ATTITUDE AND PURCHASE INTENTION TOWARDS TEMPE PRODUCERS JOINING THE PROTESTS A Study in Central Java

The theory of attribution has been widely used in marketing studies. However, this theory contains some limitations as well. Using the context of protests of tempe producers, this study aims at minimizing the limitations by integrating the attribution theory with collective action theories, prosocial theories, and expectancy-value theories in explaining and predicting the behavior of product consumers and individuals/groups that joined the protests. To answer the research objectives, this study proposed a model integrating the attribution of perceived motives of tempe producers joining the, empathy towards tempe producers, attitude towards tempe producers, and subsequent purchasing intention. The results indicate that the proposed model is supported by the data. This implies that the integration of the attribution theory with collective action theories, prosocial theories, and expectancy-value theories can minimize the limitations with regard to the criticism in neglecting motivation in attribution processes and confounding between attribution processes and attributional


INTRODUCTION
There have been numerous studies that use the attribution theory to explain, predict, and understand behavioural phenomena (see for examples Sparkman and Locander, 1980;Furse et al. 1981;Teas and McElroy, 1986;Golden and Alpert, 1987;Bitner, 1990;Bitner et al. 1994;Stern, 1994;Taylor, 1994;DeCarlo and Leigh, 1996;Raghubir and Corfman, 1999;Dixon et al. 2001;Laczniak et al. 2001;Prabhu and Stewart, 2001;Maxham and Netemeyer, 2002;Bendapudi and Leone, 2003;Rifon et al. 2004;Tsiros et al. 2004;Morales, 2005).By reviewing the previous studies that use the attribution theory, Wang (2008) identifies a number of limitations of the theory in explaining behaviour, namely: (1) lack of the holistic view of attribution; (2) confounding between attribution processes and attributional processes; (3) neglecting motivation in attribution processes; (4) blurring differences between causal attribution and trait attribution; and (5) misplacing attribution as entire cognition.This study aims at minimizing the limitations in the context of protests of tempe producers.Protests are defined as "an occasion when people come together in public to express disapproval or opposition to something" (Longman Advanced American Dictionary, 2003).Reasons for using this context of protests of tempe producers are (1) within a year, between June 2012 and September 2013 the protests of tempe producers had occurred twice.The producers halted the production for 3 consecutive days to protest the hike in the price of soybean, a raw material to make tempe, in a very short time interval; and (2) the protests were covered and reported by some large scale printed and electronic media.
According to Manalu (2009), from the beginning the perspective of social behavior has made significant contribution by stating that the root caused of any conflict, violence, social movement, and protest is dissatisfaction.These forms of dissatisfaction may be directed to the prevailing norms, the unfair social structure, the tyrannical political system, exploitative economic policies, or discrimination against particular groups or identities all of which can be perceived from different ideologies and perspectives.In our opinion, this can also be viewed from the perspective of the attribution theory.Reviewing the current literature pertaining to protests and customers, we found other gaps as follows: (1) most literature focuses on identification of the determinant factors of why consumers take part in the protest; (2) attitude is not the focal point in explaining why consumers take part in the protest; and (3) only few literature discusses consumers' attitude and puchase intention towards the products manufactured by producers taking part in the protest.Based on the aforementioned, the objectives of our study are to narrow these gaps by integrating the attribution theory with collective action theories, prosocial theories, and expectancy value theories to explain, predict, and understand consumers' attitude to individuals (groups) involved in the protests.

THEORETICAL REVIEWS Limitation of the Attribution Theory
Attribution means an attempt to understand the underlying antecedents of other people's behaviour, and in some cases, the underlying antecedents of our own behaviour (Baron and Byrne, 2005).Wang (2008) states that there are five weaknesses when using the attribution theory in marketing studies.The weaknesses are as follows: First, lack of the holistic view of attribution.Previous studies on marketing usually only adopt one single approach such as consumers when processing information about a person or object.For an example, some studies use the corespondence inference theory to investigate attribution from consumers to the message delivered by marketers.To process information, previous researchers merely view that the attribution made by consumers is based on a single factor, either the marketers or the marketplace environment.However, in an attribution process, consumer attribution may be made not only from the marketplace environment where customers make transanctions, the products offered, and/or the customers themself, but also based on the information and experience that a customer has in relation to the marketing massage; Second, confounding between attribution processes and attributional processes.The attribution theory is frequently used as the theoretical basis to explain the antecedents and consequences while by definition and concept, this theory can only explain antecedents (attribution processes).For an example, in a study, although consumers' satisfaction can be explained using this theory, the consequences (attributional processes) of attribution theory will not be able to explain quite well.In other words, the attribution theory, as a theoretical basis, is used as an independent variable, mediating or moderating variable, which is frequently inappropriately applied; Third, neglecting motivation in attribution processes.In attribution process, observers' motives significantly influence the attribution.This is apparent when observers involve "self-perception".Taylor et al. (2009) mentions that selfperception is an idea that people sometimes conclude their own attitude based on their explicit behaviour rather than internal conditions.In this situation, self-perception serves as the factor that leads to attribution bias.In spite of the importance, in some studies, this factor is often neglected; Fourth, blurring differences between causal attribution and trait attribution.The attribution theory has been used to analyze the characteristics of an object concluded by the observers.For an example, in some studies, covariation theories were used to explain why a consumer, after being told that a certain product is inferior, attributes negative comments to the product and regards the product as inferior.In this instance, this is true when consumers are supported by communication or word of mouth literature.In the example above, covariation theories can be viewed to disclose the causes attributed by the consumer based on the teller' behaviour (including word of mouth communication towards the product), but these theories are not suitable for explaining characteristics of the inferred product (that the product indeed has inferior quality).Confounding between causal attribution and trait attribution may result in false conclusions; and Fifth, misplacing attribution as entire cognition.Attribution processes are frequently considered as the entire cognitive process that leads to the formation of attitude towards a particular object.In some studies, for an example, causal attribution made by consumers is merely based on negative comments of sellers to certain products.Weaknesses of such studies: (1) the characteristics of the products fall within trait attribution, but causal attribution is used; (2) the characteristics of the products inferred simply based on the seller's acting are taken as all the information that can be acquired for cognitive information processing.In reality, however, attitude towards a product can be formed based on information that a consumer obtain or experience.In other words, attitude towards a product may be formed from advertisement persuasion, personal experience, and evaluation about the product (cognitive information processing acquired from various sources, one of them is interpersonal inference).

Adaptation of the Attribution Theory
The attribution theory has been adapted to minimize the inherent limitations.Johnson (2006) employed bibliometrics in the use of this attribution theory and found that this theory has been adapted by a number of researchers such as DeCarlo (2005); Fang et al. (2005).DeCarlo (2005) adapted this theory by integrating the attribution theory with persuasion theories to examine the impact of consumers' suspicion to the motives hidden by salesperson.
Meanwhile Fang et al. (2005) integrate the attribution theory with cultural theories to examine the relationship between sales control system (the outcome, activities, and capabilities), origin of sellers' attribution (attempts, strategies, and abilities), attribution dimensions (internal/external, permanent/temporary) and psychological consequences (work satisfaction, achievement expectation.They demonstrate that the control system differentially affects attribution processes across two cultures: U.S. and China. Mayer and Sparrowe (2013) mention that there are 4 approaches commonly employed in theoretical integration: (1) single phenomenon, two theoretical perspectives; (2) one phenomenon, two seemingly similar theoretical perspectives; (3) applying one theory to the domain of another theory; and (4) streams of research sharing a similar explanation account.This study used the first and third approaches.The approach of "single phenomenon, two perspectives" involves two theoretical perspectives that highlight similar phenomena but from different perspectives.In this approach, the two theories share variables needed to operate the integration.The two theories do not necessarily have to have overlapped domains, but they have to be overlapped as long as what they predict is something specific in a particular context.One of the problems of this integration is addressing the different assumptions of different theories in such a way that other researchers feel comfortable with the resulted integration.Two conditions are required to ensure successful model integration: (a) appreciating assumption of each theory and combining different basic assumptions; (b) ensuring the way the integrated theories will be taken and clearly disclosing why each theory fails to answer questions.
Another approach, applying one theory to the domain of another theory, involves two theoretical perspectives that express similar ideas but contain different phenomena.The application of one theory to another domain will result in more comprehensive perspectives.To be successful, this approach: (a) has to establish clear a relation between the theory and the new domain that enables productive dialogue; (b) ensures suitability between the theory's basic assumption and the new domain; integration has to find the way to anticipate any unsuitability.

Motives of Collective Incentive and Attribution
One of the limitations of the attribution theory is neglecting motivation in attribution processes.Taylor et al. (2009); Baron and Byrnes (2005) call this weakness as the actor-observer effect, which means that when we observe other people's behavior, we tend to relate their behavior to their dispositional quality, but when we explain our own behavior, we explain it based on situational effects.
Information about the protests of tempe producers will attract consumers to get to know further about the value system held by tempe producers.Considering the fact that tempe producers earn profit from selling their products, consumers will easily conclude that the protests are economicallymotivated.However when the protests are found everywhere, consumers will find it difficult to conclude that the protests are merely economically-oriented.Furthermore, since there are free riders who are involved in the protests (i.e.tempe producers who did not join the protests), one (consumers) will find it difficult make conclusions about why other people (tempe producers) behave in a particular way.With regard to free riders, Mancur Olson in Coleman (2010) mentions that in offering public goods in industries consisting of a great number of small firms, the activities that promote collective interests of the involved firms (i.e., legislative lobby for industrial favor) will be done in a lower frequency than those in industries comprising of a single or few large firms.Activities of one firm are profitable for all, and for small-scale firms, the profit generated is not an adequate reason to do the activities.For the firm which becomes the larger fraction of a particular industry, the activities deserve to be done, although the activities are also favorable for smaller firms in the industry.
This study integrates the attribution theory with the collective action theory.The assumption of the attribution theory is that individuals are motivated by particular objecttives to understand and organize environment so that they will always conclude the reasons why people have particular behavior.Meanwhile, the assumption of the collective action theory is that individuals involved in the activity, when left without monitoring, make decision to have particular behavior based on personal/selective interests instead of collective interests.Some other researchers like Olson, Oberschall, and Oliver in Klandermans (1997) state that from the perspective of motives, the core of the collective action theory is distinction between collective incentive and selective incentive.In the context of the protests of tempe producers, attribution of protest motives takes two types: collective incentive (i.e., to change the policy of the government) and selective incentive (i.e., to merely attract attention).
The two theories are integrated using the variable of consumer attribution about the perceived motives of tempe producers joining in the protests (perceived motives of collective incentives) and the variable of attitude towards tempe producers joining in the protests.Individual unit analysis of the two variables can estimate one's trust and feeling directed to a group of people.Thus, it is hypothesized that: H1a : The higher the consumers' perceived motives of collective incentives, the more positive the consumers' attitude towards tempe producers joining in the protests

Empathy and Attribution
As mentioned previously, when criticizing the act of neglecting motivation in attribution processes, we tend to observe other people from different perspectives from when we observe ourselves.Regan and Totten (1975) demonstrate that in particular situations these limitations can be reduced when we have empathy towards the people whose behavior we observe.In other words, in condition where we have empathy on someone, we tend to conclude the behavior of the person similarly as how the person perceive it.Some studies have shown that attitude towards groups of minority can be improved by creating or stimulating empathy towards the groups (of minority) (Finlay and Stephen, 2000;Vescio et al. 2003;Dovidio et al. 2004).
Empathy is a complicated response and has effective and cognitive components (Mashoedi, 2009: 128).Cognitively, empathy enables someone to understand views and to learn about cultural practices, norms, values and beliefs of external groups (Miller, 2010).Cognitive empathy can also reduce differences in perceptions that may lead to better presumption.Affective empathy can be developed from the empathy towards other people's suffering.In so doing, we will have favorable changes in attitude towards external groups.Empathy also enables people to realize that they are different from others in a positive manner.In other words, empathy towards external groups may result in establishment of positive attitude towards external groups (Stephen and Finlay, 2009).Batson et al. (1997) argue that inducing positive empathy towards the stigmatized group members will improve the attitude to the whole individuals and groups.
The study conducted by Bickman and Kamzan (1973) suggests that people in super market are more likely to give some money to someone to purchase milk rather than candies.Darren George in Taylor et al. (2009) mentions that students have better empathy and are not easily irritated to friends who have academic problems due to external control rather than to friends who have academic problems due to their laziness.This finding shows that when observers perceive an event as deriving from an internal factor, willingness to help is low.On the contrary, when an observer perceives that an event is derived from an external factor that they cannot control it internally, willingness to help is high (the people deserve help).
Assumptions of the prosocial theory is that an evolutionary approach revealing that tendency to help is a part of our genetically inherited evolution; the sociocultural perspective approach suggests the importance of social norms that rule when we should offer help to other people in need.In the context of protests, both of these teories are integrated using the variable of empathy towards tempe producers joining the protests and the variable of attitude towards tempe producers joining the protests.The individual unit analysis in both variables may estimate the belief and feeling to someone or a group of people.Hendarto et al. (2013) reveal that in response to the increasing price of soybean, tempe producers had to make some adjustments to their products.The adjustments were made to keep them from increasing the selling price to consumers.The adjustments include (1) reducing the profit resulting from the reduced production volume; (2) reducing the size of tempe; and (3) using lowerquality soybean.In fact, consumers will conclude that tempe producers have no control over the ongoing situation.Therefore, we hypothesize that: H1b : The higher the consumers' empathy towards tempe producers joining the protests, the more positive the consumers' attitude towards tempe producers joining the protests.

Purchase Intention and Attribution
Previous studies reveal that the attribution theory influences attitude and behaviour.However, as Fig. 1 shows, by definition and concept, this theory can only explain the antecedents (attribution processes).To explain consequences of the attitude, we integrated the attribution theory with the expectancy-value theory.The expectancy-value theory is the derivative of the decision making theory.The decision making theory assumes that individuals will calculate the advantages and disadvantages as well as costs and benefits of alternative actions.Initially, individuals will make some alternative actions and choose one of them.The chosen alternative is the most beneficial one.In the expectancy value theory, the available alternatives are extended by incorporating some additional elements.
In the present study, both theories are integrated using the variable of purchase intention towards the products of tempe producers joining the protests.The individual unit analysis in the variable of purchase intention has a clear and suitable relation to be integrated in order to estimate the expectation which is the consequences resulted from previous conclusions (individuals' belief and feeling to someone or a group of people).Therefore, we propose the following hypothesis: H2 : The more positive the consumers' attitude towards tempe producers joining the protests, the higher consumers' purchase intention towards the products of tempe producers joining the protests.
The conceptual model which integrates the relationship among the hypotheses (Hypotheses 1a, 1b, and 2) appears in Fig. 2.

METHODS
Because the protests under study had taken place, it is assumed that respondents have prior knowledge and know that the protests of tempe producers bear a great risk.In relation to consumer complaints, Singh and Wilkes (1996) employed the critical incident approach, where respondents are required to recall their unpleasant experience.A little bit different from what Singh and Wilkes did, to minimize memory bias, Scammon and Kennard (1983); Kim et al. (2003) required respondents to imagine an unpleasant situation that may come up in future time.Different from those studies, we employ sample and filter questions.In relation to the sample, we collected data from respondents living near tempe production centers.It is expected that they know clearly about the protests.Filter questions were asked to the respondents concerning whether they know about the protests or not.If they know and remember the protests, then they were required to go on completing the questionnaires.In case they fail to recall it, respondents were required to return the questionnaires.

Sample
The sample in this study consisted of two groups selected based on time; the first group consists of the sample collected before the second protest (n1 = 120) and the second group consists of the sample collected after the second protest (n2 = 325).The sample was collected at tempe production centers in 5 kabupatens (regencies) in the Provinces of Jogjakarta and Central Java using purposive sampling.Tempe production centers are clustered in particular regions.

Figure 2
Consumer Perceived Motives of CI, Attitude, and Purchase Intention Model Bailey (1994: 96) mentions that the advantage of this sampling technique is that researchers can employ their prior knowledge and experiences to select respondents.In conformity with the technique, the inclusion criteria are: (1) giving consent to participate in the study; (2) women (housewives); and (3) living around the cluster of tempe production centers.Housewives were recruited as the sample since, as Junaedi (2006) states, they are the purchase decision makers in their own households, especially for the respondents in Jogjakarta (Central Java).
The data were collected within one month, from August to September 2013.A total of 517 questionaires was distributed to and collected from consumers living around the cluster of tempe production centers.Of the total questionaires distributed, 67 questionaires were excluded from the sample because of incomplete responses, lack of prior knowledge, and those respondents' occupation that has a direct correlation with tempe producers.Thus, 445 sample (n1 = 125; n2 = 320) remained for the final analysis, which constitute a 86,91 percent usable response rate.In brief, most respondents hold senior high school education level (32.9%); aged between 30-39 years (49.5%);monthly expenditure ranges between 750.000 and 1.900.00(52.3%); housewives (40,4%); with more than 4 family members living in the same house (46,7%); and Moslems (98,2%).The target of both protests of tempe producers is the government.The factor triggering the protests at T1 was the increased price of soybean, the raw material to make tempe, due to draught in Midwest America.The factor triggering the protests at T2 was the increased price of soybean due to the deflated exchange rate of Rupiah to US Dollar.The root causes of the protests at T1 and T2 are similar: (1) failures of the government to be self-reliant in soybean provision so that the government barely has no control over the price; and (2) inabilities of the government to control soybean supply at national level (Hendarto, 2014).

Measures
A structured questionnaire was developed.The questionnaires consisted of three main parts.The first part consisted of filter questions.This part briefly presents the protests and asks the respondents if they know about the protests.If they reply "no", then the respondents were required to stop answering the questionnaires.The second part asks about consumers' perceived motives of collective incentives of tempe producers joining the protests, empathy towards tempe producers joining the protests, attitude towards tempe producers joining the protests, and purchase intention towards products of tempe producers joining the protests.The third part contains the profile of the respondents.This part contains another group of filter questions asking about the respondents' occupation.If the respondents' occupation is related to tempe producers, the data obtained are excluded from the analysis.
Consumers' perceived motives of collective incentives is a process with which consumers understand other people (tempe producers) joining the protests, whether these tempe producers joining the protests have collective objectives or selective objectives.The for this variable is based on the results of the content analysis of the protest-related reports in printed media.The data retrieved from national printed media related to the first protest (23 July to 30 July 2012) were collected and analyzed.Results of the analysis showed that the protest was resulted from: (1) failures of the government in maintaining soybean selfsufficiency; and (2) inabilities of the government in controlling the national soybean supply.Therefore, 2 items of measurement were developed using a bipolar scale.The measurement was made by asking consumers' opinion about producers' motives in joining the protest (influenced by the interest of soybean importers -wishing the government to have soybean self-sufficiency) and (failure in affording the increased price of soybean -wishing the government to have a better policy to control the nationnal supply of soybean).
Another scale was developed based on modification of the measurement items used in previous studies.Since the scale was in English, to ensure accurate translation, we performed back translation.We asked a language expert from State University of Yogyakarta to translate the items from English to Indonesian.The resulted Indonesian version was then back translated into English by another language expert from the Center of Asian Pacific Studies Gadjah Mada University.Further modifications were made as needed.
The scale for attitude towards tempe producers joining the protests was modified from the study conducted by Homer (1995): useless-useful; negative-positive; dislikelike.The scale for empathy towards tempe producers joining the protests and purchase intention used 5-point Likert's scale (1= strongly disagree; 5= strongly agree).The scale for empathy towards tempe producers was modified from Bagozzi and Moore (1994).Three types of scales were used: (1) feeling as if the respondents experience the problems that these tempe producers deal with (2) sharing the same feeling as what tempe producers feel; and (3) tendency to arouse internal desire/wish to provide support.Purchase intention towards the products of tempe producers joining the protests was modified from Lee et al. (2008) and Hendarto (2009).Two types of scales were used: if the price and quality are similar (between the products of tempe producers joining the protests and those who did not join the protests), then I am (1) willing to consider the products of tempe producers joining the protests (2) willing to purchase the products of tempe producers joining the protests.

Results
The data analysis was conducted in three phases.Phase 1 analyzed the measurement (i.e., the Social Desirability Response, validity, and reliability) including assessing the perceived motive of collective incentive of tempe producers joining the protests, attitude towards tempe producers joining the protests, and purchase intention towards the products of tempe producers joining the protests.Phase 2 estimated the structural relation of the construct proposed in Fig. 2 for a total sample of 445 (n = 445).Phase 3 estimated the structural relation of the construct proposed in Fig. 2 by separating the sample group before the second protest (n1 = 120) and the sample group after the second protest (n2 = 325).

Measurement Analysis
The field test aims at examining the question items (wording, time required and the instruction), the social desirability response (SDR), and the construct (validity and reliability).The pretest was conducted with 3 local residents in a face-to-face setting.The results of the pretest confirmed the adequacy of the measure item (i.e., wording, response time, and instruction).SDR is a tendency of individuals to avoid/ refuse undesirable action/ behaviour (Zerbe and Paulhus, 1987).In other words, this SDR is usually viewed as one's tendency to bring in oneself to something that other people like, although it conflicts with his/ her true feeling (Tyson, 1992).Sensitive topics such as sex or taboo topics such as suicide are likely to be responded normatively (Bailey, 1994) because respondents refuse or feel ashamed of either discussing or disliking the topic of thestudy and are afraid of the negative consequence of their answers.
Some strategies can be employed to overcome SDR.First, Crowne and Marlowe in Jo, Nelson and Kiecker (1997) developed a scale to measure the general tendency in respondents' answers.This scale consisted of 33 general question items frequently not related to the research topic so that they do not measure sensitivity of a construct.Other limitations are suggested by Podskoff and Organ (1986).They demonstrate that this scale may reduce the validity of the measurement scale and thus it is ineffective in controlling SDR.Middleton and Jones (2000) also add that there are differences in the reliability of this scale when used for cross-cultural respondents, where reliability of this scale will significantly be lower when used for respondents of eastern culture than those of western culture.Second, Bernard Phillips in Bailey (1994) proposed other strategies: (1) question items that require acknowledgment from respondents concerning the behavior not complying with social norms are arranged in such a way as if that the respondents have answered the questions; (2) question items do not mention consensus to social norms; (3) question items do not mention that certain behavior, that is not in compliance with social norms, is a violation but reveals that the behavior is extensively practiced; (4) using/selecting a euphemistic language style for those question items; and (5) the questions that require respondents to criticize some people or groups (negative) should allow the respondents to say complimentary comments (positive), so that they feel convenient for having expressed fair and polite comments.Third, Junaedi (2006) compared answers to direct questions and answers to indirect questions.Direct questions require respondents to answer about what they think while indirect questions require respondents to answer about what "other people" think/do in relation to a particular issue (Jo et al., 1997).
Based on those strategies, we compared answers to direct questions and answers to indirect questions.After obtaining all the answers (n = 25), the initial step was examining the data normality.Because the data is not normally distributed, non-parametric testing was performed.Results of the Wilcoxon Signed Ranks Test suggest that there is no significant differences between answers to direct questions and answers to indirect questions.This implies that question items that will be used in the study will not result in normative answers.
After the SDR, the validity and reliability were then tested.Direct questions were used.Why? Fisher (1993) explains that although indirect questions are proven to be more effective in controlling SDR bias (as reflected in a higher score of undesirable social behaviour between direct questions  Jo, 2000).To deal with the aforementioned problem, the present study used direct questions to ensure the testing of validity and reliability.
To test the validity, this study employs Exploratory Factor Analysis (EFA) using principal component extraction with varimax rotation, as recommended by Churchill (1979).Gudono (2011) informs that EFA is used when a theory or hypothesis concerning the number of factors (constructs) and which variables related a priori to a particular factor have not been identified.Therefore researchers are "free" in data exploration.Accordingly, the exploratory analysis is more suitable for theoretical establishment.In addition, the relation between latent variables and observed variables is not previously specified; the number of latent variables is not determined before the analysis; all latent variables are assumed to have influenced all observed variables; and error measurement must not be correlated (Wijanto, 2008).
Results of EFA (n = 75) tested using SPSS 13.1 indicate a satisfactory model fit.In EFA, convergent validity is evaluated by the significance and magnitude of the loading factors.The magnitude of the loading factors ranges from 0.72 to 0.92 and the value of Average Variance Extracted (AVE) is higher than 0.5.This finding is in line with that of Fornell and Larcker (1981), who require the discriminant and convergent validities to have: (1) a significant loading factor that is higher than 0.7 and each indicator of a latent variable is different from the indicators of another latent variable.This is shown by a loading score that is higher than the construct score; and (2) the value of AVE (Average Variance Extracted) that is above 0.5.After testing the discriminant and convergent validities, the internal consistency was examined using an alpha coefficient.Results of the reliability testing in this study range from 0.62 to 0.92.Nunnaly and Bernstein (1994) indicate that for exploratory work, reliability of 0.6 is adequate.

The Structural Model for the Total Sample
The structural model in Fig. 2 was tested using AMOS 4.01.The Model showed a good overall fit (χ 2 = 48.83;RMSEA = 0.03; GFI = 0.98; AGFI = 0.96; and CFI = 0.99).Because the model showed satisfactory fit, the coefficient of structural estimation was employed to evaluate the hypotheses.The SEM results for the total sample after and before the second protest are presented in Table 2.
As hypothesized in H1, consumers' perceived motives of collective incentive of tempe producers joining the protests positively and significantly influence attitude towards tempe producers joining the protests (γ = 0.32, t = 4.39).In H2, consumers' empathy towards tempe producers joining the protests positively and significantly influence attitude towards tempe producers joining the protests (γ = 0.32, t = 4.56).Likewise, in H3, attitude towards tempe producers joining the protests positively and significantly influence purchase intention (γ = 0.31, t = 4.24).

The Structural Model for the Total Sample After and Before the Second Protest
After testing the structural model for the total sample, each sample group was tested.The analysis showed that both and before the second protest, the results are satisfactorily fit.For the sample after the second protest occurred, χ 2 = 65.47;RMSEA = 0.06; GFI = 0.96; AGFI = 0.94; and CFI = 0.96); while for the sample after the second protest occurred (χ 2 = 25.58;RMSEA = 0.00; GFI = 0.96; AGFI = 0.93; and CFI = 1.00).

Discussion
The theory of attribution has been widely used in marketing studies.However, this theory contains some limitations as well.Using the context of protests of tempe producers, this study aims at minimizing the limitations by integrating the attribution theory with collective action theories, prosocial theories, and expectancy value theories in explaining and predicting the behavior of product consumers and individuals/groups that joined the protests.Our results indicated that the proposed model is supported by the obtained data, both the data retrieved from the total sample and from the sample before and after the second protest.Integration of the attribution theory with the collective action theory and the prosocial theory may reduce the limitation of neglecting motivation in attribution process.
Fig. 4 shows that when consumers know the protests of tempe producers, they will seek answers for the antecedents of the protests based on the underlying motives.Plus-One's action is perceived useful for a great number of people.

Selective Incentives
Dispositional Attribution Perceiving that one's action derives from such stable characteristics as personality.
Minus-One's action is perceived useless for a great number of people.

Figure 4 Integration of the Attribution Theory with the Collective Action Theory
If consumers perceive that the antecedents of the protests derive from external factors (beyond one's own will), then collective incentive motives as the antecedent of the protests will be selected by consumers.On the contrary, if consumers perceive that the antecedents of the protests derive from internal factors ("inherent characters of the individuals or groups"), then selective incentive motives as the antecedent of the protests will be selected by these consumers.
In the context of the protests of tempe producers, integration of the attribution theory with the prosocial theory can explain emotional reactions of the respondents.Perceptions that the antecedents of the protests are beyond one's own control will result in the feeling of empathy so that willingness to help is high.The hypothesis derived from the integration of the attribution theory and the prosocial theory (H1b) showed that the higher the consumers' empathy to tempe producers joining the protests the more positive the attitude towards these tempe producers who joined the protests and vice versa.
This results support the study conducted by Regan and Totten (1975) which mention that in conditions where we have empathy towards other people, we tend to conclude that the behaviour of other people is similar as the behaviour of the person who observes it.These also support the results obtained from studies conducted by Deitz et al. (1982); Deitz et al. (1984); Lambert and Raichle (2000); Sakalli-Ugurlu et al.
(2007) that empathy towards rape victims will result in positive attitude towards the victims.Meanwhile, as studied by Batson et al. (1997) for the stigmatized group (i.e., patients of AIDS, the homeless, and criminals convicted for murder), positive empathy will similarly result in positive attitude towards the stigmatized group members.
Even, positive empathy towards a stigmatized group member will be generalized to all members of the group.In a visual form, the resulted integration can be described as follows: Fig. 5 shows that when consumers perceive that the root cause of a protest is one's or group' needs that are uncontrol-lable then empathy will arise.On the contrary, if the consumers perceive that the root cause of a protest is one's or group's needs that can be controlled, it will result in dislike and irritation (the opposite of empathy).
These feelings of empathy and nonempathy will result in willingness to help at either a low or high extent.Integration of the attribution theory with the expectancyvalue theory can reduce the limitation about the confounding of attribution processes and attributional processes.

Perceived antecedents of one's need Emotional reaction to people in need Willingness to help
Uncontrollable Beyond one's own control.Empathy High -One deserves help Can be controlled Something that can be controlled by someone.
Non empathy/ irritated Low-One deserves no help

Figure 5 Integration of the Attribution Theory with the Prosocial Theory
When attitude towards an object leads to likeness (resulting from attribution processes), consumers will consider and purchase the product of the object.It implies that the more positive the consumers' attitude towards tempe producers joining the protests the higher the consumers' intention to consider and to purchase the products of those tempe producers joining the protests and vice versa.In a visual form, the integration can be described as follows: Attitude towards an object (someone/ group) Behavioural (purchasing) intention (subject' responses to an object)

Positive
Psychological tendency expressed by evaluating a particular entity that leads to likeness.
High -willing to consider and purchase the products.

Negative
Psychological tendency expressed by evaluating a particular entity that leads to disfavor.
Low -unwilling to consider and purchase the products.

Figure 6 Integration of the Attribution Theory with the Expectancy-Value Theory
Fig. 6 shows that when result of the attribution process suggests that consumers have positive attitude (psychological tendency expressed by evaluating a particular entity that leads to likeness) the resulting attributional process is high (willing to consider and purchase the products of producers joining the protests).On the contrary, if result of the attribution process of the consumers leads to disfavor to a particular entity, the attributional process will result in unwillingness to consider or purchase the products of producers joining the protests (low purchase intention).
Why is consumers' purchasing intention high when consumers have positive attitude and why is it low when they show negative attitude?This has something to do with consumers' expectancy value.The expectancy-value theory postulates that the behavior of individuals is the function of the value expected from particular behavior.The higher the probability that particular behavior has a specific result and the higher the individuals' evaluation to the result, then the higher the probability that someone will behave that way.In the context of the protests of tempe producers, the probability that after the protests tempe producers will produce tempe as usual and that consumers will be able to get tempe at equal or higher quality will be considered by consumers as "something more favorable" than the probability that consumers will switch to other products or producers.

CONCLUSIONS AND FUTURE RESEARCH
The objectives of this study are: (1) reducing limitations of the attribution theory by integrating the theory with the collective action theory, the prosocial theory, and the expectancy-value theory to explain and predict consumers' behavior to individuals (groups) joining the protests; and (2) verifying those limitations in the context of the protests of tempe producers occurring at 2 different time periods.For those reasons, a model was then proposed to reduce and verify the limitations of the attribution theory.
Our results indicated that the proposed model is supported by the obtained data, both the data retrieved from the total sample and from the sample before and after the second protest.Integration of the attribution theory with the collective action theory and the prosocial theory may reduce the limitation of neglecting motivation in attribution process.
This study is subject to several limitations.First, the study cannot specifically reduce other attribution biases (i.e., lack of a holistic view of attribution, blurring differences between causal attribution and trait attribution, misplacing attribution as entire cognition).For cultural bias and in-group bias, Choi et al. (1999) mention that the people with East Asian culture are more sensitive to situational factors.Therefore, when considering the existing social context, they do not have to assume that other people's behavior (tempe producers joining the protests) is related to internal attribution.In collective culture, respon-dents have a lower tendency to spontaneously interpret particular behavior as reflection of internal trait (Newman, 1993).For in-group bias, Takwin (2009) mentions that we tend to prefer our own group members than members of other groups.Although the study does not specifically focus on cultural and in-group bias, the SDR test at the initial stage is expected to reduce the bias.Second, measurement of the variable of perceived motives of collective incentives of producers joining the protests is different from one protest to another protest.Because our study is specific to the protests of tempe producers, future study should develop measure instruments that can be used to assess other major constructs of a protest.Third, because of the lack of sample frame from the respondents, this study employed a non-probability sampling technique.Using this sampling technique will enable generalizability of these research findings as long as it is performed cautiously.Forth, in this study, the respondents were taken from areas situated around the centers of tempe production, where the respondents are customers of those tempe producers.With some modification, future studies may be conducted among respondents who purchase tempe from the market or who purchase tempe indirectly from the producers.
Figure 1 Antecedents and Consequences of Attribution

Figure 3
Figure 3 Timeline of the Data CollectionFigure3shows that the protests of tempe producers occurred twice, i.e. in 25 to 27 July 2012 and 9 to 11 September 2013.The target of both protests of tempe producers is the government.The factor triggering the protests at T1 was the increased price of soybean, the raw material to make tempe, due to draught in Midwest America.The factor triggering the protests at T2 was the increased price of soybean due to the deflated exchange rate of Rupiah to US Dollar.The root causes of the protests at T1 and T2 are similar: (1) failures of the

Table 2 Results of the Structural Equation Model for the Total Sample and the Number of Sample Before and After the Second Protest
* Parameter estimates are standardized