [15]. Given the cross-sectional design, it is also possible that less structured family routines contribute simultaneously to greater screen exposure and less favorable dietary behaviors. Nevertheless, the consistency of the observed associations across several dietary outcomes suggests that screen-related routines are an important area for future research and may reasonably be considered in family-based nutrition guidance, alongside existing recommendations to limit sedentary screen time in young children [16].
The sensitivity analyses also provide important context for interpreting the ordinal dietary-behavior models. Evidence of violation of the proportional-odds assumption was identified for breakfast consumption, fruit and vegetable intake, and food refusal, and multinomial logistic regression models were therefore fitted for these outcomes. Several of the main substantive patterns—notably those involving screen time, healthy food availability, and variety encouragement—were generally consistent across model specifications, whereas some nominal associations from the primary ordinal models were not reproduced consistently across multinomial outcome categories. These differences support a cautious interpretation of isolated nominal findings and reinforce the emphasis on associations that were supported by FDR correction and showed consistency across sensitivity analyses.
Caregiver and childcare characteristics may also contribute to the broader feeding environment. Childcare setting was associated with breakfast consumption in the adjusted models, with children attending government or private childcare showing lower odds of more frequent breakfast consumption than children cared for at home. This finding should be interpreted cautiously because the study did not assess whether breakfast was eaten before attendance, provided within childcare, or affected by attendance schedules or institutional food policies. In the secondary FPSQ-28 regression models, caregiver sex was associated with several feeding-practice domains, with male respondents reporting higher pressure to eat, food as reward, and structured meal routine scores. Because these FPSQ-domain models were secondary analyses and were not subjected to formal multiplicity adjustment, individual nominal associations should not be overinterpreted. Nevertheless, they are compatible with literature emphasizing that mothers, fathers, grandparents, and other caregivers may all contribute to feeding routines and food provision during early childhood [3, 12].
The potential contribution of fathers and other caregivers is particularly relevant in family-centered nutrition interventions. Fathers’ nutrition knowledge and feeding involvement have increasingly been recognized as components of children’s dietary environments and obesity-prevention efforts [17, 18]. Grandparents may also influence food availability, meal routines, and feeding decisions [19]. Future family-based nutrition programs in this setting may therefore benefit from engaging the range of caregivers involved in food purchasing, preparation, and feeding rather than focusing exclusively on mothers.
Socioeconomic findings were less uniform. Household income was positively associated with parental knowledge in bivariate analysis and showed a borderline association in the primary adjusted linear regression model. When the bounded knowledge score was modeled ordinally in sensitivity analysis, income showed a positive association, indicating some sensitivity of this finding to model specification. Income was also associated with selected FPSQ-28 domain scores in the secondary regression analyses. These associations should therefore be interpreted cautiously rather than as definitive evidence of a socioeconomic gradient. In the Palestinian context, however, household resources may plausibly influence food access, dietary variety, childcare arrangements, and families’ ability to implement nutrition recommendations. Previous Palestinian evidence has linked nutritional outcomes with socioeconomic circumstances, food availability, and dietary intake [6, 7]. Nutrition-promotion strategies should therefore consider the practical circumstances in which families make food-related decisions rather than assuming that provision of information alone is sufficient.
A broader implication of the present findings is the apparent gap between generally high nutrition knowledge and variability in reported child dietary behaviors. Nearly all respondents correctly recognized the importance of breakfast and daily fruit and vegetable intake, yet unhealthy snack and sweetened drink consumption remained common in a proportion of children, and only about half of respondents correctly identified that natural juice is not a healthy substitute for whole fruit. This pattern suggests that nutrition education may be more useful when it combines factual information with practical strategies addressing common misconceptions, food availability, structured meals, management of food refusal, feeding interactions, and screen-related eating routines. Interventions that incorporate behavioral and environmental components in addition to nutrition information may be particularly relevant for preschool-aged children [1, 3, 20].
This study has several strengths. It provides evidence from the West Bank, where research integrating parental nutrition knowledge, multidimensional feeding practices, and preschool dietary behaviors remains limited [5]. The study assessed nine distinct FPSQ-28 feeding-practice domains rather than reducing feeding practices to a single overall score, examined five child dietary behaviors separately, and used multivariable models to account for relevant caregiver- and child-related characteristics. The analysis also incorporated several robustness procedures, including FDR correction for the principal substantive exposure–outcome comparisons, sensitivity analysis of screen time as a categorical predictor, multinomial sensitivity models when the proportional-odds assumption was violated, and alternative modeling of the bounded parental knowledge score. Use of both online and paper-based recruitment may additionally have improved access to participants from different community settings.
Several limitations should also be considered. First, the cross-sectional design precludes conclusions regarding causality, temporality, or the direction of associations, particularly because parent feeding practices and child dietary behaviors may influence one another. Second, convenience sampling limits generalizability, and the predominance of female respondents may restrict inference regarding male caregivers. The sample was drawn from selected northern West Bank governorates and was not intended to be geographically representative of all Palestinian families. Urban, rural, or refugee-camp residence was not collected as an individual-level variable, so differences according to residential setting could not be examined. Third, all measures were based on caregiver self-report and may therefore be affected by recall and social-desirability bias. Fourth, although the online and paper questionnaires were standardized in content, wording, order, response options, eligibility criteria, consent information, and instructions, individual response mode was not retained in the analytical dataset. Consequently, potential differences in participant characteristics or responses between administration modes could not be evaluated. Fifth, participation by more than one caregiver from the same household was possible because no household identifier was collected, and potential clustering of observations within households could not be assessed.
Additional measurement and analytical limitations should also be acknowledged. The FPSQ-28 was translated and culturally adapted for this study, but a published Arabic validation was not available to our knowledge; further psychometric evaluation in Arabic-speaking populations is warranted. The six-item parental nutrition knowledge measure was a brief factual index rather than a validated unidimensional psychometric scale and showed evidence of a ceiling effect. Although the principal screen-time association with knowledge remained consistent when the score was modeled ordinally, results involving the knowledge score should still be interpreted with this limitation in mind. Complete-case analysis resulted in a smaller sample for the primary knowledge regression because child age could not be included for all participants; however, omitting child age and including all 350 participants yielded materially similar screen-time findings. Ordered predictors were primarily entered as rank variables, which assumes a common trend across ordered categories; a categorical sensitivity analysis for screen time, the principal ordered substantive predictor in the dietary models, produced a similar overall pattern. Finally, although multiplicity was addressed using FDR correction for the 45 exploratory bivariate FPSQ–dietary-behavior comparisons and the 55 substantive exposure–outcome tests in the principal dietary models, the nine FPSQ-domain regression models were secondary analyses without formal multiplicity adjustment. Findings from those models should therefore be interpreted in terms of effect size, precision, consistency, and biological or behavioral plausibility rather than isolated nominal p-values.
In summary, parents and caregivers of preschool children in the West Bank demonstrated generally high nutrition knowledge, but feeding practices and reported child dietary behaviors varied. Child screen time showed the most consistent relationships with dietary behaviors, with higher screen exposure associated with less frequent breakfast and fruit and vegetable intake and more frequent unhealthy snack and sweetened drink consumption. Selected feeding-practice domains were also associated with specific dietary behaviors, while the independent relationships between parental nutrition knowledge and the dietary outcomes were less robust after correction for multiple testing. Taken together, the findings support consideration of family-centered nutrition approaches that combine accurate nutrition information with practical feeding guidance, supportive home food environments, structured eating routines, attention to screen-related behaviors, and involvement of the caregivers who participate in everyday child feeding.
Conclusion
Among parents and caregivers of preschool children in the West Bank, Palestine, parental nutrition knowledge was generally high, but reported feeding practices and child dietary behaviors varied. Higher child screen time was associated with lower parental nutrition knowledge and was the most consistent correlate of child dietary behaviors, showing associations with less frequent breakfast and fruit and vegetable intake and more frequent unhealthy snack and sweetened drink consumption. Selected feeding-practice domains were also associated with specific dietary behaviors, whereas associations between parental nutrition knowledge and the dietary outcomes were less robust after correction for multiple testing. Because of the cross-sectional design, causal or directional relationships cannot be inferred. These findings support consideration of family-centered nutrition approaches that combine practical feeding guidance, supportive home food environments, attention to screen-related eating behaviors, and involvement of caregivers in early childhood nutrition.
Supplementary Information
Supplementary Material 1: Supplementary File 1: STROBE checklist for cross-sectional studies.
Supplementary Material 2: Supplementary File 2: Supplementary Tables S1–S6.
Acknowledgements
Not applicable.
Authors’ contributions
LJ contributed to data analysis and led the design and writing of the manuscript. LM contributed to research conceptualization, supervision, and critical revision of the manuscript. SA, NA, LS, and NAH contributed to data collection and drafting of the manuscript. MZ contributed to data collection. All authors critically revised the manuscript, approved the final version, and agreed to be accountable for the work.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
Ethical approval was obtained from the Institutional Review Board of An-Najah National University (Approval No. Med.Nov.2025/174). Electronic or written informed consent was obtained from all participants before participation. The study was conducted in accordance with the Declaration of Helsinki.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Shahenaz Abualrob, Natal Abbasi and Lina Sawalha contributed equally to this work.
References
- 1.Jarman M, Edwards K, Blissett J. Influences on the dietary intakes of preschool children: a systematic scoping review. Int J Behav Nutr Phys Activity. 2022;19:20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Unicef. The State of the World’s Children 2019: Children, Food and Nutrition: Growing Well in a Changing World. New York: UNICEF; 2019. [Google Scholar]
- 3.Costa A, Oliveira A. Parental feeding practices and children’s eating behaviours: an overview of their complex relationship. Healthcare. 2023;11(3):400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Jansen E, Williams KE, Mallan KM, Nicholson JM, Daniels LA. The Feeding Practices and Structure Questionnaire (FPSQ-28): a parsimonious version validated for longitudinal use from 2 to 5 years. Appetite. 2016;100:172–80. [DOI] [PubMed] [Google Scholar]
- 5.Assaf E, Al Sabbah H, Al-Jawadleh A. Analysis of the nutritional status in the Palestinian territory: a review study. Front Nutr. 2023;10:1206090. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Palestinian Central Bureau of S. Palestinian Multiple Indicator Cluster Survey 2019–2020, Survey Findings Report. Ramallah, Palestine: PCBS; 2021. [Google Scholar]
- 7.Issa G, Qasrawi R, Thwib S, Amro M, Abu Ghoush R, Saghir S. A structural equation modeling approach to examine determinants of nutritional status in Palestinian children 6–59 months. PLoS ONE. 2025;20(9):e0331579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Vandenbroucke JP, von Elm E, Altman DG, Gøtzsche PC, Mulrow CD, Pocock SJ, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. PLoS Med. 2007;4(10):e297. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Jansen E, Mallan KM, Nicholson JM, Daniels LA. The feeding practices and structure questionnaire (FPSQ): construction and initial validation in a sample of Australian first-time mothers and their 2-year-olds. Int J Behav Nutr Phys Activity. 2014;11:72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Al-buobayd AA, Al-Otaibi HH, Farag HAS. Association of maternal feeding style with fruit and vegetable consumption in Saudi preschoolers: a nationwide cross-sectional study. Nutrients. 2023;15(22):4735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Channell Doig A, Lipsky LM, Shearrer G, Nansel TR. Profiles of maternal feeding practices and their associations with parent feeding styles, child diet quality, and the home food environment. Appetite. 2025;213:108038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Marr C, Breeze P, Caton SJ. A comparison between parent and grandparent dietary provision, feeding styles and feeding practices when caring for preschool-aged children. Appetite. 2022;168:105777. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Baker LN, Fuglestad AJ. Parent’s use of coercive control practices with food is associated with poorer emotion regulation and increased emotional overeating in preschoolers. Appetite. 2024;201:107608. [DOI] [PubMed] [Google Scholar]
- 14.Gubbels JS, Kremers SP, Stafleu A, Dagnelie PC, Goldbohm RA, de Vries NK, et al. Diet-related restrictive parenting practices: impact on dietary intake of 2-year-old children and interactions with child characteristics. Appetite. 2009;52(2):423–9. [DOI] [PubMed] [Google Scholar]
- 15.Han YH, Shin S, Woo EY, Park HK, Hyun T. Screen time, mealtime media use, and dietary behaviors in Korean preschoolers: a cross-sectional study. Korean J Community Nutr. 2023;28(3):206–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.World Health O. Guidelines on Physical Activity, Sedentary Behaviour and Sleep for Children Under 5 Years of Age. Geneva: World Health Organization; 2019. [PubMed] [Google Scholar]
- 17.Bawa SG, Haldeman L. Fathers nutrition knowledge and child feeding practices associated with childhood overweight and obesity: a scoping review of literature from 2000 to 2023. Global Pediatr Health. 2024;11:2333794X241263199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Litchford A, Savoie Roskos MR, Wengreen H. Influence of fathers on the feeding practices and behaviors of children: a systematic review. Appetite. 2020;147:104558. [DOI] [PubMed] [Google Scholar]
- 19.Jongenelis MI, Budden T. The influence of grandparents on children’s dietary health: a narrative review. Curr Nutr Rep. 2023;12:395–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hodder RK, O’Brien KM, Tzelepis F, Wyse RJ, Yoong SL, Stacey FG, et al. Interventions for increasing fruit and vegetable consumption in children aged five years and under. Cochrane Database Syst Rev. 2024;9(9):CD008552. 10.1002/14651858.CD008552.pub8. [DOI] [PMC free article] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material 1: Supplementary File 1: STROBE checklist for cross-sectional studies.
Supplementary Material 2: Supplementary File 2: Supplementary Tables S1–S6.
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.
