Determinants of Fast-Food Consumption in Romania: An Application of the Theory of Planned Behavior

This study explores drivers of fast-food consumption in Romania using the Theory of Planned Behavior. We analyze 532 responses to an online survey and use partial least squares path modeling to estimate the relationships between the intention to consume fast food and its possible determinants. Our results show that the most significant predictor is the subjective norms (injunctive norms: β = 0.218, p < 0.001; descriptive norms: β = 0.192, p < 0.001). Among the affective and cognitive attitudes, only the latter is statistically significant in predicting the intention (β = 0.088, p = 0.020), while perceived behavioral control is not significantly associated with intention toward fast-food consumption. We explain how our results can help policymakers to design better interventions on public health concerns about fast-food consumption and population obesity, especially children obesity.


Introduction
While children tend to consider eating at a fast-food restaurant a special event [1], obesity and other food-related diseases are on the rise [2,3]. In 2016, more than 340 million children and young adults between 5 and 19 years old were overweight or obese, and in 2020, a number of 39 million children under the age of 5 were diagnosed as obese or overweight [4]. Since 1975, global obesity has nearly tripled, and approximately 1.9 billion adults over the age of 18 are overweight, and over 650 million of them are obese [4]. According to the authors of [5], abdominal obesity was associated with fast-food consumption. Sandwich intake was linked to obesity and overweight in 35% of the cases, fried chicken in 40% of cases, and pizza in more than 80% of cases. With low nutritional properties [6], fast food includes burgers, hot chips/French fries, fried chicken, certain mass-produced pizzas [7], and also doughnuts [1], being currently served by 897.683 restaurants worldwide with a market size of USD 860 billion and more than 14 million [8].
There is significant scientific research on fast-food consumption among adolescents, driven by the high tendency of "people in their teens and twenties" to consume energydense food products [1]. The preferences for this type of food are more prevalent in lowand middle-income countries where they signal a "nutritional transition" from the country's traditional food to a Westernized diet [9], which includes excessive consumption of high-fat dietary products, processed meats, sweets and soft drinks [10]. These energy-dense foods are low in fiber and essential nutrients, with a high amount of salt, sugar, and refined The theory of planned behavior framework assumes that attitudes, subjective norms, and perceived behavioral control predict behavioral intention, which further determines the corresponding actual behavior [20]. The popularity of this theory is explained by its ability to explain a wide range of human behaviors, using this compact structure of attitudes, subjective norms, and individual control.
Attitude refers to a favorable or an unfavorable assessment of a certain outcome [21] related to the performance of a certain behavior (for example, eating fast food), and it can be divided into two categories [7,22]: affective attitude (the emotional component reflecting an individual's feelings about something) and cognitive attitude (the rational component reflecting an individual's knowledge or beliefs about something). Favorable attitudes support behavioral intention, while negative attitudes hinder it.
Two basic categories of subjective norms (SN) are discussed in the work of [23]: injunctive and descriptive. Injunctive norms refer to an individual's perception that society or a specific reference group will approve or disapprove their behavior, and descriptive norms highlight the social pressure an individual may experience comparing their behavior in a certain situation with what is generally recognized as normal conduct. In different words, injunctive norms refer to what we think that other people expect us to do, while descriptive norms refer to the perception of what other people actually do, this setting the norm with respect to a certain behavior [24]. This partition is also supported by the authors of [25], where subjective norms are split into a category that involves information (injunctive) and another category that involves peer pressure (descriptive). According to the TPB, subjective norms of both types are expected to support behavioral intention if they signal that the behavior is desirable. However, previous research shows that the mechanisms that underline these influence are different: injunctive norms acts through all five types of social influence discussed by the authors of [26], while descriptive norms drive the behavior through only two of them [23].
Perceived behavioral control captures people's self-reported control on performing a given behavior [27] through its two dimensions: perception of control, which focuses on factors from the outside of the individual (external control), such as others' behavior, accessibility, task difficulty, etc., and self-efficacy, which involves factors that come from the inside of the individual (internal control), such as motivation, ability, or personality [28]. The TPB framework assumes that both dimensions of the perceived behavioral control support behavioral intention in the case of positive behaviors and hinder the intention of undesirable behaviors.

Hypotheses Development
TPB was successfully applied to explain a large variety of behavioral intentions and actual behaviors, from recycling, travel, technology adoption, protection of privacy to smoking, drinking, the use of health services, breastfeeding, and drug use [27,29]. Healthy eating behaviors [20], genetically modified food consumption [30], organic food consumption [31], consumption and purchasing of halal food [32,33], sustainable food consumption [34,35] food neophobia and ethnic food consumption [36], functional food purchasing [37], and fast-food consumption [7,11,38,39] are food-related applications of the TPB.
According to the authors of [20], subjective norms and perceived behavioral control predict 30% of the variance in intention to adopt healthier eating behavior. In the application of TPB on genetically modified food consumption, the model had an explanatory power of 44.4%, with attitude being the strongest predictor [30], while in the study conducted on organic food consumption [31], subjective norms were the most significant predictor of the intention to consume. In the case of functional food consumption [37], both attitude and subjective norm had the strongest influence on intention. As regards the consumption and purchasing of halal food, [32] indicated that TPB was able to explain 29.1% of the variance in intention, and all the TPB's constructs were significant predictors of halal food consumption, while [33] argued that the only attitude has the highest influence on halal food consumption and the model has lower explanatory power, of only 24%. Other studies also found attitude as the strongest predictor [40] and positively related to behavioral intention [7]. All of the TPB's constructs have the power to explain on average 48% of intention to buy and consume sustainable food, according to the work of [34], although [35] suggest that social norms have the strongest ability to predict sustainable food consumption. As for the study conducted on food neophobia and ethnic food consumption [36], the TPB model was able to explain 42.1% of the variance in the intention to consume Dayak food, attitude having the strongest effect on intention compared to subjective norms and perceived behavioral control.
In an attempt to better understand how attitude influences individual behavior, previous studies [7,41,42] have debated the existence of two dimensions of the construct: affective and cognitive. The main findings of these studies indicated that measures of cognitive and affective attitudes load on distinct factors and are differently associated with intention. In the case of fast-food consumption, although mean scores for both cognitive and affective attitudes show a slightly negative attitude toward fast-food consumption [7], the findings revealed by the same study suggest that only cognitive attitude was notably significant in the prediction of the intention to consume fast-food. The context for the failure of affective attitude to predict fast-food consumption is explained by the authors of [43,44], who highlight the fact that an affective attitude is usually a predictor of healthy food consumption rather than a snack or sweet food consumption. In line with these considerations, the following research hypotheses are assumed to hold in the Romanian case too: Hypothesis 1a (H1a): Affective attitudes are positively related to behavioral intention to consume fast food.
Hypothesis 1b (H1b): Cognitive attitudes are positively related to behavioral intention to consume fast food.
Hypothesis 1c (H1c): Cognitive attitudes are stronger predictors of the behavioral intention to consume fast food than affective attitudes.
Subjective norms, especially those influenced by reference groups, such as friends, have been found to strongly affect the intention to consume fast food in Iran [45]. Both injunctive and descriptive norms were found as strong determinants of fast-food consumption in The Nederlands, although the effect sizes were small [46]. Another study conducted in Australia found that injunctive norms are predictors of intention to consume fast food [38], while other studies employing the TPB to examine dietary behaviors reveal that subjective norms are a stronger predictor of the intention in adolescents samples compared with adults samples situations [47]. Although the assumptions of the theoretical model, as well as previous research, suggest that both injunctive and descriptive norms positively support the behavioral intention to consume fast food, we expect that the two components of the subjective norms have different contributions. As the authors of [26] state, what others do and what others approve are different sources of human motivation. While descriptive norms refer to what is commonly done and indicate what is effective and adaptive, injunctive norms are moral drivers and promise social rewards. There is conflicting evidence regarding the contribution of each type of subjective norm in predicting the intention to perform a certain behavior. While some studies show that injunctive norms account for a larger variability in the behavioral intention and behavior than descriptive norms do [48,49], other contributions show the opposite [50]. As a consequence, we will assume that the descriptive and the injunctive norms are equally strong in predicting the behavioral intention to consume fast food and set the following research hypotheses: Hypothesis 2a (H2a): Injunctive norms are positively related to behavioral intention to consume fast food.
Hypothesis 2b (H2b): Descriptive norms are positively related to behavioral intention to consume fast food.

Hypothesis 2c (H2c):
The injunctive and the descriptive norms are equally strong predictors of the behavioral intention to consume fast food.
Reference [23] stated that a person's intention to act in a certain way is influenced by the perceived behavioral control, through both external control, represented by perceptions of control, and internal control, represented by self-efficacy. As for the influence exerted, perceived behavioral control has between moderate and strong influence on behavioral intention [47]. Furthermore, [1] shows that behavioral intention is strongly associated with perceived behavioral control, a result confirmed by the authors of [51] that shows that higher levels of perceived behavioral control are related to lower levels of intention to consume fast food. In terms of dimensions of the perceived behavioral control, the TPB framework assumes that both perceptions of control and self-efficacy are negatively associated with the intention to perform undesirable behaviors, as is the case with fast-food consumption. Previous findings identified these constructs as having, among all TPB determinants, the strongest influence on people's intention for healthy eating [52]. Both of them positively relate to the intention to reduce fast-food consumption, with the perception of control as the strongest predictor [53]. Other studies emphasize the importance of self-efficacy in improving adherence to healthy diets [54], which in turn suggests a negative association with the intention to consume unhealthy food. We, therefore, set the following hypotheses: Hypothesis 3a (H3a): Perception of control is negatively related to the intention to consume fast food.
Hypothesis 3b (H3b): Self-efficacy is negatively related to the intention to consume fast food.
Hypothesis 3c (H3c): Perception of control is a stronger predictor of the intention to consume fast food than self-efficacy.
Self-identification has been found as the strongest predictor of behavioral intention than other TPB constructs, especially in the domain of food choice and consumption [55]. The result is also supported by the authors of [7], suggesting that people who self-identify as healthy eaters have weaker intentions to consume fast food frequently.
Hypothesis 4 (H4): Self-identification as a healthy eater is negatively related to the behavioral intention to consume fast food.

Data
Data were collected between November 2020 and February 2021, using a self-reported questionnaire made on Google Forms and sent to individuals through different social platforms, such as Facebook, WhatsApp, and e-mail, via convenience and snowball sampling. The Ethical Committee of the University of Bucharest approved the research (decision no 146/07.07.2021). Before completing the questionnaire, the respondents were informed that their participation in this study is anonymous and entirely voluntary, and by filling in the survey, they provide consent to participation in the study. The minimum sample for a significance level of 0.05 and a power level of 0.990 was 410 if calculated using the inverse square root method and 392 if calculated using the gamma-exponential method.

Measurement
The aim of this study is to explore what are the determinants of fast-food consumption among Romanians and to what extent they influence behavioral intention. We use the theory of planned behavior and replicate the research of [7] on fast-food consumption in Australia, from which we translated the constructs. The reason for this approach comes from the fact that the original paper builds on a strong and previously tested theoretical background, enriched with control variables rooted in core principles of human psychology. Consequently, we expect that the same research design is appropriate regardless of the country. However, a direct comparison between Romania and Australia can bring more. Even if both countries have been going through a rapid "nutrition transition", Romania started to adopt a Western diet pattern almost thirty years later than Australia, after the fall of the communist regime [56,57]. If we find that, despite the time gap and the different geographical locations of the two countries, the drivers of fast-food consumption in Romania match the Australian ones, the extant literature can suggest research and inform practical interventions in Romania that align with those adopted worldwide.

Behavioral Intention to Consume Fast Food
The survey measured the intention to consume fast food, attitudes and subjective norms, and the respondents' perceived behavioral control over fast-food consumption. Intention to consume was measured using two 7-point Likert-scale items concerning the respondents' likelihood to buy or consume fast food over the next month. To assess the key predictors of behavioral intention, namely attitude, subjective norm, and perceived behavioral control, we divided every dimension in two, as the authors of [7] suggest, and as can be observed in Table 1. Table 1. Measurement items by latent constructs as provided by extant literature.

Self-efficacy
Items that capture self-efficacy: PBC3, PBC4 Consideration of Future Consequences [7,58] Items: CFC1-CFC14 Fear of Negative Evaluation [59] Items that capture how fearful are people about others' negative evaluation of their fast-food consumption: FNE1-FNE12 Self-identification as a healthy eater [7] How much a person self-identifies as a healthy eater: SI1, SI2, SI3, SI4 Control beliefs [7] Facilitating factors Items that capture factors that facilitate fast-food consumption: BS17-BS20 Impeding factors Items that capture factors that inhibit fast-food consumption: BS21-BS24

Attitudes
Attitudes were split into cognitive and affective. The cognitive attitudes were measured using the words "To me, eating fast food frequently is . . . " followed by a list of five sets of adjectives (e.g., harmful-beneficial, cheap-expensive, etc.), all rated on a 7-point scale. To measure affective attitudes, we used the following: "Eating fast food frequently makes me . . . " followed by twelve sets of adjectives (e.g., happy-unhappy, worried-calm, etc.) rated on a 7-point scale.

Subjective Norms (SN)
Subjective norms were also split into injunctive norms and descriptive norms. Injunctive norms refer to the individual's perceptions about what others expect regarding fast-food consumption (e.g., "Those close to me expect me to eat fast food regularly"), and were measured using 2 items presented on a 7-point scale (1-definitely false to 7-definitely true). In order to capture how respondents perceive their peer group's behavior (descriptive norms) regarding fast-food consumption (e.g., "Those who are close to me eat fast food regularly"), another 2 items rated on the same 7-point scale were added.

Perceived Behavioral Control (PBC)
In line with [7], perceived behavioral control was measured by using two subdimensions: perceptions of control and self-efficacy. Perceptions of control were assessed using two items (e.g., "I have complete control over the number of times I will eat fast food over the next month.") on a 7-point scale between 1 (definitely false) and 7 (definitely true). Self-efficacy was also measured using two items, both on a 7-point scale (e.g., "It would be impossible for me not to eat fast food regularly over the next month").

Control Variables
Apart from TPBs dimensions, and following the research design adopted by the authors of [7,38], we added as controls: Consideration of Future Consequences Scale (CFC) [58,60], Fear of Negative Evaluation Scale (FNE) [59], the extent to which the respondents identify themselves as healthy eaters, facilitating factors and impeding factors. A detailed review of the theoretical background that underlines these influences is presented in the work of [7,38,61].
According to the authors of [7], facilitating factors, such as not cooking at home, eating alone, nearby fast-food restaurants, etc., influence people to consume fast food, while the impeding factors, such as concerns about weight, health, or fast-food cost, should inhibit to some extent the appetite for fast-food products.
The relevance of CFC in foregoing eating behaviors was proven in extant research [58,62], although in the concrete case of fast-food eating, the concern for the future negative effects of the dense food can be offset by the short-term benefits experienced by the eater [7,38,63]. The fear of negative evaluation, defined as the extent to which an individual fears disregard from others [59], was already identified as an inhibitor of food consumption [64,65]. Although originally accounted for as an important predictor of eating disorders [64,66], other authors took it into consideration as a control variable in other food-related contexts, including fastfood consumption [7,61]. To the CFC Scale's original items, another two items were added for an appropriate investigation of CFC in connection with diet (i.e., "I often avoid certain foods because I am concerned about my health." And "I usually choose food because it is convenient or tasty rather than because it is good for my health.") We used the original FNF scale comprising 12 items (examples are "I am unconcerned even if I know people are forming an unfavorable impression of me" or "I am usually worried about what kind of impression I make".) Furthermore, apart from CFC Scale and FNE Scale, we used as control variables facilitating and impeding factors of fast-food consumption previously accounted for by the authors of [67][68][69]. We operated with the 7 items out of 24 found as statistically significant in previous research, 4 items to measure facilitating factors (e.g., "I am more likely to eat fast food if I have a little spare time."), and 3 items to measure impeding factors (e.g., "I feel guilty if I eat fast food.") [7].
Finally, using a 4-item scale, we measured how respondents self-identified as healthy eaters by asking questions such as "I think of myself as someone who is concerned with the health consequences of what I eat" and "I think of myself as a healthy eater".

Method
Supported by the lack of normality of our data, we use a partial least squares algorithm [70] that provides information about how much variance in the result can be explained by the TPB predictors along with the control variables. The algorithm that underlines partial least squares estimation is iterative, composed of two parts: a measurement (or outer) model that provides scores of the latent constructs, and a structural (inner) model, that assesses the relationships among variables. To conduct our analysis, we used WarpPLS software, version 7.0.

Results
Our sample consists of 532 respondents (370 women and 162 men), with an average age of 22 years. A total of 83% of the respondents have ages between 18 and 24 years old, this being the predominant age group in our sample. A total of 67% of the respondents have their provenience in urban areas, and 89.3% of them declared they are students. This can explain why more than half of them have an income under 1400 lei (minimum net wage in Romania). A complete sample description is available in Table 2.

The Measurement (Outer) Model
After a preliminary investigation, the Consideration of Future Consequences Scale was excluded since only one item has a loading over 0.7 [71]. Out of all latent variables that measure the TPB dimensions, we have to drop three items from affective attitude, two items from cognitive attitude, and both items that measured self-efficacy since the items' loading were below the 0.7 threshold. The final structure of the latent constructs is presented in Appendix A. Table 3 presents the reliability of measurement for each remained construct. The composite reliability values range between 0.769 and 0.947 and are above the recommended threshold of 0.70 [72]. The Cronbach's alpha values are higher than 0.70, indicating a suitable internal consistency [72] with two exceptions: cognitive attitude, with a Cronbach's alpha of 0.617, and perceptions of control with 0.634. There are recommendations stating that if the items' number is small, the acceptable Cronbach's alpha value can be above the threshold of 0.5; we keep the latent constructs in the analysis. Furthermore, Table 3 also shows that the average variance extracted (AVE) for each variable exceeds 0.50, except for cognitive attitudes, whose value is 0.421. However, according to the work of [73], this AVE is still adequate since the composite reliability coefficient associated with this construct is higher than 0.60 (its actual value is 0.769). Table A2 (Appendix B) shows that the discriminant validity of the measurement holds. The diagonal values in this table are higher in all cases than the corresponding off-diagonal values. Moreover, none of the off-diagonal correlations is greater than the recommended value of 0.8 [74]. Table A3 (Appendix B) presents the combined loadings and cross-loading of all indicator items of the latent construct. The loadings of all reflective items range from a lower bound of 0.716 to an upper bound of 0.947, thus higher than the required threshold 0.7, and all off-diagonal values are lower than the diagonal value for each block of measures items. Cognitive and affective attitudes were derived using formative measurement since each of them has some contribution in explaining the attitude of the respondent regarding fast-food consumption. We, therefore, decide that convergent validity holds. Table 4 presents the estimated coefficients of the model along with the corresponding effect sizes. The amount of variance explained (R 2 ) for the endogenous construct, behavioral intention is 0.39. We checked for multicollinearity and found that all VIF values are lower than 2.1, and the average block VIF (AVIF) is 1.441, which is below the recommended threshold of 5. In terms of overall model fit, the standardized root means square residual (SRMR) is 0.08, below the acceptable threshold recommended by the authors of [75].

The TPB Dimensions
Cognitive (β = 0.088, p = 0.020) and affective (β = 0.041, p = 0.171) attitudes show positive association with the intention to consume fast-food products. However, only cognitive attitudes are statistically significant; therefore, H1a is rejected, and H1b is accepted. As for the predictive power of affective and cognitive attitudes, the latter has a higher effect size (0.088) than the former (0.015), showing that the cognitive attitude exerts a stronger influence on the behavioral intention to consume fast food than the affective attitude. Both injunctive (β = 0.218, p < 0.001) and descriptive (β = 0.192, p < 0.001) norms are predictors of behavioral intention and show a positive influence on fast-food consumption, a result that supports both H2a and H2b. However, injunctive norms have a higher effect size (0.102) than descriptive norms (0.081), showing that the former are stronger predictors than the latter. From the two dimensions of perceived behavioral control, self-efficacy was excluded since it does not have factor loadings above 0.7. Perception of control shows a negative correlation with behavioral intention; however, the value is not statistically significant (β = −0.024, p = 0.288). Therefore, H3a was rejected, while we were not able to test H3b due to the fact that self-efficacy had to be removed from the analysis.

The Control Variables
Facilitating factors have a significant positive association with behavioral intention (β = 0.19, p < 0.001), while impeding factors are negatively and significantly correlated with intention (β = −0.095, p < 0.014). Therefore, both H4a and b are supported. Other two factors that exert an influence on behavioral intention are self-identification as a healthy eater, which has a significant negative relationship with intention to consume fast food (β = −0.112, p < 0.005), and sex, which exerts a positive influence on intention toward fast-food consumption (β = 0.074, p < 0.042). Though fear of negative evaluation scale (β = −0.032, p < 0.228) and age indicate a negative influence on intention, the values are not statistically significant in predicting one's behavioral intention, so they are not taken into consideration (β = −0.012, p < 0.390). We conclude that H5 supported. Table 5 summarizes our findings in terms of accepted and rejected research hypotheses.

Hypothesis Supported/Rejected
H1a Affective attitudes are positively related to the intention to consume fast food Rejected H1b Cognitive attitudes are positively related to the intention to consume fast food Supported

H1c
Cognitive attitudes are stronger predictors of the behavioral intention to consume fast food than affective attitudes Supported

H2a
Injunctive norms are positively related to the intention to consume fast food Supported H2b Descriptive norms are positively related to the intention to consume fast food Supported H2c Injunctive norms are stronger predictors of behavioral intention than descriptive norms Rejected H3a Perception of control is negatively related to the intention to consume fast food Rejected H3b Self-efficacy is negatively related to the intention to consume fast food Untestable

H3c
Perception of control is a stronger predictor of the intention to consume fast food compared with self-efficacy Untestable H4 Self-identification as a healthy eater is negatively related to the intention to consume fast food Supported

Effect Sizes
In terms of effect size, injunctive norms have the highest contribution to the intention to consume fast food (0.102), thus indicating that interventions on injunctive norms may have a meaningful effect since they are strong enough to be considered relevant from a practical point of view [76]. Still relevant, although with smaller effects, are descriptive norms, facilitating factors (0.080, respectively 0.081), cognitive attitude (0.032), self-identification as a healthy eater (0.039), and with a very small effect, impeding factors (0.025). Sex effect is below the minimum threshold of 0.02, thus too weak to be practically relevant, even though the relationship with intention is statistically significant [77].

Discussion
We explored determinants of the intention to consume fast food among Romanians, using the theory of planned behavior as the main theoretical background. Facilitating and impeding factors, fear of negative evaluation, and self-identification as a healthy eater were used as additional variables. Our model explains 39% of the variations in behavioral intention, less than the 42% identified in the work of [7], or the 67% identified by the authors of [45], and overall less than the average explanatory power of 44.3% found across different studies by the authors of [47]. However, our study explains more than the 25.7% found in the work of [51] and the 34.7% identified by the authors of [78].
We found that affective attitudes are not significantly related to the intention to consume fast food, while cognitive attitudes positively relate to the intention. Our result clarifies the conflicting evidence provided by extant literature. Studies conducted on Korean [1] and Iranian [79] high school students found that out of the three TPB dimensions, attitude was not a significant predictor for behavioral intention, as opposed to another study conducted on an Iranian sample of high school students [51] where attitude was found as the strongest predictor of the intention to consume fast food. All these studies measures attitude as one single construct. Our study takes the path suggested by the authors of [7] and splits the construct into cognitive attitudes and affective attitudes. We confirm the results of [7] and [44] by showing that only cognitive attitudes predict the intention to consume fast food. The authors of [43] suggest a possible explanation of this result in the sense that they found that affective attitudes significantly predict healthy food consumption, but not energy-dense food consumption.
Furthermore, we found that subjective norms have a positive and significant effect on intention, as other previous studies confirmed [1,51,79]. Nevertheless, subjective norms in our case were split into descriptive and injunctive norms, and both were found significant in predicting intention, unlike the results of [7], where only injunctive norms were significant. Our result shows that the respondents' intention to consume fast food is affected by how their reference group (i.e., family, friends, colleagues, etc.) think and behave regarding fastfood consumption. This aligns with findings from a study on the psychobiology of appetite: "when we eat with others, we have a natural tendency to use their behavior as a guide" [80]. The results of [1] indicate that family, teachers, and friends are the most influential people among young adults, although friends have a more significant influence on respondents' decision to consume fast food. The authors of [79] show that parents, especially mothers, are also having a strong influence on children's decisions to consume a certain type of food. Considering that previous research shows that health experts' opinions on fast food were not significant predictors of the intention to consume [7], our findings are highly important for public policymakers, especially those who are designing nutrition education programs.
As for the influence of perceived behavioral control on the intention toward fastfood consumption, the results indicate that neither of the two constructs that measured this dimension, namely self-efficacy and perceptions of control, predict the intention to consume fast food. These findings are partially in line with those revealed by the original study [7], which shows that respondents' perception of control over individual fast-food consumption was not statistically significant in predicting intention to consume fast food. On the other side, the same study found that self-efficacy does have a significant negative relation with behavioral intention (β = −0.27, p-value < 0.001), resulting in an inhibition to consume energy-dense food products. Considering that self-efficacy was dropped in our research, we cannot either concur or diverge from this finding.
We found that facilitating factors and impeding factors play an important role in predicting the intention to consume fast food. The intention is higher for those who cannot cook, have more spare time, eat alone, or have cravings. On another side, weight and health status concerns, feelings of guilt, and the price paid for buying fast-food products hinder fast-food consumption. Self-identification as a healthy eater also plays a significant role in inhibiting fast-food consumption and a higher predisposition for a healthier meal or a low-fat diet, a finding also supported by the authors of [7,67].
In terms of effect size, the most significant predictor of the extended TPB model regarding fast-food consumption in our Romanian sample is the injunctive norms. This result aligns with [51], which found that subjective norms are the strongest predictor in the extended TPB model, unlike in the basic TPB model where the attitude was the most important predictor is behavioral intention. In our Romanian sample, this result may be explained by the tendency of the Romanian psychosocial and cultural model to pay excessive attention to what others think about them, especially close people [81]. As an additional factor, global society exerts powerful and pervasive influences on dietary and eating habits, many studies showing that social norms are important players in the development and maintenance of obesity around the world [82].
Considering that subjective norms were the predictor with the highest contribution in terms of effect size and therefore the most suitable for practical interventions, a recommended step for behavioral public policymakers, working on reducing fast-food consumption in Romania, should be the design and implementation of nutrition education programs aiming to change societal norms around food choices. Given the obvious influence of the subjective norms, it would be helpful to highlight, through different marketing platforms, the behavior of specific persons with whom people can identify themselves in order to guide the societal change in a suitable direction [83]. In this sense, celebrities or individuals that have an influence on a considerable number of people (e.g., social media influencers, teachers, etc.) and who adopted a healthier lifestyle or avoided fast-food consumption can be involved to spread the awareness of negative fast-food implications on health. In universities and student campuses, especially in diner rooms, it would be effective to show up messages encouraging people to make healthier food choices, using framing effects (e.g., loss aversion, gain seeking, etc.) Some messages could also include statistics about their colleagues' healthy food choices, nudging them to compare their food behavior with their colleagues' behavior, and, consequently, to make a change.
According to our findings, young adults between 18 and 24 years old are the group that consumes the most fast-food products, the majority of them have an income under minimum net wage, and they indicated that price and convenience are some of the factors that influence their decision to consume fast-food. A practical incentive to support better decisions regarding food consumption is to encourage these people to choose healthier menus.
In this sense, a decrease in price for healthier food products would have a positive effect on the intention to buy healthy food products. The advantages and benefits of this incentive are also supported by the authors of [84], who debated in a scientific paper the positive or negative effects of different interventions on promoting healthier ready-to-eat meals. Other intervention proposals regarding healthier alternatives for fast-food products encounter difficulties in collaboration with food outlet/chain manager [85]. In this case, the intervention approach should focus on customers or suppliers, or, in the extreme case, the government could apply taxes for unhealthy food products, such as carbonated drinks and saturated fats, and to subsidize food items containing fruits and vegetables in a significant proportion. This might have an impact on individual dietary habits, leading to an improvement in the overall health status.
The research on fast-food consumption in Romania is almost inexistent, and therefore our results cannot be judged against local theoretical backgrounds or previous findings. Although the subject deserves attention in all age groups, the majority of our respondents are students aged between 18 and 35 reached through convenience combined with snowball sampling. A study on a sample representative may deepen the understanding of what drives fast-food consumption in this age group.
Furthermore, a significant number of scientific papers examine fast-food consumption among teenagers, but there is less known on fast-food consumption among older age groups. Further studies need to address the gap, not only in Romania. Concerning this aspect, one neglected issue is the adult consumer's perception of the definition of fast food [86], since many adults tend to perceive as fast food only the items served by the wellknown Western fast-food chains [87]. However, the fast-food items (e.g., burgers, French fries, fried chicken, pizza, etc.) can also be made at home or served by regular restaurants, among other food items. Consequently, a further study should focus on fast-food products that older people eat, but not necessary at a fast-food restaurant.
Future research should relate eating habits and fast-food consumption with objective, clinical data, able to prove not only the relationships between this type of products and a number of noncommunicable diseases but also to measure the magnitude of the relation, allowing for assessing the concrete impact of reducing fast-food consumption on health.

Conclusions
The patterns of fast-food consumption in Romania are an unexplored niche, and so are their implications for public health and the economy. This study is the first step toward a better understanding of what drives the Romanian population in their choice for dense food consumption. We not only provide an understanding of the Romanians' perceptions, motivations to consume, and factors that support fast-food consumption, but we also compare findings obtained in a middle-income country with results obtained in developed countries. Informed Consent Statement: Before completing the questionnaire, the respondents were informed that their participation in this study is anonymous and entirely voluntary, and by filling in the survey, they provide consent to participation in the study.

Data Availability Statement: Data is available on request.
Acknowledgments: The authors thank the University of Bucharest for supporting the APC.

Conflicts of Interest:
The authors declare no conflict of interest. Table A1. A review of the measurement items.

Dimension as Included in [7] Item Abbreviation Item
Intention to consume INT 1 On a scale of 1 to 7, where 1 means "strongly disagree" and 7 means "strongly agree", given your lifestyle and/or taste preferences, how much it is likely that you will eat fast food regularly over the next four weeks?

INT2
On a scale of 1 to 7, where 1 means "definitely false" and 7 means "definitely true", how much you are likely to eat fast food regularly over the next month?