Nutrition knowledge and health belief model-based dietary perceptions among in-school female adolescents in Ogun State, Nigeria

Abstract

Background

Adolescence is a critical period for establishing dietary behaviours that influence immediate health, reproductive outcomes, and long-term risk of non-communicable diseases. Female adolescents are particularly vulnerable to poor nutrition due to increased biological demands and sociocultural constraints. Examining the relationship between nutrition knowledge and behavioural perceptions within a theoretical framework is essential for developing effective interventions.

Methods

A descriptive cross-sectional study was conducted among 290 in-school female adolescents aged 13–17 years in public and private secondary schools in Odeda Local Government Area, Ogun State, Nigeria. A pretested, interviewer-administered questionnaire assessed sociodemographic characteristics, nutrition knowledge, and perceived dietary behaviour using the Health Belief Model (HBM) framework. Independent samples t-tests, Pearson’s correlation, chi-squares test, and six parallel linear regression models were fitted one per HBM construct as the dependent variable with nutrition knowledge as the primary predictor and age, school setting, household income, and household size as covariates. Statistical significance was set at p < 0.05.

Results

The mean age of participants was 15.19 ± 1.05 years. Overall nutrition knowledge was moderate, with a mean score of 56.52 ± 13.26. Nutrition knowledge was positively associated with perceived severity, perceived benefits, cues to action, and self-efficacy, and negatively associated with perceived barriers. Nutrition knowledge was not significantly associated with perceived susceptibility. In multivariate analysis, nutrition knowledge remained significantly associated with perceived severity (β = 0.196, η² = 0.041), perceived benefits (β = 0.179, η² = 0.033), perceived barriers (β = −0.344, η² = 0.118), cues to action (β = 0.286, η² = 0.083), and self-efficacy (β = 0.225, η² = 0.052), but not perceived susceptibility (p = 0.124).

Conclusions

Nutrition knowledge is significantly associated with HBM-based dietary perceptions among in-school female adolescents. Strengthened, theory-driven nutrition education, particularly in public schools, may reinforce favourable dietary cognitions theoretically linked to healthier eating practices.

Introduction

Adolescents aged 10–19 years constitute approximately 16% of the global population, representing an estimated 1.3 billion individuals [1]. This developmental period, second only to infancy in the pace of physical and cognitive growth, encompasses profound biological, psychological, and social transformation, beginning with puberty and culminating in legal and social independence [2–4].

Globally, nearly one in three individuals experiences some form of malnutrition, and adolescents in low- and middle-income countries bear a disproportionate share of this burden [5]. The double burden of malnutrition, which is the coexistence of undernutrition and overnutrition within the same population, is of particular concern among female adolescents in developing countries [5, 7]. Estimates indicate that malnutrition affects approximately 49 million adolescent girls (8%) and 154 million women (10%) worldwide [6, 8]. Cross-sectional evidence further documents prevalence rates of 13.4% for underweight, 16.9% for overweight, and 7.5% for obesity among adolescents, with abdominal obesity significantly more prevalent among females (16.9%) than males (2.5%) [9, 10]. These nutritional vulnerabilities are compounded by suboptimal dietary diversity, excessive carbohydrate consumption, poor nutrition knowledge, and unhealthy perceived dietary behaviour [10, 11].

The health consequences are substantial and extend beyond adolescence. Thinness is associated with delayed maturation, reduced muscle strength, diminished work capacity, and lower bone density [8], while being overweight elevates obstetric risk in females. Adequate nutrition during adolescence is essential to support muscle and fat accretion, brain development, linear growth, and reproductive maturation [1, 8, 9]. In Nigeria, rapid nutrition transition, urbanisation, and the proliferation of ultra-processed foods have intensified these dietary risks among young females [4, 6].

Adolescents from higher-income households benefit disproportionately from nutrition education through school curricula and parental influence, further widening nutritional disparities [12]. Malnutrition among female adolescents is simultaneously a cause and consequence of gender inequality [13]. Discriminatory norms, unequal access to food and nutrition services, limited decision-making power, and the heightened nutritional demands of growth and menarche collectively amplify vulnerability in this group, with downstream consequences including anaemia, poor reproductive outcomes, and intergenerational transmission of malnutrition [13–15].

Although school-based nutrition interventions are widely implemented, evidence consistently demonstrates that nutrition knowledge alone does not reliably translate into healthier dietary behaviour [16, 17]. Behavioural change requires engagement with the perceptions, beliefs, and perceived control that underpin health-related decisions [18]. The Health Belief Model (HBM) provides a useful framework for this purpose, encompassing perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action, and self-efficacy [18]. Within this framework, nutrition knowledge functions as a distal cognitive enabler: it is theorised to heighten perceived susceptibility and severity by grounding risk appraisal in factual understanding, reinforce perceived benefits, reduce perceived barriers by building practical competence, and strengthen self-efficacy. As proximal determinants of behavioural intent, self-efficacy and perceived benefits are the constructs through which knowledge is most likely to exert its influence on dietary behaviour [18]. Elucidating these pathways, however, requires longitudinal or experimental designs.

Despite growing research interest in adolescent nutrition in Nigeria, few studies have simultaneously examined the relationship between nutrition knowledge and perceived dietary behaviour using a multidimensional theoretical framework, particularly among female adolescents in semi-urban settings such as Odeda Local Government Area, Ogun State. This study, therefore, aimed to examine that relationship among in-school female adolescents, specifically assessing [1] differences in nutrition knowledge and HBM constructs by school setting, and [2] the independent association between nutrition knowledge and each HBM construct after adjusting for sociodemographic and economic factors.

Methods

Study design and setting

This study employed a descriptive cross-sectional design in accordance with STROBE recommendations for observational studies. The research was conducted between August and November 2024 in Odeda Local Government Area (LGA), Ogun State, Nigeria. Odeda LGA is located along the Abeokuta–Ibadan Road, approximately 10 km from Abeokuta, the state capital. The area comprises several towns and villages with predominantly agricultural livelihoods and includes both public and private secondary schools, providing a heterogeneous educational and socioeconomic context suitable for adolescent nutrition research.

Study population and eligibility criteria

The study population comprised in-school female adolescents aged 13–17 years enrolled in public or private secondary schools in Odeda LGA at the time of data collection. Adolescents were eligible if they were present during the study period, provided written assent, and had written informed consent from a parent or legal guardian. Adolescents with chronic illnesses, diagnosed eating disorders, diabetes, or other medical conditions likely to influence dietary behaviour or anthropometric outcomes were excluded. Adolescents unwilling to participate, those without parental consent, or those enrolled in other nutrition-related interventions were also excluded.

Sampling procedure

Odeda LGA is administratively divided into three zones: Odeda, Ilugun, and Opeji. The Opeji zone was selected using simple random sampling. Within this zone, five communities exist: Obantoko, Adao, Alabata, Opeji, and Obete. The Obantoko community was purposively selected due to its larger population size and diverse sociodemographic composition, encompassing both urban and peri-urban settings and hosting a mix of public and private secondary schools.

From the list of registered secondary schools in Obantoko, four schools (two public and two private) were purposively selected to ensure representation across school types. Within each selected school, students were stratified by class level (Senior Secondary School 1–3). Proportional allocation was applied to determine the number of participants per class, and simple random sampling was used to select eligible students. This process yielded a final sample of 290 female adolescents aged 13–17 years.

Sample size determination

The minimum sample size was calculated using Cochran’s formula for cross-sectional studies. The expected prevalence of adequate dietary diversity (56.5%) reported by Ademiluyi & Uthman-Akinhanmi [12] was used. A 95% confidence level (Z = 1.96) and a margin of error of 5% (d = 0.05) were applied, producing an initial sample size of 379 participants. Given the finite population of eligible female adolescents in the selected schools, finite population correction was applied, resulting in an adjusted sample size of approximately 264. To account for potential non-response, a 10% allowance was added, yielding a final minimum sample size of 290 participants.

The finite population of eligible female adolescents across the four selected schools was N = 350 (based on class registration lists at the time of data collection). Applying the Cochran formula (p = 0.565, q = 0.435, Z = 1.96, d = 0.05) yielded an initial sample size estimate of 379. Finite population correction reduced this to n ≈ 264; with a 10% non-response allowance, the final minimum sample size was 290. Of 298 questionnaires administered, 8 were excluded due to incomplete responses (> 20% missing items), yielding a final analytical sample of 290 participants (response rate: 97.3%).

Data collection procedures

Data were collected using a pretested, interviewer-administered, semi-structured questionnaire. The questionnaire obtained information on sociodemographic and economic characteristics, nutrition knowledge, and perceived dietary behaviour toward healthy eating based on the Health Belief Model (HBM). Data collection took place during regular academic sessions and outside examination periods to minimise disruption and participant fatigue. Trained research assistants administered the questionnaires within the school environment. Class teachers assisted only with logistical coordination and were not present during data collection to reduce response bias.

Measurement of variables

Sociodemographic and economic characteristics

Information collected included age, school type, parental education, parental occupation, household income, household size, and number of siblings.

Nutrition knowledge

Nutrition knowledge was assessed using a validated instrument originally developed by Turconi et al. [19] and adapted by Ademiluyi et al. [4]. The tool assessed knowledge of healthy diet composition, consequences of unhealthy eating, and basic nutrition principles. Multiple-choice responses were scored, with correct answers summed to generate a total score. Scores were converted to percentages and categorised as poor (0–40%), moderate (50–70%), or good (> 70%) in line with Ademiluyi et al. (2025).

Perceived dietary behaviour (Health Belief Model constructs)

Perceived dietary behaviour was assessed using an HBM-based questionnaire adapted from Keshani et al. [20] and Saghafi-Asl et al. [21]. Six constructs were measured: perceived severity, perceived susceptibility, perceived benefits, perceived barriers, cues to action, and self-efficacy.

Perceived severity assessed the seriousness of unhealthy eating across emotional and mental health, physical fitness, short- and long-term health outcomes, and social well-being. Perceived susceptibility measured the perceived likelihood of experiencing nutrition-related health problems. Perceived benefits captured beliefs about the advantages of healthy eating, while perceived barriers assessed practical and personal challenges such as cost, availability, preferences, and motivation. Cues to action evaluated internal and external triggers, including family support, peer influence, and exposure to nutrition information. Self-efficacy assessed confidence in maintaining healthy eating across various contexts, including at home, with peers, and in public settings.

Consistent with the original HBM formulation, perceived barriers are conceptualised as obstacles that inhibit health-promoting dietary behaviours; accordingly, lower perceived barriers are indicative of a more favourable behavioural disposition. To maintain directional consistency with other HBM constructs (where higher scores are favourable), perceived barriers items were reverse-scored before summation, such that a higher composite score on the perceived barriers subscale reflects fewer or less severe perceived obstacles, that is, a more favourable barrier profile. This approach has been applied in prior HBM dietary behaviour research [20, 21] and ensures that the classification categories (unfavourable, moderate, favourable) remain comparably interpretable across all six constructs.

All items were rated on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Construct scores were summed, converted to percentage scores, and categorised as unfavourable (0–39%), moderate (40–69%), or favourable (70–100%) following Heydari et al. [22] and Ademiluyi et al. [4].

Instrument validity and reliability

The questionnaire was pretested among 30 female adolescents from a school in a neighbouring community with similar sociodemographic characteristics; participants from this pilot were excluded from the main study. Content validity was assessed using Lawshe’s [23] method as updated by Sürücü and Maslakçı [24]. Ten experts in public health nutrition reviewed the instrument for relevance, clarity, and cultural appropriateness. Of the initial 77 items, 31 were retained in their original form, 25 were revised, and 21 were removed, resulting in a final 56-item questionnaire. Item revisions were concentrated in the perceived barriers subscale (where seven items referencing food costs and family food norms required rewording to reflect the Nigerian adolescent context), the perceived severity subscale (where six items were simplified for age-appropriateness and comprehension), and the cues to action subscale (where four items referencing media and healthcare sources were adapted for local media exposure patterns). Removed items failed to meet the minimum Content Validity Ratio threshold of 0.62 as recommended by Lawshe [23] and were predominantly items from the original perceived susceptibility domain that were deemed culturally incongruent or redundant after expert review. The adaptations are acknowledged to introduce the possibility of reduced construct equivalence relative to the original instruments; however, the retained CVI values (0.74–1.00) and Cronbach’s alpha coefficients across all subscales (ranging from 0.73 to 0.91) support the internal consistency and content representativeness of the adapted tool within this study population. Categorisation cut-offs for nutrition knowledge (poor: 0–40%; moderate: 41–69%; good: ≥70%) and HBM construct scores (unfavourable: 0–39%; moderate: 40–69%; favourable: ≥70%) were adopted from Ademiluyi et al. [4] and Heydari et al. [22], which applied these thresholds in populations of comparable age, educational level, and cultural context. These cut-offs are consistent with the tripartite classification approach widely used in HBM-based dietary behaviour research [20, 21]. Although empirical derivation of population-specific cut-offs through receiver operating characteristic (ROC) analysis would be methodologically optimal, this was beyond the scope of the present study and is recommended for future instrument development efforts in this population.

Statistical analysis

Data were analysed using appropriate statistical software. Descriptive statistics were used to summarise participants’ sociodemographic and economic characteristics. Continuous variables were reported as means and standard deviations, while categorical variables were presented as frequencies and percentages. Figures were used to illustrate the distribution of nutrition knowledge categories, differences in HBM constructs by school setting, and the magnitude of associations between predictors and HBM constructs based on effect sizes. Independent samples t-tests were used to compare mean nutrition knowledge scores and HBM construct scores between public and private school adolescents, as comparisons involved continuous outcomes across two independent groups. The Pearson chi-square test was used to compare the nutrition knowledge between public and private school adolescents. Statistical significance was set at p < 0.05.

Pearson’s correlation analysis was conducted to assess linear relationships between nutrition knowledge and perceived dietary behaviour constructs, as well as interrelationships among the HBM constructs. Correlation coefficients (r) were interpreted based on strength and direction, with significance determined at p < 0.05. Six separate linear regression models were fitted, one per HBM construct as the dependent variable, with nutrition knowledge as the primary predictor and school setting, age, household income, and household size as covariates. Nutrition knowledge was included as the primary predictor, while school setting, age, household income, and household size were entered as covariates. Model significance was evaluated using F statistics, and effect sizes were reported using partial eta squared (η²) to quantify the magnitude of associations. Statistical significance was defined as p < 0.05.

Before multivariate multiple linear regression analysis, key statistical assumptions were verified. Normality of residuals was assessed using the Shapiro-Wilk test and visual inspection of Q-Q plots; residuals for all six HBM outcome variables did not significantly deviate from normality. Homoscedasticity was evaluated using Levene’s test of equality of error variances, and no evidence of significant heteroscedasticity was observed. Independence of errors was confirmed by the study design, as data were collected from distinct individuals with no nested or repeated observations.

Multicollinearity among predictors was assessed using the Variance Inflation Factor (VIF); all VIF values ranged from 1.02 to 2.46, confirming no problematic collinearity. Outliers and influential observations were assessed using Cook’s distance; no observation exceeded the conventional threshold of 4/n = 0.014. Normality of residuals was confirmed via Shapiro-Wilk tests (all p > 0.05) and visual inspection of Q-Q plots.

Results

Sociodemographic and economic characteristics of the female adolescents

Table 1 shows the result of the sociodemographic and economic characteristics of the participants. The study involved 290 female adolescents who were predominantly mid-adolescents, with most aged 15 years and an overall mean age of 15.19 ± 1.05 years. Slightly more participants attended public schools (55.2%) than private schools (44.8%). Parents generally had good educational backgrounds, as most fathers had secondary (46.2%) or tertiary education (47.6%), and most mothers also had secondary (40%) or tertiary education (46.9%). Fathers were mainly civil servants (35.2%) or engaged in trades, crafts, and farming (33.8%), while mothers were largely involved in trades, crafts, and farming (57.9%), with about one-quarter working as civil servants. Over half of the households (53.1%) earned more than ₦100,000 per month, indicating a relatively moderate to high income level among many families. Household sizes were commonly between 5 and 8 members (65.5%), with a mean household size of 5.38 ± 1.84 persons. However, most adolescents had between 0 and 4 siblings (90.3%), with an average of 2.86 ± 1.28 siblings.

Table 1.

Sociodemographic and economic characteristics of the female adolescents

Variable	Frequency	Percent
Age (years)
13	14	4.8
14	51	17.6
15	137	47.2
16	43	14.8
17	45	15.5
Total	290	100
Mean ± SD: 15.19 ± 1.05
School Settings
Public School	160	55.2
Private School	130	44.8
Total	290	100
Father’s highest level of education
No formal education	6	2.1
Primary education	12	4.1
Secondary education	134	46.2
Tertiary education	138	47.6
Total	290	100
Mother’s highest level of education
No formal education	10	3.4
Primary education	28	9.7
Secondary education	116	40
Tertiary education	136	46.9
Total	290	100
Father’s Occupation
Civil Servant	102	35.2
Professionals	55	19
Technicians	31	10.7
Trades, Crafts and Farming	98	33.8
Unemployed	4	1.4
Total	290	100
Mother’s Occupation
Civil Servant	78	26.9
Professionals	40	13.8
Technicians	2	0.7
Trades, Crafts and Farming	168	57.9
Unemployed	2	0.7
Total	290	100
Household’s Monthly Income†
Less than 20,000	40	13.8
20,000–50,000	52	17.9
50,001–100,000	44	15.2
More than 100,000	154	53.1
Total	290	100
Number of Households
0–4	86	29.7
5–8	190	65.5
9–12	14	4.8
Total	290	100
Mean ± SD: 5.38 ± 1.84
Number of Siblings
0–4	262	90.3
5–8	28	9.7
Total	290	100

SD Standard Deviation

Mean ± SD: 2.86 ± 1.28

†Household monthly income was reported in Nigerian Naira (₦)

Nutrition knowledge classification by school setting

Table 2 shows differences in nutrition knowledge (NK) between school settings. Poor nutrition knowledge was reported by 11.0% of public school and 9.0% of private school adolescents. A higher proportion of public school students had moderate nutrition knowledge (81.0%), whereas good nutrition knowledge was more prevalent among private school students (28.0%) compared with public school students (8.0%). Overall differences in the distribution of nutrition knowledge classifications between school settings were statistically significant (χ²(2) = 27.61, p < 0.001). The overall mean nutrition knowledge score was 56.52 ± 13.26. Figure 1 illustrates the distributional pattern of nutrition knowledge categories among participants. While Table 2 provides the exact frequencies and school-level breakdown, Fig. 1 highlights the overall skew toward moderate knowledge and the relative scarcity of participants in the good knowledge category, a visual pattern useful for appreciating the scope of the educational gap that nutrition interventions must address.

Table 2.

Classification of Nutrition Knowledge of the Female Adolescent by School Setting

Classification of NK	Public School	Private School
	N (%)	N (%)	χ²	P-Value↕
Poor Nutrition Knowledge	18 (11)	12 (9)	27.61	0.001*
Moderate Nutrition Knowledge	130 (81)	82 (63)
Good Nutrition Knowledge	12 (8)	36 (28)
Total	160 (100)	130 (100)

Mean ± SD = 56.52 ± 13.26

(↕) Pearson chi-square test; χ² (2) = 27.61, p < 0.001. *Significant difference at p < 0.05

Fig. 1.

Overall nutrition knowledge of the female adolescents

Health belief model construct scores by school setting

The significant relationship of Health Belief Model constructs by school setting was examined using an independent t-test. Table 3 shows the comparison of mean scores of Health Belief Model constructs by school setting. Private school adolescents had significantly higher mean scores for perceived severity (PSE), while public school adolescents had significantly higher mean scores for perceived susceptibility (PSU), perceived barriers (PBA), cues to action (CTA), and self-efficacy (SEF). No statistically significant difference was observed between school settings for perceived benefits (PBE).

Table 3.

Health Belief (HB) Construct by school setting

				95% CI of the Difference
HB Construct	School Settings	N	Mean	[Lower, Upper]		P-Value
PSE	Public School	160	57.83 ± 8.92	-4.637, -0.266		0.028*
	Private School	130	60.28 ± 9.97
PSU	Public School	160	18.21 ± 4.58	0.294, 2.716		0.015*
	Private School	130	16.71 ± 5.89
PBA	Public School	160	21.40 ± 4.58	1.522, 3.678		0.001*
	Private School	130	18.80 ± 4.71
PBE	Public School	160	36.67 ± 4.64	-2.028, 0.363		0.172
	Private School	130	37.51 ± 5.71
CTA	Public School	160	20.19 ± 2.89	0.159, 1.755		0.019*
	Private School	130	19.23 ± 4.01
SEF	Public School	160	49.90 ± 6.67	2.36, 5.963		0.001*
	Private School	130	45.74 ± 8.91

PSE Perceived Severity, PSU Perceived Susceptibility, PBA Perceived Barriers, PBE Perceived Benefits, CTA Cues to Action, SEF Self-efficacy

*Signify a significant difference at p < 0.05

Perceived dietary behaviour toward healthy eating by school setting

Figure 2 visually contrasts the HBM construct profiles between public and private school adolescents across all six constructs simultaneously. Unlike Table 3, which presents mean scores and significance values. Figure 2 presents visual proportional distribution across unfavourable, moderate, and favourable categories for each construct and school type side by side, providing an accessible overview of the heterogeneity in dietary behaviour perceptions across the two school settings.

Fig. 2.

HBM classification by school settings

Favourable PSU was high in both public (68%) and private schools (72%). PSE was mostly moderate in public schools (65%), while private schools had a higher proportion of unfavourable scores (25% vs. 3%). PBE were largely moderate, especially in private schools (80% vs. 66%), although favourable scores were higher in public schools (30% vs. 9%). Favourable PBA were common in both groups (90% public; 85% private). CTA and SEF were more favourable among public school adolescents (91% and 80%, respectively) compared with private school adolescents (69% and 49%), who showed more moderate and unfavourable responses.

Relationships between nutrition knowledge and perceived dietary behaviour

Table 4 presents Pearson correlation coefficients describing linear relationships among nutrition knowledge and perceived dietary behaviour constructs. Nutrition knowledge was positively correlated with PSE (r = 0.249, p < 0.01), PBE (r = 0.388, p < 0.01), CTA (r = 0.248, p < 0.01), and SEF (r = 0.174, p < 0.01), and negatively correlated with PBA (r = − 0.250, p < 0.01). Nutrition knowledge was not significantly correlated with PSU (r = 0.061, p > 0.05). Additional significant interrelationships were observed among Health Belief Model constructs, including strong positive correlations between PBE and CTA, and between CTA and SEF.

Table 4.

Linear relationship between nutrition knowledge and perceived dietary behaviour towards healthy eating

Variable	Score	PSE	PSU	PBA	PBE	CTA	SEF
Score	1.00	0.249**	0.061	-0.250**	0.388**	0.248**	0.174**
PSE		1.00	0.149*	0.042	0.614**	0.285**	0.440**
PSU			1.00	0.367**	-0.043	0.026	-0.007
PBA				1.00	-0.194**	-0.200**	-0.187**
PBE					1.00	0.618**	0.509**
CTA						1.00	0.506**
SEF							1.00

The negative correlation between nutrition knowledge and PBA (r = − 0.250) indicates that higher nutrition knowledge is associated with lower perceived barrier endorsement. Although PBA items were reverse-scored for classification purposes (such that higher composite scores reflect a more favourable, lower-barrier profile), the correlation with nutrition knowledge is reported in the original directionality to preserve interpretive consistency with the HBM literature

Statistical significant difference at *p < 0.05 and **p < 0.05

Association of nutrition knowledge and sociodemographic factors on health belief model constructs among female adolescents

Table 5 presents the results of the multivariate multiple linear regression examining the independent associations between nutrition knowledge and perceived dietary behaviour, measured by six HBM constructs, after adjusting for school setting, household income, age, and household size. Nutrition knowledge was significantly and independently associated with perceived severity (F = 12.136, p = 0.001, η² = 0.041), perceived benefits (F = 9.526, p = 0.002, η² = 0.033), perceived barriers (F = 37.686, p < 0.001, η² = 0.118), cues to action (F = 25.490, p < 0.001, η² = 0.083), and self-efficacy (F = 15.582, p < 0.001, η² = 0.052). Nutrition knowledge was not significantly associated with perceived susceptibility (F = 2.385, p = 0.124). School setting was also significantly associated with perceived barriers (F = 15.956, p < 0.001, η² = 0.054), cues to action (F = 11.284, p = 0.001, η² = 0.038), and self-efficacy (F = 32.970, p < 0.001, η² = 0.105), but not with perceived severity, perceived susceptibility, or perceived benefits.

Table 5.

Association of nutrition knowledge and sociodemographic factors on health belief model constructs among adolescents (n = 290)

Predictor	HBM Construct	β	95% CI (Lower – Upper)	F	p-value	Partial η²	Sig.
School Settings	PSE	0.032	−0.084 to 0.146	0.479	0.632	0.001	NS
	PSU	0.071	−0.045 to 0.184	1.386	0.240	0.005	NS
	PBA	−0.232	−0.339 to − 0.120	15.956	< 0.001	0.054	Yes
	PBE	0.063	−0.052 to 0.177	1.015	0.315	0.004	NS
	CTA	0.195	0.082 to 0.303	11.284	0.001	0.038	Yes
	SEF	0.324	0.217 to 0.423	32.970	< 0.001	0.105	Yes
Household Income	PSE	0.138	0.023 to 0.249	1.787	0.150	0.019	NS
	PSU	0.118	0.003 to 0.230	1.347	0.259	0.014	NS
	PBA	−0.110	−0.222 to 0.006	1.132	0.337	0.012	NS
	PBE	0.155	0.040 to 0.265	2.318	0.076	0.024	Trend
	CTA	0.063	−0.052 to 0.177	0.374	0.772	0.004	NS
	SEF	0.155	0.040 to 0.265	2.297	0.078	0.024	Trend
Age (years)	PSE	0.071	−0.045 to 0.184	1.550	0.214	0.005	NS
	PSU	0.084	−0.032 to 0.197	1.916	0.167	0.007	NS
	PBA	−0.055	−0.169 to 0.061	0.981	0.323	0.003	NS
	PBE	0.000	−0.115 to 0.115	0.076	0.783	0.000	NS
	CTA	0.063	−0.052 to 0.177	1.123	0.290	0.004	NS
	SEF	0.032	−0.084 to 0.146	0.285	0.594	0.001	NS
Nutrition Knowledge (Score)	PSE	0.196	0.083 to 0.309	12.136	0.001	0.041	Yes
	PSU	0.090	−0.024 to 0.204	2.385	0.124	0.008	NS
	PBA	−0.179	−0.279 to − 0.079	9.526	0.002	0.033	Yes
	PBE	0.344	0.241 to 0.447	37.686	< 0.001	0.118	Yes
	CTA	0.286	0.183 to 0.389	25.490	< 0.001	0.083	Yes
	SEF	0.225	0.122 to 0.328	15.582	< 0.001	0.052	Yes
Household Size	PSE	0.045	−0.071 to 0.159	0.429	0.513	0.002	NS
	PSU	0.045	−0.071 to 0.159	0.623	0.431	0.002	NS
	PBA	−0.045	−0.159 to 0.071	0.606	0.437	0.002	NS
	PBE	0.045	−0.071 to 0.159	0.515	0.474	0.002	NS
	CTA	0.000	−0.115 to 0.115	0.110	0.740	0.000	NS
	SEF	0.032	−0.084 to 0.146	0.276	0.600	0.001	NS

For Nutrition Knowledge (Score), β and 95% CI are standardised regression coefficients. For all other predictors (School Settings, Household Income, Age, Household Size), β represents the correlation coefficient; Negative β values for PBA indicate that higher values on the predictor are associated with lower perceived barriers (i.e., a favourable direction). For School Settings, a negative PBA β indicates that private school students report fewer barriers

Yes =p< 0.05 (statistically significant); Trend =p< 0.10; NS = not significant (p≥ 0.10)

Figure 3 illustrates the pattern and magnitude of associations between selected predictors and the six Health Belief Model (HBM) constructs using effect sizes (Partial η²). Asterisk indicates statistically significant associations (p < 0.05), while lighter tones represent non-significant effects.

Fig. 3.

Association between selected predictors and the six Health Belief Model (HBM) constructs

Figure 3 provides a visual summary of the relative effect sizes (partial η²) for all predictor–HBM construct associations in the multivariate model. This heatmap-style representation helps to compare the relative strength of nutrition knowledge versus sociodemographic predictors across all six constructs at a glance, complementing the detailed F-statistics and p-values in Table 5 with an interpretively accessible visual of effect magnitude and statistical significance patterns.

In the present study, the association between nutrition knowledge and perceived barriers yielded a moderate effect size (η² = 0.118), approaching the threshold for a large effect and indicating that nutrition knowledge accounts for approximately 11.8% of the variance in perceived barrier profiles, a substantial proportion for a single predictor in a complex multivariate model. The associations with cues to action (η² = 0.083) and self-efficacy (η² = 0.052) represent medium-sized effects, underscoring the practical importance of nutrition knowledge in shaping motivational and volitional aspects of dietary behaviour. Smaller, yet significant, effect sizes were observed for perceived severity (η² = 0.041) and perceived benefits (η² = 0.033), consistent with the expectation that these constructs are also associated with lived experience, social environment, and developmental factors beyond nutrition knowledge alone.

Discussion

This study examined the relationship between nutrition knowledge (NK) and perceived dietary behaviour toward healthy eating among in-school female adolescents in Ogun State, Nigeria, using the Health Belief Model (HBM). The findings contribute to the growing evidence base on adolescent girls’ nutrition by elucidating how cognitive and behavioural perceptions interact with sociodemographic factors associated with dietary behaviour during a critical stage of growth and future reproductive health.

The study population consisted predominantly of mid-adolescent girls (mean age 15.19 ± 1.05 years) drawn from both public and private secondary schools. Most parents had attained secondary or tertiary education, and over half of households reported monthly incomes exceeding ₦100,000, suggesting relatively favourable socioeconomic conditions compared with many adolescent populations in Nigeria. Nevertheless, the average household size of 5.38 persons reflects a common household structure in which food resources may be shared among multiple dependents, potentially constraining individual dietary quality. These findings indicate that although parental education and income may improve access to nutrition information and food resources, they do not necessarily ensure optimal dietary behaviours among adolescents. Similar patterns have been reported in Ogun State, where gaps between nutrition knowledge and dietary perceptions persist due to misconceptions, affordability constraints, and food availability challenges [4, 25].

Significant differences in nutrition knowledge were observed between school settings, with private school adolescents demonstrating higher proportions of good NK compared with public school adolescents (p = 0.001). This finding aligns with evidence suggesting that school context, including instructional quality and access to educational resources, plays a critical role in shaping adolescents’ health knowledge. Higher nutrition knowledge has been associated with more favourable dietary perceptions and, in some settings, healthier eating behaviours [26]. Although the overall mean NK score was moderate (56.52 ± 13.26), the predominance of moderate knowledge indicates substantial scope for strengthening school-based nutrition education, particularly within public schools where structured curricula and tailored instructional resources may be limited.

Consistent with these findings, a recent study among in-school adolescents in Ogun State similarly reported predominantly moderate nutrition knowledge, with significantly higher levels among private school students [12]. This school-based disparity in nutrition knowledge has similarly been documented in other low- and middle-income country contexts. Raut et al. [26] reported significant improvements in nutrition knowledge among Nepalese school adolescents following structured education interventions, with baseline knowledge deficits particularly concentrated in public school settings. Abubakar et al. [6] likewise identified educational resource disparities as a key driver of knowledge gaps among Nigerian adolescents across different school types. Together, these findings suggest that even in relatively socioeconomically advantaged contexts, adolescents may lack sufficiently comprehensive and actionable nutrition knowledge. The persistence of moderate knowledge highlights the need to improve not only the coverage but also the depth and practical relevance of nutrition education for female adolescents.

Across HBM constructs, favourable scores were most evident for perceived susceptibility, perceived barriers, cues to action, and self-efficacy among public school adolescents, whereas private school adolescents demonstrated higher perceived severity. These findings suggest that while adolescents across school settings recognise vulnerability to unhealthy dietary practices and acknowledge both benefits and barriers to healthy eating, the internalisation of these perceptions differs by context. Higher cues to action and self-efficacy among public school adolescents may reflect collective social environments or school-level health messaging, whereas higher perceived severity among private school adolescents may be attributable to greater exposure to health information emphasising long-term consequences of unhealthy diets [4].

Nutrition knowledge showed significant positive correlations with PSE, PBE, CTA, and SEF, and a significant negative correlation with PBA, indicating that higher nutrition knowledge is associated with fewer perceived dietary barriers. However, nutrition knowledge was not significantly correlated with PSU, suggesting that factual nutritional knowledge may be insufficient, in isolation, to alter adolescents’ personal appraisals of their susceptibility to diet-related health risks. This is consistent with evidence demonstrating that nutrition knowledge, attitudes, and self-efficacy are key determinants of adolescents’ dietary behaviour, with adolescents who possess greater knowledge and confidence being more likely to adopt healthier eating patterns [27].

After adjustment for sociodemographic characteristics, nutrition knowledge remained independently and significantly associated with five of the six HBM constructs: PSE, PBE, PBA, CTA, and SEF. These findings support the role of knowledge as a foundational cognitive determinant that is simultaneously associated with motivational appraisals, barrier recognition, and behavioural confidence. Previous studies similarly report that higher nutrition knowledge is associated with healthier dietary patterns, though effect sizes are often modest once broader contextual factors are considered [26, 28]. The consistent association of nutrition knowledge with PSE suggests that factual understanding of dietary consequences equips adolescents to recognise the seriousness of diet-related health risks, an important precondition for motivating protective dietary behaviour.

The absence of a significant independent association between nutrition knowledge and perceived susceptibility (PSU) warrants deeper theoretical reflection. PSU, defined as an individual’s perceived likelihood of experiencing nutrition-related health consequences, may be more strongly shaped by personal or vicarious experience of illness and by social comparison processes than by formal nutritional knowledge. Among adolescents, the well-documented phenomenon of optimistic bias, the tendency to underestimate personal risk relative to peers, may attenuate the relationship between factual knowledge and perceived personal vulnerability, even when nutritional knowledge is adequate [29]. Furthermore, PSU is inherently ego-involved and emotionally laden in ways that abstract nutritional facts may not easily penetrate. These findings suggest that nutrition education alone may be insufficient to alter susceptibility appraisals; supplementary strategies such as personalised risk communication, peer testimonials, or narrative-based approaches that render dietary risk personally proximate and emotionally vivid may be required. Future longitudinal studies should examine whether accumulating nutritional knowledge over time eventually shapes susceptibility appraisals as adolescents mature and encounter real-world diet-related consequences.

The strongest associations were observed between nutrition knowledge and SEF, as well as cues to action. From a social cognitive perspective, knowledge is a prerequisite for perceived behavioural capability, as individuals are unlikely to feel confident performing behaviours they do not understand [28]. Evidence from a systematic review by Vaitkeviciute et al. [30] corroborates these findings, demonstrating that food literacy, a construct closely related to nutrition knowledge, is consistently associated with dietary self-efficacy and behavioural intent among adolescents across diverse cultural settings, suggesting that the knowledge-SEF pathway observed in the current study is generalisable beyond the Nigerian context. Similarly, Sharma et al. [27] reported significant positive associations between nutrition knowledge, SEF, and dietary attitudes among community-based adolescents in India, further supporting the cross-cultural relevance of these associations.

Evidence suggests that nutrition interventions combining education with skill-building components such as goal-setting, practical food selection guidance, and peer modelling are more effective than information-only approaches in improving adolescents’ dietary behaviours [30]. The association with cues to action further highlights the role of supportive environments and social prompts in translating knowledge into behaviour, particularly within school contexts [31, 32].

Public Health Implications

Female adolescents represent a critical population for public health intervention due to their increased nutritional requirements during growth and the implications of their dietary behaviours for future reproductive health and intergenerational nutrition outcomes. Adolescents with stronger nutrition knowledge and supportive behavioural perceptions are more likely to adopt healthy eating practices that reduce the risk of micronutrient deficiencies, anaemia, and diet-related non-communicable disease conditions that disproportionately affect adolescent girls [25]. Internationally, interactive and skills-based nutrition education programmes, including peer-led, digital, and community-linked approaches, have demonstrated effectiveness in improving adolescents’ dietary knowledge and self-efficacy beyond knowledge-only interventions [17, 32]. These approaches align with the findings of the present study, in which the strongest associations were observed for the motivational and volitional constructs (self-efficacy and cues to action) that are most amenable to behaviour change strategies.

The moderate distribution of nutrition knowledge and its consistent association with key behavioural determinants underscore the need to strengthen school-based nutrition education, particularly within public schools. Formal curricula tailored to adolescent girls have the potential to improve nutrition knowledge and reinforce positive HBM constructs. Interactive, context-specific approaches including digital, peer-led, and community-linked strategies have demonstrated promise in enhancing nutrition knowledge and supporting behavioural intent among adolescents [32]. Public health programmes should incorporate behavioural frameworks such as the HBM to address perceived barriers and strengthen self-efficacy, while also engaging families and communities to support healthy food environments. Explicit consideration of gendered social norms and adolescents’ autonomy in food choices is essential for designing equitable and effective interventions.

Study strengths and limitations

This study has several limitations that should be considered when interpreting the findings. First, the cross-sectional design precludes causal inference; the associations reported between nutrition knowledge and HBM constructs are correlational in nature, and directionality cannot be established without longitudinal evidence. Second, reliance on self-reported data for both nutrition knowledge and perceived dietary behaviour introduces the possibility of social desirability bias and recall error. Third, the multi-stage sampling approach, which incorporated purposive selection of the Opeji zone, Obantoko community, and four secondary schools, was adopted to ensure representation across school types and to capitalise on the heterogeneous sociodemographic composition of this community. The sample is most appropriately representative of female adolescents attending public and private secondary schools in semi-urban peri-Abeokuta communities within Odeda LGA. The observed knowledge and perception patterns should be interpreted as hypothesis-generating and context-specific. Fourth, the study focused exclusively on female adolescents; male adolescents and those in other school or community contexts were not included, which constrains the breadth of inference. Future research employing probability-based sampling across multiple LGAs and incorporating both sexes would improve generalisability.

Conclusion

This study demonstrates that nutrition knowledge is positively and independently associated with several key HBM-based dietary perceptions, including perceived severity, perceived benefits, reduced perceived barriers, cues to action, and self-efficacy among in-school female adolescents in Ogun State, Nigeria. These perceptual constructs represent proximal cognitive and motivational determinants of dietary behaviour within the HBM framework, rather than direct measures of dietary intake or food consumption. Strengthened, theory-driven nutrition education, particularly within public schools, combined with behavioural strategies that target perceived barriers and build self-efficacy, may reinforce the favourable dietary cognitions that are theoretically linked to healthier eating practices. However, longitudinal research that incorporates direct dietary intake assessment is essential to confirm whether improvements in these cognitive constructs translate into sustained behavioural change. The findings have implications for the design of school-based adolescent nutrition interventions in Nigeria and comparable low- and middle-income country settings.

Abbreviations

CI

Confidence Interval

CVI

Content Validity Index

CVR

Content Validity Ratio

HBM

Health Belief Model

NK

Nutrition Knowledge

WHO

World Health Organisation

PBE

Perceived Benefits

PBA

Perceived Barriers

PSE

Perceived Severity

PSU

Perceived Susceptibility

CTA

Cues to Action

SEF

Self-Efficacy

Dare D. Ademiluyi

is a nutrition and public health researcher with research interests in Maternal Nutrition, adolescent nutrition, dietary behaviour, and theory-driven nutrition interventions in low- and middle-income countries. A lecturer in the Federal University of Medicine and Medical Sciences in the department of Human Nutrition and Dietetics. He is also in a research collaboration with the University of Ibadan.

Authors’ contributions

DD conceived and designed the study and was solely responsible for the study concept; developed the methodology; supervised data collection; conducted formal analysis; interpreted the data; managed project administration; and drafted the original manuscript. YO contributed to methodology refinement, assisted with data collection, and reviewed the manuscript. AR and ED contributed to investigation and data curation and reviewed the manuscript. AE, AB, and OB provided academic supervision, contributed to methodology development, and critically reviewed and edited the manuscript. OA and PM assisted with data collection and manuscript review.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

The datasets generated and/or analysed during the current study are not publicly available due to ethical restrictions related to participant confidentiality but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study involved human participants. Ethical approval was obtained from the Federal Medical Centre, Abeokuta Health Research Ethics Committee (FMCA/470/HREC/01/2023/56). All study procedures were conducted in accordance with the ethical standards of the approving committee and the principles of the Declaration of Helsinki. Written informed consent was obtained from parents or legal guardians of all participating adolescents, and written assent was obtained from the adolescents before data collection. Participation was voluntary, and confidentiality and anonymity were ensured throughout the study.

Consent for publication

Not applicable. The manuscript does not contain any person’s data in an identifiable form, including images, videos, or personal identifiers.

Competing interests

The authors declare no competing interests.