Healthy Eating Determinants: A Study among Malaysian Young Adults

This study aimed to examine the effect of health consciousness, knowledge about healthy food, attitudes toward healthy food, subjective norms, and perceived behavioural control on the intention to consume healthy food, which subsequently affects the consumption of healthy food among Malaysian young adults. The current study also examined the moderating effect of perceived barriers on the association between intention to consume healthy food and the consumption of healthy food. This study adopted a cross-sectional design and collected quantitative data from 1651 Malaysian young adults (between the age of 18 and 40 years) by sharing a Google form link through social media. The findings reveal that health consciousness, knowledge about healthy food, attitude toward healthy food, subjective norms, and perceived behavioural control had a significant positive effect on the intention to consume healthy food. Findings also show that the intention to consume healthy food has a significant positive effect on the consumption of healthy food among Malaysian young adults. Furthermore, the findings reveal the positive and significant mediating effect of the intention to consume healthy food and the significant moderating effect of perceived barriers on the association between the intention to consume healthy food and the consumption of healthy food. The multi-group analysis revealed that the effect of perceived barriers on the consumption of healthy food and the moderating effect of perceived barriers were significantly higher among urban respondents. Health and agriculture policymakers should focus on the attributes of healthy eating practices and their health benefits to promote the mass adoption of healthy food among Malaysian young adults.


Introduction
Modern society is confronting increased health issues as the population's eating habits and the lack of healthy food consciousness had caused obesity and poor nutrition and eating conditions among young adults [1]. About 2.8 million people die worldwide because due to being overweight or having melancholic obesity each year [2]. Since 2000, Malaysians are facing issues of obesity and eating disorders [3]. Diet-related diseases are on the rise in Malaysia, and this is increasing the socioeconomic burdens on middle-income households [1]. Scientific evidence shows that unhealthy and unbalanced food increases the risk of hypertension, cardiovascular diseases, and diabetes [4]. Whole grains, fruits, vegetables, and legumes are essential for a healthy life besides reducing certain medical conditions [5].
Thirty-nine percent of the world population is overweight, and about 13% of the population is obese [3]. Malaysians are the most obese citizens in Southeast Asia, in which 48% of the population is intention and consumption of healthy foods. Individual attitude, subjective norm, and perceived behavioural control can significantly influence the intention to consume healthy foods [5].

Consumption of Healthy Food (CHF)
Healthy foods are gaining higher acceptability among the general population in recent times [8]. Factors that can lead to higher acceptability are the more significant concern for personal health, a higher rate of obesity among young adults, and increasing health disorders [1,3]. Individuals that have the understanding and knowledge of healthy food can improve the intention to use healthy foods [4]. Moreover, the concern of personal health can help to formulate healthy eating habits. Governmental agencies are increasingly reporting upsurges in health ailments among young adults [3]. Moreover, advancements in nutrition research enable a better understanding of the human body's nutritional requirements [4]. The healthy balance lifestyle relies on sensible daily food based on dietary recommendations and guidelines. Young consumers are willing to use conventional and newly developed food supplements to achieve the benefits of food, and they are willing to pay premium prices for nourishment food products [3,7].
Healthy foods improve individual health. Healthy foods provide adequate nutritional ingredients that reduce disease risks and improve health issues among individuals [1]. Healthy foods are a mix of commonly available food that have health-related beneficial effects on human health [11]. Healthy food has nutritional and physiological effects on the human body [3]. Health-related food offerings are increased in Malaysia, and the percentage of healthy food accounts for about 40% of total food offerings [10].

Health Consciousness (HTC)
Health consciousness (HTC) is the perceived importance of health in an individual's daily life routines. It reflects the individual's willingness to adopt a healthy routine, food, and lifestyle [13]. Healthy food is indispensable for a healthy lifestyle and provides the necessary minerals and proteins to boost health and reduce the risk of diseases (6]. HTC is vital for a healthy life. An individual's health consciousness can significantly affect the intention to use healthy foods [4]. Singh and Verma [6] postulated that HTC positively and significantly (β = 0.18, p = 0.01) influences the intention to consume healthy food (IHF) among Indian consumers. Hypothesis 1 (H1). HTC has a significant positive effect on IHF among Malaysian young adults.

Knowledge about Healthy Food (KHF)
Knowledge is recognised as a critical factor for human behaviours. Food knowledge influences an individual's eating behaviours [6]. Food-selecting behaviour is influenced by product knowledge, and knowledge enhances product understanding and healthy food behaviour [11]. Low levels of healthy food knowledge demonstrate poor eating behaviours among young adults [5]. Consumer awareness and knowledge can develop the intention to use environmentally friendly products and innovations [6]. Lee et al. [11] postulated that KHF significantly (β = 0.297, p = 0.000) influence the intention to use healthy food among Korean adults.
Hypothesis 2 (H2). KHF has a significant positive effect on the IHF among Malaysian young adults.

Attitude toward Healthy Food (AHF)
Attitude represents the overall evaluation of the perceived consequences of particular behaviour under the consideration of an individual [14]. A positive attitude toward behaviour can guide the Foods 2020, 9, 974 4 of 18 intention to perform that behaviour [15]. Rezai et al. [10] postulated that ATFs positively and significantly (β = 0.116, p = 0.000) influences IHF among the Malaysian sample. Moreover, Nguyen et al. [8] reported that attitude toward functional food (β = 0.353, p = 0.000) positively affects the intention to purchase function food among Vietnamese youth. Hence, this study proposed the following hypothesis: Hypothesis 3 (H3). AHF has a significant positive effect on IHF among Malaysian young adults.

Subjective Norms (SBNs)
Subjective norms (SBNs) are the perceived social pressure on an individual to perform or not to perform certain behaviours [14]. Perception of social pressure induces the social behaviours of individuals. Rezai et al. [10] postulated that subjective norm positively and significantly (β = 0.198, p = 0.000) influences the intention to purchase healthy food among Malaysian consumers. Furthermore, Menozzi et al. [5] postulated that SBN significantly (β = 0.56, p = 0.000) influences the intention to use green food among Italian students. Therefore, the following hypothesis is proposed: Hypothesis 4 (H4). SBNs have a significant positive effect on IHF among Malaysian young adults.

Perceived Behavioural Control (PBC)
Individual perception of ability effects the performance of the behaviour [10]. Perceived behavioural control (PBC) is an individual's understanding of the ease or difficulty associated with the performance of a behaviour. PBC affects the intention of green behaviours [15]. Menozzi et al. [5] postulated that PBC significantly (β = 0.69, p = 0.000) influences the intention to use green food among Italian students. Therefore, the following hypothesis is proposed: Hypothesis 5 (H5). PBC has a significant positive effect on IHF among Malaysian young adults.

Intention to Consume Healthy Food (IHF)
Intention is the first outcome of the TPB based on attitude, SUN, and PBC [14]. There is empirical evidence that intention leads to the consumption behaviours [15]. Nguyen et al. [8] postulated that PBC significantly (β = 0.35, p = 0.000) influences CHF among young Vietnamese consumers. This study proposed the following hypothesis: Hypothesis 6 (H6). IHF has a significant positive effect on CHF among Malaysian young adults.

Mediating Effect of the Intention to Consume Healthy Food
Intention is the integral outcome of the TPB that leads to a particular behaviour. Singh and Verma [6] reported that intention mediates the three factors of TPB for the consumption behaviours for healthy organic food among Indian consumers. Moreover, the current work expanded the TPB with the factors of HTC and KHF. This study proposed the following hypothesis: Hypothesis 7 (H7). IHF mediates the relationship between HTC, KHF, ATF, SUN, and PBC on CHF among Malaysian young adults.

Moderating Effect of Perceived Barriers (PBS)
Healthy foods are perceived as beneficial and good for health. Instead of having HTC and KHF, the CHF is scant [3]. Perceived barriers (PBS) restrict CHFs [10]. Healthy foods are perceived as the difficulty to find, cook, and eat [6]. These PBS reveal the individuals' belief that healthy food is costly, difficult to procure, and time-consuming to cook [7]. PBS have significant unfavourable effects on the intention to consume healthy foods [3]. Rezai et al. [10] postulated that PBS significantly (β = −0.083, p = 0.000) reduces the IHF among the Malaysian sample. Higher intention leads to higher consumption behaviour toward healthy food. PBS have adverse effects on the consumption behaviour of healthy food. This study, therefore, examined the moderating effect of PBS between the IHF and the CHF. Hence, the following hypothesis is proposed: Hypothesis 8 (H8). PBS moderate the relationship between IHF and CHF among Malaysian young adults.
All association hypothesized and tested associations are presented in Figure 1.

Data Collection and Study Sample Design
This study examined the effect of HTC, KHF, ATF, SUN, and PBC on IHF, which subsequently affects CHF among Malaysian young adults. In Malaysia, "youth" can be defined as those aged between 15 and 40 years old [16]. However, in order to avoid ethical issues and/or parental permission requirements for the collection of data, those aged below 18 years were excluded from this study, with those aged between 18 and 40 years meeting the criteria for Malaysian young adults. This study adopted the cross-sectional design and collected quantitative data from 1651 Malaysian young adults through an online survey for the first two weeks of April 2020. This study designed a Google form, highlighted the purpose, reported the procedure of the study, and collected informed consent from all respondents before they participated in the survey. The questionnaire was distributed by sharing the link of the questionnaire form using social media.

Survey Instrument
Explicit and straightforward statements were designed to gauge responses to the given constructs. This approach can obtain an appropriate and accessible understanding of the survey respondents. A total of five questions measuring HTC were adopted from several studies [7,8]. This study measured KHF using five questions adopted from several studies [13,17]. Five questions measured AHF adopted from several studies [5,10]. Five questions were adopted from several studies to measure SUN [7,10]. Five questions to measure PBC were adopted from several studies [5,15]. Five questions were adopted from several studies to measure PBS [10,13]. Four questions measuring IHF were adopted from several studies [5,10]. One question was adopted from a study by Menozzi et al. [5] to measure the CHF. All question items were assessed against a 7-point Likert scale, except for CHF, as this was measured as 'yes' or 'no'. All questions are presented in Appendix A.

Assessment of Common Method Variance (CMV)
CMV issue is normal in social science research due to the data collection methods and techniques

Data Collection and Study Sample Design
This study examined the effect of HTC, KHF, ATF, SUN, and PBC on IHF, which subsequently affects CHF among Malaysian young adults. In Malaysia, "youth" can be defined as those aged between 15 and 40 years old [16]. However, in order to avoid ethical issues and/or parental permission requirements for the collection of data, those aged below 18 years were excluded from this study, with those aged between 18 and 40 years meeting the criteria for Malaysian young adults. This study adopted the cross-sectional design and collected quantitative data from 1651 Malaysian young adults through an online survey for the first two weeks of April 2020. This study designed a Google form, highlighted the purpose, reported the procedure of the study, and collected informed consent from all respondents before they participated in the survey. The questionnaire was distributed by sharing the link of the questionnaire form using social media.

Survey Instrument
Explicit and straightforward statements were designed to gauge responses to the given constructs. This approach can obtain an appropriate and accessible understanding of the survey respondents. A total of five questions measuring HTC were adopted from several studies [7,8]. This study measured KHF using five questions adopted from several studies [13,17]. Five questions measured AHF adopted from several studies [5,10]. Five questions were adopted from several studies to measure SUN [7,10]. Five questions to measure PBC were adopted from several studies [5,15]. Five questions were adopted from several studies to measure PBS [10,13]. Four questions measuring IHF were adopted from several studies [5,10]. One question was adopted from a study by Menozzi et al. [5] to measure the CHF.
All question items were assessed against a 7-point Likert scale, except for CHF, as this was measured as 'yes' or 'no'. All questions are presented in Appendix A.

Assessment of Common Method Variance (CMV)
CMV issue is normal in social science research due to the data collection methods and techniques [18]. Harman's [19] one-factor test was suggested to estimate the impact of CMV on study constructs [18]. One-factor Harman's test revealed that CMV was not a critical matter for study, as the main factor accounted for 31.84% variance and less than the recommended limit of 50% [18].

Multivariate Normality
SEM-PLS is not associated with multivariate normality in the data, as it is a non-parametric analysis instrument [20]. Multivariate data normality was tested as suggested by Peng and Lai [21] using an online tool of web power (https://webpower.psychstat.org/wiki/tools/index) to confirm data normality. The test results confirm that the data set is not as normal as Mardia's multivariate coefficient p-values that are less than 0.05 [22].

Data Analysis Method
Partial least squares structural equation modelling (PLS-SEM) was used with Smart-PLS software 3.1 for data analysis. PLS-SEM is a multivariate analysis instrument used to gauge the path models that have latent constructs with composites [20]. PLS-SEM empowers the researcher to tackle non-normal and small data sets. Furthermore, PLS-SEM has a casual-predictive nature with an undisturbed supposition of goodness-of-fit estimation compared to covariance-based SEM [23]. Two-step techniques analysed data with PLS-SEM, and the first measurement was performed to test the model's reliability and validity at the constructs' level [20]. The second stage was executed for the estimation of the structural model and the investigation of study hypotheses with significance levels [23]. Model estimation was performed with r 2 , Q 2 , and the effect size f 2 that describe the path effect from the exogenous construct for the endogenous construct [20].
Multi-group analysis (MGA) in PLS-SEM permits the researchers to distinguish the differences between the pre-defined groups [24]. MGA is a convenient procedure to evaluate the differences between the groups inside the data set [20]. The MGA evaluates the distinctions among the structural paths of several groups in the data sets [24]. MGA was performed with the development of groups within data based on the categorical variables of interest like age, gender, or income. Then, the path coefficients for the groups were analysed whether two groups were significantly different from each other or not based on the procedures suggested by Henseler et al. [24]. The differences within the data set were based on the characteristics of samples that may not be noticeable in the collected data. Path coefficients of the group data can confirm the statistical variance using MGA to establish significant statistical differences among data based on categorical bases [24].
Importance-performance map analysis (IPMA) categorises the study constructs into relatively high to low by their corresponding importance and performance of the endogenous construct [23]. IMPA distinguishes the possible area of improvement from the managerial and literature perspective. IPMA analysis transforms the total effect of the rescaled variables totals in the un-standardised technique [25]. Rescaling is recognised for every latent constructs' score between 0 and 100. The mean value of the latent variable score represents the performance of the latent variable, where 0 indicates the least and 100 indicates the maximum importance in the performance of the endogenous construct [20].

Reliabilities and Validities
Following the approval of Hair et al. [20], the reliabilities for study's latent constructs can be achieved and assessed by Cronbach's alpha (CA), DG rho, and composite reliability (CR). Cronbach's alpha values for each construct are above the threshold of 0.70, and the minimum value of Cronbach's alpha value achieves 0.781 [23]. The results are reported in Table 2. Furthermore, all DG rho values are above the threshold of 0.70, where the minimum value of DG rho is 0.783 [20]. Moreover, CR values are well beyond the threshold of 0.70, where the lowest value of CR value is 0.850 [23]. These outcomes indicate that the latent constructs realised the suitable reliabilities, and they performed well for the later stage of analysis. AVE for all items for each construct must be above 0.50 score to the extent the adequate convergent validity to support the uni-dimensionality concept for each construct [20]. Items display that the constructs have acceptable convergent validity (see Table 2.). All the VIF values for each construct are below the threshold of 3.3 that reveals no concern of multicollinearity [23]. The item loading and cross-loading for the confirmation of construct discriminant validity are described in Tables 3 and 4, respectively.    All the study constructs have appropriate discriminant validities (see Table 3). Additionally, the Fornell-Larcker criterion (1981) and HTMT ratio had achieved the discriminant validity of each study construct. The Fornell-Larcker criterion was assessed with the square root of the respective construct's AVE, and the square root of AVE for the construct must be higher than the correlation among other constructs [20]. HTMT ratio needs to be less than 0.85 to establish discriminant validity for each study construct [26]. Tables 3 and 4 show that the study has adequate discriminant validity for each construct.

Path Analysis
The reliabilities and validities from the structural assessment of the study model are satisfactory. The next measurement assessment examined the study hypothesis. The adjusted r 2 value for the five exogenous constructs (i.e., HTC, KHF, ATF, SUN, and PBC)) on IHF explains the 50.3% change in the intention to consume healthy food. The predictive relevance (Q 2 ) value for the part of the model is 0.343, indicating a large predictive relevance [23]. The adjusted r 2 value for the exogenous construct (i.e., intention to consume healthy food) on the CHF elucidates 8.2% change in the CHF. The predictive relevance (Q 2 ) value for the part of the model is 0.078, indicating small predictive relevance [23].
Model standardised path values, t-values, and significance level are illustrated in Table 5. The path coefficient between HTC and IHF (β = 0.344, t = 10.825, p = 0.000) indicates a significant and positive effect of HTC on the intention to consume healthy food. The result forms significant statistical support for H1. The path value for KHF and IHF (β = 0.203, t = 6.556, p = 0.000) shows the impact of KHF for the intention to consume healthy food, which is positive and significant; hence, it offers significant statistical support for H2. The path between AHF and IHF (β = 0.109, t = 4.289, p = 0.000) shows the influence of AHF in influencing the intention to consume healthy food, which is positive and significant; it supports H3. The path coefficient for SBN and IHF (β = 0.076, t = 2.815, p = 0.003) shows a positive and significant effect; it supports H4. The path between PBC and IHF (β = 0.102, t = 3.381, p = 0.000) shows the influence of PBC in influencing the intention to consume healthy food, which is positive and significant; it supports H5. The path coefficient for IHF and CHF (β = 0.267, t = 11.570, p = 0.000) shows a positive and significant effect; it supports H6. Table 5 shows the path coefficients.

Mediation Analysis
The mediation effect of IHF was tested with H 7A for the relationship between HTC and CHF. The result reveals that IHF mediates the relationship between HTC and CHF (β = 0.092, CI min = 0.072, CI max = 0.112, p = 0.000) and supports H 7A . For H 7B , the relationship between KHF and CHF is mediated by IHF. The result shows that IHF mediates the relationship between KHF and CHF (β = 0.054, CI min = 0.039, CI max = 0.071, p = 0.000); it supports H 7B . For H 7C, the relationship between AHF and CHF is mediated by IHF. The result shows that IHF mediates the relationship between AHF and CHF (β = 0.029, CI min = 0.017, CI max = 0.041, p = 0.000); it supports H 7C . For H 7D, the relationship between SBN and CHF is mediated by IHF. The result reveals that IHF mediates the relationship between SBN and CHF (β = 0.020, CI min = 0.008, CI max = 0.033, p = 0.004); it supports H 7D . For H 7E , the relationship between PBC and CHF is mediated by IHF. The result reveals that IHF mediates the relationship between PBC and CHF (β = 0.027, CI min = 0.014, CI max = 0.042, p = 0.001); it supports H 7E . The mediation results are presented in Table 6.

Multi-Group Analysis
Multi-group analyses were executed to match the results for different groups based on gender, living area, and education. One non-parametric test was employed to evaluate the differences in the vital association between the model based on gender, areas of living, and education of the sample. Table 7 shows the path values for two groups with the differences within the groups with the p-values as recommended by Henseler et al. [24]. P MGA represents the p-values using the multi-group analysis of PLS-SEM as the measure for the significance of the difference between groups [24].

Effects of Gender
The results of the groups are based on gender in the sample. Gender shows no significant difference in the relationships of the model. The variance of gender does not influence the relationship between study models.

Effects of Living Area
The results of the two groups are based on the living area-namely, urban and rural. Living area shows a significant difference in the relationship between PBS and CHF, IHF, and CHF for CHF. Living area does not influence the variance between the model's other paths.

Effects of Education
The results of the two groups are based on the education of the sample. The variance of education does not influence the variance between the study's paths.

Effects of Household Income
The results of the three groups (below RM2500, between RM2501-RM5000; below RM2500 and between RM5001-RM7500; RM2501-RM5000 and RM5001-RM7500) presented in Table 7 are based on the respondents' household income. The findings revealed a significant difference in the relationship between the effect of AHF on IHF, and PBC on IHF among the respondents with a household income below RM2500 and between RM2501 and RM5000. The findings also showed a significant difference in the relationship between the effect of PBC on IHF among respondents with a household income between RM2501-RM5000 and RM5001-RM7500. Figure 2 and Table 8

Discussion
The first five hypotheses evaluated the effects of HTC, KHF, AHF, SBN, and PBC on IHF. The study findings support the argument that HTC (f 2 =0.110) has a medium effect on IHF, and KHF (f 2 =0.032) has a small effect on IHF. However, the effects of AHF (f 2 =0.012), SBN (f 2 =0.006), and PBC (f 2 = 0.010) have a significant but small effect on IHF among Malaysian young adults [23]. Study findings are parallel to the findings by Hoque et al. [4] that HTC and knowledge influence the intention to consume healthy food. HTC and food knowledge were also found to significantly influence intention in developing countries as well [8]. Furthermore, the findings from the study revealed that AHF, SBN, and PBC affected IHF, which correspond with the results in a study by Menozzi et al. [5]. However, the effect sizes of the AHF, SBN, and PBC on IHF were significant but below the small effect threshold compared to the results of Menozzi et al. [5]. Accordingly, this indicates the low level of AHF, SBN and PBC among the Malaysian respondents in having the intention to consume healthy food.

Discussion
The first five hypotheses evaluated the effects of HTC, KHF, AHF, SBN, and PBC on IHF. The study findings support the argument that HTC (f 2 =0.110) has a medium effect on IHF, and KHF (f 2 =0.032) has a small effect on IHF. However, the effects of AHF (f 2 =0.012), SBN (f 2 =0.006), and PBC (f 2 = 0.010) have a significant but small effect on IHF among Malaysian young adults [23]. Study findings are parallel to the findings by Hoque et al. [4] that HTC and knowledge influence the intention to consume healthy food. HTC and food knowledge were also found to significantly influence intention in developing countries as well [8]. Furthermore, the findings from the study revealed that AHF, SBN, and PBC affected IHF, which correspond with the results in a study by Menozzi et al. [5]. However, the effect sizes of the AHF, SBN, and PBC on IHF were significant but below the small effect threshold compared to the results of Menozzi et al. [5]. Accordingly, this indicates the low level of AHF, SBN and PBC among the Malaysian respondents in having the intention to consume healthy food. The next hypotheses proposed the effects of PBS and IHF on CHF. The study findings support the argument that PBS (f 2 =0.019) has a small effect on CHF, match with the results reported by Nguyen et al. [13] in which the influences of PBS are both significant and negative regarding the use of green products. The results of our study also suggest a similar pattern in that PBS negatively influences the CHF and reduces the CHF among the study sample. However, the effect of IHF (f 2 =0.077) has a small, positive, and significant effect on CHF [23]. Although the findings from our study are comparable to those claimed by Menozzi et al. [5] and Maichum et al. [27] in which intention significantly and positively affects consumption behaviour.
The next mediating effect of IHF was assessed with five mediation hypotheses. H 7A investigated the mediating effect of IHF between HTC and CHF. The finding approves the meditating effect of IHF (β = 0.092, p = 0.000) for the relationship between HTC and CHF among Malaysian young adults for the CHF. The findings of this study support several studies [15,27]. H 7B hypothesised about the meditating effect of IHF between KHF and CHF. The finding confirms the meditating effect of IHF (β = 0.054, p = 0.000) for the relationship between KHF and CHF for the healthy food consumption among Malaysian young adults. The finding of this study is supported by Maichum et al. [27].
The next hypothesis, H 7C , evaluated the meditating effect of IHF between AHF and CHF. The finding confirms that the significant mediating effect of IHF (β = 0.029, p = 0.000) for the relationship between AHF and CHF. The study results are supported by Yadav and Pathak [15]. Furthermore, H 7D estimated the meditating effect of IHF between the relationship of SBN and CHF. The finding confirms the mediating effect of IHF (β = 0.020, p = 0.004) for the relationship between SBN and CHF. The study results are supported by Yadav and Pathak [15]. H 7E assessed the mediating effect of IHF between PBC and CHF. The finding confirms the meditating effect of IHF (β = 0.027, p = 0.001) for the relationship between PBC and CHF. Further, IHF significantly mediates between all the factors (i.e., HTC, KHF, AHF, SBN, and PBC) and relationships with the CHF, whereby intention significantly enhances the relationship for the subject factors on the CHF.
The moderating effect of PBS was evaluated for the relationship between IHF and CHF. Study findings suggest that PBS significantly moderates the relationship between IHF and CHF. The perception of barriers reduces CHF. However, the moderating effect of PBS had a reduced effect on the relationship between the IHF and CHF. Moreover, high intention reduced the effect of PBS for CHF. However, PBS needs to be contained so as to increase the consumption behaviour for healthy foods [13]. Our study is pioneering in testing the moderating effect of PBS for the relationship between IHF and CHF and is therefore important to understand that consumers having high intention felt less about PBS than CHF and vice versa.
The multiple-group analysis estimated the effect of respondents' personal features of gender, residence area, and education. The PLS multi-group analysis technique investigated the effects of respondents' characteristics. Study results reveal no significant variance for respondents' gender on the study paths, and there is no significant difference between study paths based on gender. There is a significant difference between PBS and CHF for the respondents' living area-namely, urban and rural areas. There is a significant difference in the path between IHF and CHF. However, there is no significant difference for other paths and no significant difference based on respondents' area of living. Moreover, there is no significant difference for other paths of the study model based on the respondents' education. Multigroup analysis also revealed that the effect of AHF on IHD was significantly higher among the lower-income group compared to the higher income group. Moreover, the effect of PBC on IHF was much lower among the middle-income group than that of the other two groups.
Subsequently, this study estimated the performance of CHF with the factors of HTC, KHF, AHF, SBN, PBC, PBS, and IHF. The most critical three factors for the performance for CHF are AHF, KHF, and PBC. Besides, the fourth and fifth most important factors for the performance of CHF are HTC and SBN for the CHF.

Conclusions
It is important to have healthy nations, and the health of a nation depends on healthy food consumption by the youth of that nation [3]. The current study explored the effect of HTC and KHF that impact the IHF by factors of attitude, SUN, and PBC. The study also included the factor of PBS for healthy food in influencing CHF among Malaysian young adults.
Young people around the world have significant consumers at a global level [28]. The young Malaysian population is increasingly interested in having a healthy lifestyle and getting involved in healthy food consumption [3]. Healthy eating is increased with the personal pro-health behaviours, and it is affected by the PBS for healthy food products [4]. Global youth is encouraged to get engaged in pro-social and personal health-related consumption [29].
Study findings have several implications in developing effective strategies for healthy food consumption. The effects of HTC, KHF, AHF, SUN, and PBC positively influence IHF among Malaysian young adults. Attitude is the most significant contributor to the intention to consume healthy food. Marketers and government agencies must increase the information and promotion of healthy food [3]. It helps to enhance the level of information and knowledge of general consumers and also helps to promote healthy eating habits [12], as government intervention can ensure the reduced prices for healthy food. KHF is important for the intention to consume healthy food. CHF is significantly reduced by PBS. PBS needs to be controlled by the provision of healthy foods at superstores. Reduced prices, availability, and general consumer attitude toward healthy food can also aid in addressing the issue of obesity and empower the public to lead a healthy lifestyle [10]. The information and promotional activities need to be activated to enhance awareness and influence knowledge and consciousness of healthy food.
The study has the following three limitations. The study analysis was performed on the cross-sectional data that have limited generalisability. Future research should consider the longitudinal data to understand the time lag between IHF and CHF. However, the study model can be utilised to explore the consumption of organic food. PBS can be utilised to understand the restricting factors for CHF among study samples. PBS is higher among urban respondents than rural samples. Future studies can explore the factors to tackle the PBS in improving the CHF. This study contributes to the healthy food adoption model by adding the factor of PBS. Future research can evaluate the role of different barriers for IHF. The current study estimated that the general perception of healthy food consumption and knowledge of healthy food is inconsistent and requires further investigation [6]. This may be seen as a further limitation in generalising the findings of this study to a wider population. However, general knowledge of consumers regarding the influence of healthy food is a social and environmental concern [3]. In this regard, future studies could use specific knowledge of healthy food in establishing the intention and behaviour of consumers toward a vast range of healthy food products.