SIGNIFICANT OTHER’S ANTICIPATED EMOTIONS INCREASE STUDENT’S INTENTION TO AVOID SMOKING

The research started with a premise that individuals could take the initiative and regulate their behavior to generate Significant Others’ Anticipated Emotions (SOAEs). Could the SOAEs function as a social element of behavior in addition to subjective norms (SN)? The research aimed to answer this question. Therefore, the researcher extended the Theory of Planned Behavior (TPB) and argued that the SOAEs influenced behavioral intention through attitude toward behavior based on cognitive balance theory. In the smoking abstinence behavioral context, the researcher tested the extended model. The data from 242 respondents chosen conveniently, analyzed using structural equation modeling, revealed that Significant Others’ Anticipated Joyfulness (SOAJ) for smoking abstinence behavior and Significant Others’ Anticipated Distress (SOAD) for smoking behavior positively influences anti-smoking behavior through attitude. Moreover, the sole influence of the SOAJ and cumulative influence of SOAJ and SOAD on smoking abstinence intention are higher than that of the SN. As a new component of TPB, the SOAEs complement and do not rival the SN. Other researchers can utilize a longitudinal research design and test the extended model in different contexts of behaviors.


INTRODUCTION
Subjective Norms (SN) represent social influence on one's behavior (Ajzen, 1991(Ajzen, , 2020Fang et al., 2017). This construct consists of two components, i.e., the belief that significant others in one's life (such as family members, friends, seniors, teachers) want an individual to perform or not to perform a behavior, and the individual's motivation to comply with that desire (Ajzen, 1991). In this approach, it is clear that the starting point is significant others' initiatives, and the endpoint is an individual's behavior. Can it be seen the individuals as the initiator that intentionally regulate their behavior? More specifically, can individuals regulate their behavior regarding a concern for significant others' emotional well-being? Ajzen's (1991) Theory of Planned Behavior (TPB), the leading theory of behavior, including its extended models, has not covered this perspective. The researcher sees that Significant Others' Anticipated Emotions (SOAEs) can fill this gap.
Significant others are prominent persons in individuals' life, such as family or friends (Ajzen, 1991(Ajzen, , 2020, teachers, parents, managers, nuclear family, and extended family. Individuals can regulate their behavior to make significant others feel good or avoid feeling bad (Simamora, 2021). As Mayer and Salovey's emotional intelligence has stated that individuals have the emotional intelligence to predict significant others' emotions (Mayer, Carosu, & Salovey, 2016), then manage their behavior to generate or avoid particular SOAE.
Previous studies have found that the TPB determinants could not fully explain a behavior. There are around substantial part of intention formation that is uncovered by the three variables alone, i.e., attitude, subjective norms, and perceived behavioral control (Kan & Fabrigar, 2017;Rivis & Sheeran, 2003;Sun, 2019;Vamvaka et al., 2020;Walsh et al., 2018). To anticipate this, Tommasetti et al. (2018) have said that it might be useful to use additional variables in the model. The research answers this call by adding the SOAEs as the new determinant of behavioral intention.
Following Ajzen's (2020) call for studying a specific behavior, the researcher chooses smoking abstinence as the behavior under investigation concerning the following consideration. As the country with the third-largest tobacco user (Rukmi, 2019), it is no surprise to find the high smoking prevalence among youth in Indonesia (Fithria et al., 2021). However, there are large numbers of students who remained untouched by smoking. The controlling roles of parents, peers, and health workers for that behavior are significant for that behavior but weak (Suharyanta, Widiyaningsih, & Sugiono, 2018). Consequently, it believes that the student's decision to stay away from smoking mostly relied on their considerations.
In this consideration, an individual's relationship with significant others should be a critical factor. Schoebi and Randall (2015) have specified that emotions play a critical role in relationships. It can create, maintain, increase, or weaken interpersonal relationships. Shoham et al. (2007) have underlined that smoking behavior can induce emotions that strengthen or weaken the interpersonal relationship between close people. Interpersonal relationships can also cause smoking behavior (Omasu, Uemura, & Yukizane, 2015). Therefore, the preference for non-smoking acted by a large part of Indonesia's students, as mentioned, can occur following this way. Specifically, based on Ruvalcaba-Romero et al. (2017), the willingness to maintain or increase others' emotional well-being, which is beneficial for interpersonal relationships between individuals and significant others, can be the reason for the students to practice non-smoking behavior.
The individual can anticipate the emotional impact of their behavior on others (Mayer et al., 2016). Therefore, it is interesting to clarify whether students anticipate significant others' emotions that motivate them to practice anti-smoking behavior. To answer that curiosity, the research problem is, how do significant others' anticipated emotions influence students' intention to avoid smoking?
The research extends the Theory of Planned Behavior (TPB) to answer its problem. With the extended model, the research questions are how the SOAE influences anti-smoking behavioral intention? How is the comparison of the SOAEs and SN efficacies in influencing that behavior? As social components of behavior, do SOAEs complement or rival the SN?
The research aims to answer these questions and propose SOAEs as an additional component of social influence. So far, this concept is still new to behavioral science. Therefore, the concept and its operationalization can be viewed as the original contributions of this research. This concept may also motivate positive behaviors or demotivate negative behaviors by reminding them about their significant others' anticipated emotions from a practical perspective.
Social influence on behavior is inevitable, especially in a collective society commonly found in Asian countries. There are many concepts and myriad researches about this issue. However, the researcher limits the discussion on subjective norms and anticipated emotions as two notable behavior elements.
In Ajzen (1991), the SN is defined as the person's perception that most people who are important to him/her should or should not perform the behavior, representing the social factor of behavior. Mathematically, the SN is formulated as: (1) In this equation, the NBi states the normative belief (i.e., the person's belief that prominent persons in their lives or individuals think they should or should not perform a behavior). The MCi represents the motivation to comply with the referent, and 'n' is the number of referents. Although useful theoretically, this mathematical model is rarely used in empirical studies. Scientists usually only use the NBi to represent the SN.
The SN's activation as the representation of social influence can be traced Theory of Reasoned Action (TRA) model (Ajzen, 1991). However, the more advanced model of behavior, especially the Theory of Planned Behavior (TPB) and Model-of-Goal Directed Behavior (MGB) and the most recent researches still rely on that component (Esposito et al., 2016;Ham, Jeger, & Frajman, 2015;Holevova, 2018;Passafaro, Livi, & Kosic, 2019). In the theory of TRA, behavioral Intention (BI) is figured out as the result of attitude (Ab) and subjective norms (SN). The TPB adds perceived behavioral control (PBC) as the determinant of BI. The most notably additional component in the MGB to those three components is anticipated emotions (AE).
The predictive ability of subjective norms is limited (Kan & Fabrigar, 2017;Nystrand & Olsen, 2020;Walsh et al., 2018). It is not clear in what population types or contexts the social norms work effectively (Silva & John, 2017). Some researchers (Fang et al., 2017) see that the SN's weak operationalization causes this limitation. Others (Nystrand & Olsen, 2020;Passafaro, Livi, & Kosic, 2019;Sun, 2019) view that the SN is general in nature and should be adapted to cover the contexts' specificity. There is also the question of whether the SN covers the full extent of social influence on behavior. In responding to these questions, Niemiec et al. (2019) and Najafi et al. (2017) have reminded that the SN's relative impact should be varied across behaviors and populations. In the next section, the researcher describes some variations of social influence activation in explaining behavior.
Subjective norms in the traditional TPB do not consider the potential influence of other normative belief categories, such as moral norms (Fang et al., 2017;Sun, 2019). Moral norms deal with a classification of actions as permissible or impermissible without specifying which outcomes are more or less desirable. It can be understood as a principle or rule that specifies actions that are required, permissible, or forbidden independently of any legal or social institution (Fang et al., 2017;Liu, Liu, & Mo, 2020). Moral identity appears to be a useful additional component in predicting intention to engage over and above the basic TPB model (Sun, 2019). The involvement of moral norms in the traditional TPB shows mixed results. Moral norms' contribution to predicting ethically value-laden behaviors is low. There is no contribution of moral norms in explaining behavioral intention. Rivis and Sheeran (2003) have concluded from 46 articles that moral norms' contribution is only 3% in their metaanalysis. However, Sun (2019) has revealed that moral norms play an essential role in morally appropriate behaviors. The essential role of moral norms is also apparent in more recent value-laden behavior studies, such as recycling, pro-environment behaviors, viewed as part of ethical behavior.
There are various understandings of norms, but all of them have the same focus, i.e., the regulation of proper and acceptable behavior. Merriam-Webster Dictionary (n.d.) has stated that the norm is a principle of right action binding upon the group members and serving to guide, control, or regulate proper and acceptable behavior.
Yahdanmehr and Wang (2016) have suggested that the operationalization of norms in a study should consider the conditions under which norms are taken to guide action. Conditions-related norms are called social norms. They say that there are two required conditions to make social norms function in a given population. First, there must be a sufficient number of individuals who recognize the existent of the norms. Second, there must be an adequate number of followers or individuals who have a conditional preference to comply with the norms if their expectations are satisfied. The expectations, as Yahdanmehr and Wang (2016) have described further, come into two categories. Normative expectations refer to what an individual thinks others expect from him or herself. Empirical expectations refer to what an individual has observed or knows about others' behavior in similar situations. Normative expectations are the same as injunctive norms and empirical expectations with descriptive norms. The SN is an injunctive social norm because it deals with an individual's belief about perceived social pressure on individuals to perform or not to perform a behavior (Rivis & Sheeran, 2003).
Social norms owned by people who live in the same living space are local norms. In such a context, traditional subjective norms might fail to regulate behaviors that have collective and spatially defined implications, such as household waste. The higher the residence proximity, the stronger the local norms (Passafaro, Livi, & Kosic, 2019).
In short, the endeavors to fully understand the social influence in the TPB have not arrived yet at its final destination. This concept is still in its growth stage. To support the growth, the author offers significant others anticipated emotions (SOAEs) as another member of the social influence component of behavior. In the theory of Future-Oriented Thinking (FOT), Koh and Leung (2019) have stated that individuals could consider the future and the possible consequences of their actions. Luo et al. (2018) have emphasized that anticipation is related to the concept of consequence; instead, the concept of prediction. Prediction usually relates to outcomes, and satisfaction is usually tied with anticipation. Both of them (i.e., prediction and anticipation) are part of the expectation. As a cognitive skill, the FOT requires a self-ability assessment of predicting possible consequences of behavior for the self in the future, including future emotions (Bagozzi et al., 2016;Koh & Leung, 2019). The learning system works in anticipation. The more experienced the individuals, the more accurate their anticipation is (Luo et al., 2018).
People can anticipate their own emotions (Bagozzi et al., 2016;Israelashvili, Sauter, & Fischer, 2019;Mayer, Carosu, & Salovey, 2016) and others' emotions (Simamora, 2016(Simamora, , 2021. More specifically, Mayer, Carosu, and Salovey (2016) have stated that the emotional intelligence concept views people as capable of monitoring their and others' feelings and emotions and adapting their thinking and actions to avoid unexpected and create expected emotions.
Besides doing or not doing a behavior, emotions can also be anticipated due to success or failure in achieving goals in the future (Simamora, 2016(Simamora, , 2021. Such emotions are called anticipated emotions (Kotabe, Righetti, & Hofmann, 2019). If experienced by significant others, they are called significant others' anticipated emotions or SOAEs.
Significant others are people who have a good direct relationship with or expect good things to happen to an individual, such as family, friends, teachers. An individual can put anticipated emotions or the SOAEs as goals, which can stimulate current effort to achieve them in return. Significant others are viewed as part of proponents described by Simamora (2016Simamora ( , 2021 as any parties with whom an individual has or has not a social relationship that expected the good fortunes happened on an individual. The effort to meet goals can be an obligation to prominent persons (such as family); hence, individuals' goal achievement is beneficial to themselves and significant others around them, especially if that achievement is a gift to significant others (Simamora, 2016(Simamora, , 2021. That is why, as Curelaru et al. (2020) have stated, people may pursue personal goals (i.e., goals constructed by individuals' autonomous interests) and family goals (i.e., goals related to the family well-being). Consequently, people can regulate their behavior to avoid failure or achieve success concerning significant others' anticipated emotions. Another example, having a successful business without sacrificing study in college should be responded to with positive emotions because it is a positive thing to do. On the other hand, not having a successful business during their study should not be followed by Significant Others' Anticipated Distress (SOAD). It is not an ordinary thing for students.
This effort's mechanism has the same tone as promotion and prevention focuses on Higgins's selfregulatory focus theory (Higgins, 2018;Koopmann et al., 2019). A behavioral regulation to generate significant others' joyfulness can be viewed as a promotion focus. Behavioral regulation to avoid significant others' distress has the same spirit as a prevention focus.
The researcher believes that SOAEs' work is in line with the above schemas. To prove that belief, the researcher adopts the TPB in accommodating the role of SOAEs to predict behavioral intention.
It is a robust model and has successfully proved its efficacy in abundant studies (Ajzen, 1991(Ajzen, , 2020 and is significantly better than its predecessor (TRA) in predicting behavior (Ajzen, 1991). The TPB asserts that behavior can be predicted through behavioral intention. Attitude toward behavior (Ab), Subjective Norms (SN), and behavioral control create behavioral (PBC) (Ajzen, 1991) are determinants of the intention.
The research uses behavioral intention as the dependent variable. It means that the TPB is not wholly but partly utilized. The SN and SOAEs are used simultaneously, besides Ab and PBC, as determinants of behavioral intention. This approach enables the researcher to verify whether SOAEs are a companion of SN and the stronger one among the two to predict behavioral intention.
Several studies confirm the efficacy of anticipated emotions (AE) to explain behavioral intention. Londono, Davies, & Elms (2017) have incorporated anticipated negative emotions (regret) into the TPB to predict consumers' acceptance of and engagement in ecological behavior. Pelsmaeker et al. (2017) have revealed that anticipated positive emotions influence the intention to consume chocolate. The involvement of anticipated emotions increases the efficacy of the TPB to predict behavioral intention and actual behavior.
In the model (Figure 1), anticipated emotions occupy a position as the Ab's antecedent. The attitude toward a behavior is constructed by the belief that behavior has particular outputs (bi) and the evaluation toward that output (ei). Mathematically, the equation of attitude toward behavior (Ab) is Ab=Ʃbi*ei. To operationalize this construct, it is needed to make sure that the outcomes of behavior should be defined (Ajzen, 1991(Ajzen, , 2020 to enable the measure of belief (bi) and evaluation (ei).

Ab=Ʃbi*ei
(2) Perceived Behavioral Control (PBC) is defined as the perception of the easiness or difficulties to conduct behavior caused by the existence or the absence of required resources and opportunities. This construct is composed of Control Beliefs (CB) and Perception of Facilitation (PF). The CB is the subject assessment about the existence or the absence of resources needed to execute a behavior. The PF is the role of that resources in achieving behavior. The mathematical expression of the PBC is PBC= CBi*PFi (Ajzen, 1991).

PBC=CBi*PFi
(3) Figure 1 Conceptual Model (Source: Ajzen, 1991) Notes: SOAJ=Significant others' anticipated emotions for non-smoking behavior, SOAD=Significant others' anticipated distress for smoking behavior, Ab=Attitude toward non-smoking behavior, SN=subjective norms, PBC=perceived behavioral control, BI=Intention to continue non-smoking behavior The research, as exhibited in Figure 1, conceptualizes the SOAEs as determinants of Ab. Two considerations color this conceptualization. The first is Heider's balance theory (Belaza et al., 2017;Munroe, 2019), which asserts that people tend to maintain cognitive and emotional balance, especially with whom they have a close relationship. Cognitive and emotional congruence between the individuals with others will strengthen the balance. When cognitive and emotional incongruence occurs, Reid, Davis, and Green (2019) and Hu et al. (2019) have said individuals would be involved in attitude alignment. Through that practice, individuals will change their or others' opinions to achieve greater attitudinal congruence. This change is required because they said that individuals would experience psychological discomfort when they discover that their attitudes are inconsistent with those who have a close relationship with them. Attitude change is the only way to increase attitude congruence and reduce discomfort.
Second, Ajzen and Fishbein (1980) warned long ago that people should distinguish the SN's social influence as part of the Ab. They say that when someone says, "I believe that my children think I should buy chocolate," it deals with normative belief.
However, when he/she says, "Buying chocolate will please my children," then he/she talks about behavior outcomes. They assert that a belief for particular behavior outcomes is part of an attitude.
According to Solomon (2018), people's attitudes are based on what they think, feel, and act. It means that others' feelings reflected in the statement, "Buying chocolate will please my children," is not part of his/her attitude, but others' attitudes. However, as stated in Heider's balance theory (Belaza et al., 2017;Munroe, 2019), others' attitudes (emotions) will influence individuals' attitudes. More specifically, if an individual anticipates that significant others will like (dislike) him/her to perform (not perform) a behavior, he/she will be more (less) likely to perform (not perform) that behavior. The point is that significant others' anticipated distress for doing behavior influences an individual's attitude for performing that action negatively.
Based on those arguments, the researcher formalizes the following hypothesis. First, significant others' anticipated joyfulness (SOAJ) for performing (not performing) a behavior influences the attitude of performing (not performing) the behavior positively. In the smoking abstinence behavioral context, the higher the significant others' anticipated joyfulness (SOAJ), the higher the individuals' attitude toward smoking abstinence behavior. Second, significant others' anticipated distress (SOAD) for performing (not performing) a behavior influences the attitude of not performing (performing) the behavior positively. Concerning the smoking abstinence behavior, the higher the significant others' anticipated distress (SOAD), the higher the individuals' attitude toward smoking abstinence behavior.
Hundreds of TPB-related studies have confirmed the influence of Ab, SN, and PBC on behavioral intention. Therefore, it dares to say that attitude toward behavior, subjective norms, and perceived behavioral control positively influences behavioral intention. In smoking abstinence behavior, the third hypothesis states that attitude, subjective norms, and perceived behavioral control positively influence behavioral intention.

METHODS
The behavior's context should be specified in the TPB-related studies (Ajzen, 1991(Ajzen, , 2013(Ajzen, , 2020. For that reason, the researcher intentionally chooses students who practice smoking abstinence behavior. Omasu, Uemura, and Yukizane (2015) have found that non-smoking behavior is more apparent when individuals have a good relationship with their family and vice versa. The researcher believes that the same phenomena can be found in Indonesia.
Metaj-Macula (2017) has indicated that emotional intelligence works within a social context marked by a positive association with social interaction. Therefore, the higher tendency to practice non-smoking behavior among youths who have a good relationship with their family (Omasu, Uemura, & Yukizane, 2015) indicates the more intensive emotional intelligence work (Metaj-Macula, 2017). In other words, Ryan and Deci (2000) have said that the significant others' expectation has been internalized and become the individuals' intrinsic motivation in that context. With such a good relationship, according to Simamora (2021), the individuals are motivated to maintain significant others' emotional well-being by practicing that behavior, although there is a high opportunity not to do so. In short, smoking abstinence behavior among youth is a relevant context to study significant others' anticipated emotions conceptual efficacy.
Data are collected online. The questionnaire is placed on the website ecampus.bilsonsimamora.com. The researcher invites the respondents to visit the web by sending a questionnaire to this link https://ecampus. bilsonsimamora.com/riset-perilaku-konsumen/ through WhatsApp, Facebook, and Instagram. The respondent will be arriving at the questionnaire by merely clicking the link.
In the introduction, the researcher stresses that the research is academic, and the students' involvement is voluntary. The researcher also does not ask about the identity of the respondents who are treated anonymously. They are free to leave the site or continue to fill the questionnaire. This freedom is expected to increase the objectivity of their responses. Moreover, the absence of personal contact enables them to avoid personal bias.
To avoid position bias, the order of the question is intentionally randomized. There is no missing data because the system requires the respondents to respond to all questions to get permission to finish and submit their responses. As many 242 non-smoker students (120 males and 122 females) have selected with convenience sampling and filled the questionnaires. They come from several universities in West Java, Indonesia, mainly from Kwik Kian Gie School of Business and Information Technology (144 students), Maranatha Christian University (65 students), and various universities in Jakarta (33 students). The average age is 21,67 years. The data are collected from June 2019 to April 2020.
The Ab (e.g., avoiding smoking in life is good for me), SN (e.g., your parents, brothers, sisters, close friends, or anybody who loves you suggest you avoid smoking), SOAJ (e.g., if you avoid smoking, your parents, brothers, sisters, close friends, or anybody who loves you will be happy), PBC (e.g., you can avoid smoking if you want to behave so), and BI (e.g., I intend to avoid smoking) are connected with smoking abstinence behavior. These constructs' instruments use five levels of a Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree).
The positive direction of the emotions in SOAD is expressed by the statement, not by the score. For example, when a respondent 'disagrees' with that statement, the score is two, which means that the happy emotion is low. The SOAD (e.g., if you become a smoker, your parents, brothers, sisters, close friends, or anybody who loves you will feel sad) is related to smoking behavior. This construct's instruments use five Likert-type scale levels ranging from 1 (strongly disagree) to 5 (strongly agree). The negativity of the emotions is stated in the emotional items, not in the measurement scores. For example, in the above statement, the emotion's negativity is represented by the word 'sad'. A score of 4, for example, indicates that the sad feeling is high.
All constructs use a multi-item approach, except the PBC. This construct uses the single-item approach because, according to Sheeran, Trafimow, and Armitage (2003), the TPB in this form is more accurate in predicting the BI than in the multi-item form. The questionnaire is designed to activate the submit button after all the questions are filled in to avoid missing data.
Confirmatory factor analysis with LISREL 8.8 is conducted to investigate constructs validity. A construct is viewed as valid when satisfying factor loadings (FL)>0,5, average variance extracted (AVE)>0,5, construct reliability (CR)>0,60, with Cronbach`s Alpha (CA) coefficient of 0,70 or higher, as suggested by Hair et al. (2016). The attitude (Ab) has a perfect measurement model as shown by the X 2 =0,000 (p-value=1,000) and RMSEA=0,000. The construct also satisfies the validity and reliability criteria specified above. This result supports Ajzen's (2013) notion that the measurement is proven valid and reliable by countless researches.
Subjective norms have a perfect fit initial measurement model as shown by the X 2 =0,000 (p-value=1,000) and RMSEA=0,000. However, the item "I feel pressure from important people in my life (e.g., parents, relatives, close friends, friends) so that I do not smoke" is not valid (FL=0,13). When that item is removed, the remaining items cannot satisfy statistical requirements to construct an independent measurement model. However, by using the factor loadings from the complete model, the remaining two items are valid and reliable (Table 1). This result is different from Solesvick et al. (2012), who have found all items as valid. The failure to confirm the third item validity in this research may come from the item's improper adaptation and translation.
Significant others' anticipated joyfulness (SOAJ) and distress (SOAD) demonstrate a goodmeasurement model as both have RMSEA<0,08, CFI>0,90, and GFI>0,90. Those items are also valid and reliable according to the above criteria. These results are satisfying because the research is the first to conceptualize and investigate them empirically.
The behavioral intention has an initial perfect fit model as shown by the X 2 =0,000 (p-value=1,000). However, the item "How likely are you be to live by not smoking?" is not valid, as shown by factor loading=0,23. By using the factor loadings found in the complete model, the remaining items are valid and reliable (Table 1). This result has the same tone as Fishman, Lushin, and Mandell (2020), who have found that the correlation between "I intend to avoid smoking" (BI1) and "I plan not to smoke" (BI2) is high (r=0,9), while with "how likely are you to live by not smoking", both have low correlations (r=0,5).

RESULTS AND DISCUSSIONS
Descriptive analysis reveals that the level of all constructs is high, that shown by their mean that surpasses the threshold for that category (X̅ i >4,00), and the highest one is behavioral intention (X̅ BI =4,81). It means that respondents' intention to continue, attitude, subjective norm, perceived self-control for that behavior are very high. The respondent also significant others' high anticipated joy for not smoking Notes: FL=factor loading, AVE=average variance extracted, CR=composite reliability, CA=Cronbach`s alpha, X 2 =Chisquare, RMSEA=root mean square error of approximation, CFI=comparative fit index, GFI=goodness of fit index, NC=not calculated, R=removed. and anticipated distress for smoking behavior. The data are provided by request.
The data are all skewed, as shown by the absolute value of the skewness that is greater than 1. The negative direction indicates the flatter tail on the left side of the distribution curve. The Kurtosis scores show that all the variables have a high peak, except for SOAJ, which is flat. These results indicate the data's non-normal distribution as usually found in social science (Bono et al., 2017). The LISREL counted this risk in its goodness of fit indicators.
As exhibited in Table 2, the correlation of attitude and non-smoking intention is very high (r=0,96), while with significant others' anticipated joy and distress are low. Low correlations are also found between behavioral intention with the subjective norm and perceived behavioral control. This result indicates the dominant power of attitude to explain behavioral intention, as shown in the following structural analysis.
The researcher uses structural equation modeling with the maximum likelihood model to test the hypotheses. LISREL program requires 27 iterations to estimate specified structural relationships. Structural model is good fit according to, Comparative Fit Index (CFI) = 0,98, Root Mean Square Residual (RMR) = 0,026, Root Mean Square Error of Approximation (RMSEA) = 0,07. In the structural model (Figure 2), coefficient γ11=0,31 represents the path concerning the influence of significant others' anticipated joyfulness (SOAJ) on attitude toward non-smoking behavior (Ab). With t-value=2,79, the path shows its significance with α<0,01; therefore, H1 is confirmed. Coefficient γ12=0,22 indicates the influence of significant others' anticipated distress (SOAD) on attitude toward nonsmoking behavior (Ab). With the t-value of 2,01, this path is significant at α<0,05, and H2 is supported. The SOAJ and SOAD explain the Ab's variance as much as 25%, as shown by the determinant coefficient (R2).
The power of the direct and indirect determinants of behavior is exhibited in Table 3. The attitude toward behavior has the highest determination on behavioral intention. Based on its relative t-value, this construct represents 43,21% of the total influence of the determinants on behavioral intentions. Surprisingly, the sole influence of SOAJ represented by its relative influence of 13,26% and the cumulative influence of SOAJ and SOAD (13,26% + 9,40%=21,66%) is higher than that of subjective norms (12,36%) ( Table 3). In other words, the AESO is slightly more powerful than the SN.
As expected, attitude (Ab) (β41=0,90, t=9,19,  Figure 2). Standardized coefficient and t-value represent the relative influence of independent variables on a dependent variable (Hair et al., 2016). Because each construct has a considerable standardized error, the research uses t-values instead of path coefficients. The research reveals that SOAJ and SOAD influence students' smoking abstinence continuance behavior positively and significantly. Bagozzi et al. (2016) have found that the works of positive and negative anticipated emotions (AEs) may vary according to the situation differences. These results are confirmed by Gray's BAS/BIS system (Merchan-Clavellino et al., 2019). In this system, approach and avoidance motivations function separately. Approach social motivation is produced by Gray's Behavioral Activation System (BAS), and avoidance social motivation is generated by Gray's Behavioral Inhibition System (BIS). People will fall only into one category. He said that people that high in BAS will be low in BIS, and vice versa.
The research shows that the influence of significant others' anticipated emotions on attitude toward a behavior is inclusive. This inclusivity can be explained using Nikitin and Freund's (2018) work. They have said that approach and avoidance motivations are essential to create and maintain strong affiliations with significant others in an affiliation context. This affiliation is important because, in their research, they still depend financially and emotionally on their family. Even more, high family affiliation is a general trait in Asian countries. With that dependency and high family affiliation trait, it is normal for the student not to smoke to please significant others or avoid experiencing negative emotions.
The coherent view also comes from Scheneider et al. (2017), who described others-oriented personality as a bi-dimensional concept consisting of anticipated guilt and anticipated gratitude. They say that people could be high on particular or both dimensions. When high on both dimensions, people will pursue anticipated gratitude and avoid anticipated guilt simultaneously by performing or not performing a behavior. The research's result implies that the decision to perform smoking abstinence behavior is motivated by individuals' anticipated guilt when they smoke and anticipated gratitude for a stay away from smoking. Significant other's anticipated joyfulness and distress and subjective norms show their efficacies in influencing behavioral intention (Table 3). As social influences, it is interesting to ask whether redundancy occurs among them. Statistically, the influence of the SOAJ and SOAD on behavioral intention is mediated fully by attitude toward behavior. On the other hand, attitude toward behavior and subjective norms influence behavioral intention exclusively. Therefore, it can be said that there is no redundancy between the SOAJ and SOAD and the SN. In other words, the SOAJ and SOAD do not rival but complement the SN as the elements of social influence on behavior.
Conceptually, the difference between the significant others' anticipated emotions (SOAEs) from the subjective norm is as follows. In subjective norms, the initiative is at the hand of significant persons. Ajzen (1991) has defined subjective norms as an individual's belief that important persons in their lives suggest they do or not to do a behavior. In this definition, the initiative is at the hand of significant others, as reflected in the word 'suggest'. It means that significant others are in a position of expecting an individual to perform or not perform a behavior without a preliminary observation about whether the individual has a willingness to confirm that expectation or not. In the TPB, the subjective norm is injunctive because it is related to perceived social pressures from significant others for an individual to perform or not to perform a behavior. In the SOAEs concept, the initiatives are in the hand of the individuals. The social environment's pressure does not generate the intention to perform or not to perform a behavior, as specified in the subjective norm concept (Niemiec et al., 2019). Somewhat, the intention is generated by individuals' willingness to please or to avoid significant others from negative emotions.
The stronger influence of the SOAEs on behavioral intention than that of subjective norms means that in performing smoking abstinence behavior, individuals are the initiators or problem solvers who are more motivated by the willingness to please significant others or to avoid them from feeling sad instead of just following the significant others' will. Lastly, the research shows that attitude toward behavior (Ab) is the strongest and dominant behavioral intention predictor. It means that the intention not to perform smoking behavior is mostly determined by the individuals' considerations. This result is consistent with the smoking tradition in Indonesia. This country has the highest percentage of smoking people in South-East Asian countries. It means that subjective norms have a weak relative influence on the intention of not performing smoking behavior, and the stronger one is individuals' attitude, as mentioned before. Simamora (2016) has investigated the influence of anticipated emotions on goal achievement intention. He has described anticipated emotions as emotional reactions to success or failure to achieve goals in the future. Although those studies are concerned with the individuals' emotions, future research may study the influence of significant others' anticipated emotions on individuals' motivation to achieve goals.
The research investigates the intention for a continuance of a predetermined behavior. It gives no idea about how significant others' anticipated emotions function in a situation in which people are free to choose to perform or not to perform a behavior. Future research needs to be concerned about this limitation. Shoham et al. (2007) have found that smoking behavior can weaken the relationship when smoking is practiced by one party in a dyadic personal relationship and not by another party. The research has not made a consideration about whether the significant others are smokers or non-smoker. The researcher believes that this status will moderate the result, in which the results are more potent when significant others are non-smokers, and vice versa. Future research can investigate this issue.
The respondents majorly are from two universities. Social norms can vary for different populations (Passafaro, Livi, & Kosic, 2019;Silva & John, 2017). It can also be expected that a different society has different emotional reactions to behavior. Although there is no statistical evidence for the moderation of the university origin to the result of the research, other researchers are encouraged to consider this.
It is also interesting to investigate the efficacy of significant others' anticipated emotions in behavior that is almost taboo to discuss openly but strictly guarded by injunctive norms, for example, pre-marital sex in Asian countries. It is expected that the SOAD, not SOAJ, has a strong influence on the effort to avoid it. Moreover, future research can also test the possibility of the exclusive influence of SOAD or SOAJ on other behaviors.

CONCLUSIONS
Significant others' anticipated emotions influence attitude toward behavior directly and behavioral intention indirectly. The influence occurs amid the presence of subjective norms and perceived behavioral control as behavioral intention determinants. Significant others' anticipated emotions do not rival but complement subjective norms as the elements of social influence of behavior in a way that the first has a stronger influence. Those who are concerned can use the concept to motivate people to perform positive behaviors or not perform negative behaviors by reminding them about significant others' anticipated emotions.
Lastly, the research utilizes a single crosssectional design. Its design does not enable the research to check significant others' anticipated emotions before and after performing or not performing a behavior. The interesting questions are, do the SOAJ and SOAD strengthen, weaken, or disappear in line with the time flows? Can they function interchangeably? Future research can use a longitudinal research design to answer these research questions.