FACTORS INFLUENCING FARMERS' PURCHASE BEHAVIOR – THE CASE OF UZ45 IN LAM DONG PROVINCE In Partial Fulfillment of the Requirements of the Degree of MASTER OF BUSINESS ADMINISTRATION In Marketing By Mr: Nguyen Gia Thinh ID: MBA04037 International University - Vietnam National University HCMC August 2013 FACTORS INFLUENCING FARMERS' PURCHASE BEHAVIOR – THE CASE OF UZ45 IN LAM DONG PROVINCE In Partial Fulfillment of the Requirements of the Degree of MASTER OF BUSINESS ADMINISTRATION In Marketing By Mr: Nguyen GIA THINH ID: MBA04037 International University - Vietnam National University HCMC August 2013 Under the guidance and approval of the committee, and approved by all its members, this thesis has been accepted in partial fulfillment of the requirements for the degree. Approved: ---------------------------------------------Chairperson --------------------------------------------Advisor ---------------------------------------------Committee member --------------------------------------------Committee member ---------------------------------------------Committee member --------------------------------------------Committee member i Acknowledge I would like to thank a lot to Prof. Le Nguyen Hau, School of Industrial Management, HCMC University of Technology, for your patience, helpful advices and consults. I also remember in mind the contribution from lecturers of International University – National University HCMC. For my family, my company and my customers, thank you so much for your support. ii Plagiarism Statements I would like to declare that, apart from the acknowledged references, this thesis either does not use language, ideas, or other original material from anyone; or has not been previously submitted to any other educational and research programs or institutions. I fully understand that any writings in this thesis contradicted to the above statement will automatically lead to the rejection from the MBA program at the International University – Vietnam National University Hochiminh City. iii Copyright Statement This copy of the thesis has been supplied on condition that anyone who consults it is understood to recognize that its copyright rests with its author and that no quotation from the thesis and no information derived from it may be published without the author’s prior consent. © Nguyen Gia Thinh/MBA04037/2013 iv Table of Contents Abstract ........................................................................................................................ x CHAPTER ONE – INTRODUCTION ..................................................................... 1 1. Research Background ..................................................................................... 1 2. Statement of the Problem ............................................................................... 4 3. Research Objective.......................................................................................... 5 4. Significance of the Research ........................................................................... 5 5. Research Scope ................................................................................................ 5 6. Limitation of the Research ............................................................................. 5 CHAPTER TWO – LITERATURE REVIEW ........................................................ 6 1. Theory Review ................................................................................................. 6 1.1. Theory of Reasoned Action ......................................................................... 6 1.2. Theory of Planned Behavior ........................................................................ 7 1.3. Technology Acceptance Model .................................................................... 9 1.4. Model of Consumer Decision Making ...................................................... 10 2. The Proposal Framework ............................................................................. 12 2.1. Components of TRA Model ....................................................................... 14 2.1.1. Behavior............................................................................................... 14 2.1.2. Intention .............................................................................................. 14 2.1.3. Attitude................................................................................................. 15 v 2.1.4. Salient Belief ....................................................................................... 16 2.1.5. Subjective Norm .................................................................................. 17 2.1.6. Normative Belief.................................................................................. 17 2.2. Relation among TRA’s Components ........................................................ 18 2.2.1. Behavior, Intention, Attitude and Subjective Norm .......................... 18 2.2.2. Attitude and Behavioral Belief ........................................................... 19 2.2.3. Subjective Norm and Normative Belief .............................................. 20 3. Previous Studies ............................................................................................ 20 4. Research Model and Hypothesis .................................................................. 22 CHAPTER THREE – RESEARCH DESIGN........................................................ 27 1. Qualitative Research ..................................................................................... 27 2. Quantitative Research .................................................................................. 27 2.1. Sampling Procedure ................................................................................... 27 2.2. Measurement Scales ................................................................................... 27 2.3. Questionnaire .............................................................................................. 30 2.4. Data Analysis .............................................................................................. 31 CHAPTER FOUR – RESULT ................................................................................. 32 1. Result of Qualitative Study .......................................................................... 32 2. Sample Description ....................................................................................... 33 2.1. Age ............................................................................................................... 33 2.2. Gender ......................................................................................................... 33 vi 2.3. Education .................................................................................................... 33 2.4. Farm Area ................................................................................................... 33 2.5. Farming Experience ................................................................................... 34 2.6. Crop ............................................................................................................. 34 2.7. Selling Price of Farming Products ............................................................ 34 2.8. Payment Term of Purchasing Pesticides Product ................................... 34 2.9. Weather ....................................................................................................... 35 2.10. 3. Income ...................................................................................................... 35 Data Screening............................................................................................... 35 3.1. Multivariate normality ............................................................................... 35 3.2. Outliner ....................................................................................................... 37 3.2.1. Univariate outliner. ............................................................................. 37 3.2.2. Multivariate Outliner .......................................................................... 38 4. Measurement Model ..................................................................................... 39 4.1. Specification ................................................................................................ 39 4.2. Re-specifying Record ................................................................................. 41 4.3. Identification ............................................................................................... 42 4.4. Construct Reliability and Validity ............................................................ 43 4.4.1. Reliability ............................................................................................. 43 4.4.2. Validity ................................................................................................. 43 4.5. Testing Measurement Model ..................................................................... 46 vii 5. Structural Model ........................................................................................... 48 5.1. Specification ................................................................................................ 48 5.2. Identification ............................................................................................... 50 5.3. Testing Structural Model ........................................................................... 51 6. Hypothesis Testing ........................................................................................ 52 7. Discussion ....................................................................................................... 55 7.1. Attitude, Subjective Norm and Consumption Level ............................... 56 7.2. Belief about Quality, Belief about Cost and Attitude .............................. 57 7.3. Belief about Quality, Normative from Colleague and Retailer .............. 58 7.4. Normative Belief from Colleague, Retailer and Subjective Norm ......... 58 7.5. Application – Marketing Plan for UZ45 .................................................. 60 7.5.1. Positioning........................................................................................... 60 7.5.2. Marketing Mix .................................................................................... 61 7.5.3. Integrated Marketing Communication ............................................ 62 CHAPTER FIVE – CONCLUSION ....................................................................... 65 1. Conclusion...................................................................................................... 65 2. Suggestion for Managers .............................................................................. 65 3. Limitation ...................................................................................................... 66 References .................................................................................................................. 68 Appendixes ................................................................................................................. 70 viii List of Tables Table 1: Overall Measurement Scale ......................................................................... 29 Table 2: Skew and Kurtosis Index ............................................................................. 36 Table 3: Observed Variables Contain Potential Outliners ......................................... 37 Table 4: Mutivariate Outliners ................................................................................... 39 Table 5: Reliability of Scales ..................................................................................... 43 Table 6: Standardized Regression Weights (Factor Loadings) .................................. 44 Table 7: Correlation ................................................................................................... 45 Table 8: Fit Indices - Measurement Model ................................................................ 48 Table 9: Goodness of Fit Indices - Structural Model ................................................. 51 Table 10: Unstandardized Regression Weights - Structural Model ........................... 52 Table 11: Standardized Regression Weights - Structural Model ............................... 53 Table 12: Covariance and Correlation ....................................................................... 55 ix List of Figures Figure 1: Global Pesticide Trading Value 2005-2011. ................................................ 1 Figure 2: Pesticide Import Value of Vietnam 2008-2012 ............................................ 2 Figure 3: Sales History of UZ45 2008-2012 ................................................................ 3 Figure 4: Growth rate of UZ45 2008-2012 .................................................................. 4 Figure 5: Theory of Reasoned Action Diagram. .......................................................... 7 Figure 6: Theory of Planned Behavior Diagram .......................................................... 8 Figure 7: Technology Acceptance Model .................................................................... 9 Figure 8: A Model of Consumer Decison Making..................................................... 11 Figure 9: Proposed Research Model .......................................................................... 24 Figure 10: Measurement Model ................................................................................. 40 Figure 11: Structural Model ....................................................................................... 49 x Abstract To understand consumers’ purchase behavior in the case of product UZ45 (fungicide) in Lam Dong province, based on Theory of Reasoned Action, researcher proposed research model and hypotheses. Measurement scales were developed to measure constructs. Full measurement model was tested by CFA procedure. Then, structural equation model was tested. Models fit the data well. Proposed hypotheses were supported and interpreted. Factors influencing consumers’ purchase behavior includes consumers’ attitude toward consuming UZ45, consumers’ belief about UZ45’s quality and price, subjective norm, normative belief from colleague and normative belief from retailer. Among them, consumers’ attitude toward consuming UZ45, consumers’ belief about UZ45’a quality and normative belief from colleague are more important factors. Consumers’ attitude toward using the product impacts stronger on consuming behavior than subjective norm. Consumers’ belief about product’s quality influences stronger on consumers’ attitude toward using product than consumers’ belief about product’s price. Normative belief from colleague affects stronger on subjective norm than normative belief from retailer. It also impact consumer’s belief about product’s quality and price. The research then suggested that in order to increase sales of UZ45, the company should position on consumer mind that the product is good quality and on retailers’ mind that UZ45 is a profitable product. Some features were suggested to be included in the product such as solubility, sticky and fungi control. The research also mentioned the possibility of increase consumers’ price to fund communication. Retailers’ profit was also noted as an issue in distribution channel. An integrated marketing communication strategy was suggested. Marketing activities should focus more on farmer than retailer because farmer can affect other farmers better. Advertising and event were suggested to communicate to farmers. A farmer can be considered as a good spoken person in the communication plan because he can influence well consumers’ perception about quality. But care also should be spent to retailers because they indirectly affect consumers’ attitude, perception about quality and cost beside direct effect on subjective norm. Trade promotion is the recommended tool to communicate with retailers. xi This page is intentionally left blank 1 CHAPTER ONE – INTRODUCTION 1. Research Background From the time when the first pesticide was used in around 1000 BC, the global pesticide industry trading value was worth more than 23 billion USD in 2011 (FAO statistic, 2013). The global trading value increased continuously from 16 billion USD in 2005 to 25 billion USD in 2008, then faced a little bid fluctuation from 2009 to 2011 as shown in Figure 1. In 2011, France, Germany and USA are three largest pesticide exporters in the world. Global Pesticide Trading Value 2005-2011 (Million USD) 25,065 16,516 2005 24,608 23,411 20,114 23,469 16,509 2006 2007 2008 2009 2010 2011 Figure 1: Global Pesticide Trading Value 2005-2011. (Source: FAO statistic, 2013) In 2010, Vietnam was one of 14 largest pesticides import country in the world (FAO Statistic, 2013). As shown in Figure 2, annual pesticide import value of Vietnam continuously increases in 2009-2011 period. In 2012, the import value reaches 745 million USD, a little lower than 2011. There are totally 1,405 active ingredients and 3,514 brand names and around 200 companies operating in this market (Circular No.10/2012-NN-PTNN, Vietnam government 2012). 2 Vietnam Pesticides Import Value 2008-2012 (Mil USD) 752 745 2011 2012 577 541 521 2008 2009 2010 Figure 2: Pesticide Import Value of Vietnam 2008-2012 (Source: A&A Co,.Ltd) Joined Vietnam pesticide market from 2002, A&A Co,.Ltd1 – an Indian company has been developing sustainably with annual growth rate exceeding 20%. It possessed a rich product portfolio consisting of herbicides, insecticides and fungicides. Its key fungicide is Mancozeb. Mancozeb is a complex of zinc and maneb containing 20% Mn and 2.5% Zn (FAO specification No.34/1/ts/9, 1980). It has been used as an active ingredient to control harmful fungi named as Phytophthora infestansa which cause Downy Mildew disease in tomato and potato. It is an off-patent chemical product so that every company can manufacture it as long as they have enough technological capacity to do. Because many companies produce and sell Mancozeb, they have to use a brand name to distinguish their product from other competitors. For example, Mancozeb brand name of Dow Agrochem Company (USA) is Dithane and A&A Company’s Mancozeb brand name is UZ452. (1). Company name A&A has been changed for confidential purpose. (2). Product name UZ45 has been changed for confidential purpose. 3 Launched in 2008, UZ45 has grown impressively in the first three years as shown in Figure 3. In last 2 years, although its sales increased year by year, the growth rate has tended to decrease and toughed 7% in 2011 and 3% in 2012 as shown in Figure 4. UZ45 Sales History 2008-2012 (Ton) 150 140 155 120 80 2008 2009 2010 2011 2012 Figure 3: Sales History of UZ45 2008-2012 (Source: A&A Co.,Ltd, 2013) Currently, the company’s marketing plan for UZ45 is spread and just based on subjective assumptions instead of confirmed facts. The word “spread” here briefly described how the company is allocating marketing efforts. They are positioning that UZ45 is both high quality and reasonable cost because they don’t know between quality and cost, which one is more valuable to consumers. They are dividing some budget for consumer’s promotion and some for retailer’s discount because they don’t have evidence that is reliable to be sure that who is more important between consumers and retailers. “Successful marketing builds demand for products and services” (Kotler & Keller, 2012, p.4). The decrease of sales growth rate mentioned above changes 4 approach of board of director. They decide that this is the time for developing a better and more structural marketing plan for UZ45 so that marketing efforts can be allocated rightly to enhance sales of the product. UZ45 Growth Rate History 2008-2012 (%) 50% 17% 7% 0% 2008 3% 2009 2010 2011 2012 Figure 4: Growth rate of UZ45 2008-2012 (Source: A&A Co.,Ltd, 2013) Kotler and Keller (2012) also argued that successful marketing requires that companies fully connect with their customers, understand customers a 360-degree view of both their daily lives and the changes that occur during their lifetimes so the right products are always marketed to the right customers in the right way. So, in order to develop a better marketing plan for UZ45, a research on consumers’ behavior accounts for a crucial role. 2. Statement of the Problem The existing marketing plan is no longer appropriate in supporting sales development of UZ45. The company requests a better and more structural marketing plan so that they can allocate accurately marketing resources to maintain and increase the 5 product’s sales. A deep insight of consumer purchase behavior accounts for a crucial role in this mission. 3. Research Objective The research is aimed to find out main factors influencing farmers’ purchase behavior in the case of product UZ45 in Lam Dong province and suggest marketing strategy to managers to develop the product’s sales. 4. Significance of the Research Knowing main factors influencing consumers’ purchase behavior will help managers to allocate rightly marketing efforts to maintain and grow sales of the product. For instance, if finally the research finds that consumers’ buying behavior is affected more strongly by normative belief from important others rather than his own beliefs, the company’s marketing communication campaign should set his important others as targeted audiences instead of himself. 5. Research Scope This research is carried out within Lam Dong province where accounts for 80% sales of UZ45 in Vietnam market. The population is defined as a set of farmers who have demand of Mancozeb products, within Lam Dong province and no matter what brand he/she is familiar with. 6. Limitation of the Research Findings from this research can be generalized to explain purchase behavior of famers in Lam Dong province and in case of UZ45 product only. 6 CHAPTER TWO – LITERATURE REVIEW This chapter is assigned to present available theories about consumer behavior and the reasons for selecting the most appropriate one to adopt for forming theoretical framework of this research. The background of selected theory is presented in detail. Previous studies which were based on the selected theory are reviewed. Finally, the research model and hypotheses are proposed. 1. Theory Review Several theories of consumer’s behavior or consumer’s decision making are available and can be adopted to explain behavior of consumers in a specific market or product like UZ45 in this research. Hereafter, some theories are listed and briefly described. 1.1. Theory of Reasoned Action According to Ajzen (1985), many behaviors in normal life can be considered as “under volitional control” in sense that people can easily perform those behaviors if they want to do, for example, watch evening news on TV or buy toothpaste at a drug store. The theory of reasoned action is designed to predict volitional behaviors of this kind (Ajzen, 1985). The theory of reasoned action (TRA) is based on the assumption that human being usually behaves in a sensible manner (Ajzen, 1985). According to the authors, behavior can be predicted immediately from its determinant – intention (B ≈ I), intention’s determinants are attitude and subjective norm (I = w1 A + w2 SN), attitude will be predicted from individual’s beliefs about outcomes of performing that behavior and individual’s evaluation about those outcomes (A = ∑bi ei) and finally, subjective norm’s determinants are individual’s perceived social pressure put 7 on him to perform that behavior and individual’s motivation to comply with those social pressure (SN = ∑bjmj) (Ajzen, 1985). Figure 5: Theory of Reasoned Action Diagram. (Source: Fisbein & Ajzen, 1975) Bentler & Speckart (1979, cited in Jerold, Brian & Kathryn, 2002) and Langer (1989) argued that the aim of the TRA is to explain volitional behaviors and its scope excludes wide range of behavior such as those that are spontaneous, impulsive, habitual…because their performance might not be voluntary. To expand the range of behaviors encompassed by the TRA, Ajzen (1985) proposed the theory of planned behavior (Jerold et al, 2002). 1.2. Theory of Planned Behavior The theory of planned behavior (TPB) is an extension of the theory of reasoned action (Ajzen & Fishbein, 1980; Fishbein & Ajzen, 1975) made necessary by the original model’s limitations in dealing with behaviors over which people have incomplete volitional control (Ajzen, 1991). The components of the theory of planned behavior mirror those of TRA, except that perceived behavioral control is added to TPB (Jerold at al, 2002). Perceived behavioral control is “one’s perception 8 of how easy or difficult it is to perform the behavior” (Eagly & Chaiken, 1993 cited in Jerod at al, 2002). The same as TRA, in TPB, immediate predictor of behavior is intention. Intention, a little different, is a function of attitude, subjective norm and perceived behavioral control. As in TRA, attitude is a function of belief about outcomes of performing the behavior and perceived importance of those outcomes and subjective norm is a function of normative belief from important others and motivation to comply. Perceived behavioral control is also assumed to be a function of beliefs, this time beliefs about the presence or absence of factors that facilitate or impede performance of the behavior (Ajzen, 2005). Mathematical view of perceived behavioral control can be described by the formula PBC = CiPi, where “Ci is the control belief that a given factor i will be present; Pi is the power of factor i to facilitate or inhibit performance of the behavior” (Ajzen, 2005). Specially, perceived behavioral control, together with behavioral intention, can be used directly to predict behavioral achievement (Ajzen 1991). Figure 6: Theory of Planned Behavior Diagram (Source: Ajzen 2005) 9 1.3. Technology Acceptance Model Technology Acceptance Model (TAM) was proposed by Davis (1985) in his Ph.D. thesis in Sloan School of Management - Massachusetts Institute of Technology. Stated by author – Davis (1985), the model was developed with two objectives: first, it should improve understanding of user acceptance process, providing new theoretical insights into successful design and implementation of information systems; second, it should provide theoretical basis for a practical “user acceptance test” methodology that would enable system designers and implementors to evaluate proposed new systems prior to their implementation. Figure 7: Technology Acceptance Model (Source: Davis 1985) Attitude refers to the degree of evaluative affect that an individual associates with using the target system in his or her job (Davis, 1985). Davis (1985) also defined perceived usefulness is "the degree to which a person believes that using a particular system would enhance his or her job performance” and perceived ease of 10 use is “the degree to which a person believes that using a particular system would be free of effort”. According to Davis (1985), TAM can be briefly described that: a potential user’s overall attitude toward using a given system is hypothesized to be a major determinant of whether or not he actually uses it (USE = β1 ATT+ ); attitude toward using, in turn, is a function of two major beliefs: perceived usefulness and perceived ease of use (ATT = β1*USEF + β2*EOU+ ) and perceived ease of use has a causal effect on perceived usefulness. Design features, donated by X1, X2 and X3, directly influence perceived usefulness and perceived ease of use (where βi stands for standardized partial regression coefficient and stands for random error). 1.4. Model of Consumer Decision Making Schiffman & Kanuk (2004) proposed a model of consumer decision making. This model is designed to tie together many of the ideas on consumer decision making; it does not presume to provide an exhaustive picture of the complexities of consumer decision making; rather, it is designed to synthesize and coordinate relevant concepts into a significant whole (Schiffman & Kanuk, 2004). According to the model, consumers function as “information processors”. Consumers receive information from input sources then process that information and create outputs. Sources where consumers can get information from consist of companies’ marketing efforts and sociocultural environment. Companies try to push information about their products or services to consumers mainly via communication campaigns such as advertising, promotion, event…Consumers can get information from and be affected by social group like family, friends or discussion groups in social networks… The influence of social class, culture and subculture, although les 11 tangible, are important input factors that are internalized and affect how consumers evaluate and ultimately adopt or reject products (Schiffman & Kanuk, 2004). Figure 8: A Model of Consumer Decison Making (Source: Schiffman & Kanuk 2004) In the process stage, consumers will be influenced by psychological concepts such as motivation, perception, learning, personality and attitudes. The act of marking a consumer decision consists of three stages: need recognition, pre-purchase search and evaluation of alternatives (Schiffman & Kanuk, 2004). The recognition of a need usually occurs when one gets a stress or problem and wants to satisfy it. Then, they will search for a product or service which can be used to settle their problem then evaluate them by some criterion before purchase. Which channels consumers 12 will search on and which criterion they will use to evaluate is make sense with marketers. After processing information, consumer may try to use product or service then evaluate them. They may repeat purchase if they are satisfied. Otherwise, they can turn back to process stage to search for alternatives. 2. The Proposal Framework Among several theories, the Theory of Reasoned Action (TRA) developed by Fishbein and Aijen in 1975 is adopted for examining factors influencing farmers’ purchase behavior in case of UZ45 in Lam Dong province, because of following reasons. Firstly, theory of reasoned action is based on the assumption that human being usually behave in sensible manner; that they take account of information and implicitly or explicitly consider the implications of their actions (Ajzen, 1985). In this research, farmers buy UZ45 just for protecting their crops so that it can be considered as an input material instead of consuming product. In this case, farmers’ buying behavior can be viewed as organization’s purchase behavior (B2B) then it is appropriate to assume that those buying behavior is “sensible” or rational. Therefore, TRA is an appropriate selection. Secondly, consuming UZ45 is so easy and nothing different than “watch evening news on TV or buy toothpaste at a drug store” (Ajzen 1985). Because agrochemical shops are available, farmers can buy a product like UZ45 any time even if they don’t have cash, they will be offered credit by retailers. Additionally, mixing agrochemical products (pesticides) in a sprayer and spraying it on field is a so simple work that farmers can do it automatically. So, this behavior can be 13 considered as “under volitional control behavior”. The theory of reasoned action is designed to predict volitional behaviors of this kind (Ajzen, 1985). Therefore, the selection of TRA is appropriate. Thirdly, TRA is more appropriate than Theory of Planned Behavior (TPB). The co-author of the theory, Ajzen, stated that “TRA is a special case of the TPB. The only difference between the two theories is that the TPB includes perceived behavioral control as an additional determinant of intentions and behavior. In the development of the TRA it was assumed that people have volitional control over the behavior of interest. Under these conditions, perceived behavioral control becomes irrelevant and the theory of planned behavior reduces to the theory of reasoned action”. Fourthly, as described above, the behavior under interest is a very simple action (consuming an agrochemical product) as the view of farmers (consumers) in this research. Therefore, the concept “Ease of Use” in Technology Acceptance Model (TAM) is no longer needs to be measured or accessed. So, TAM is not as good as TRA to examine this behavior. It may be suitable for other behaviors which should be more complicated. Finally, the Model of Consumer Decision Making proposed my Schiffman and Kanuk (2004) is actually a good candidate. As described by authors, it concludes almost concepts mentioned in other model on consumer behavior. All constructs of TRA like behavior, attitude or subjective norm… can be inferred from this model. But containing several concepts makes this model more complicated to be operationalized and researcher contemporarily lacks of supporting facts in 14 operationalizing this model. So, it should be adopted in researcher’s future researches. 2.1. Components of TRA Model 2.1.1. Behavior A key assumption in TRA is that behavior concept, which is the main interest in the model, is under volitional control. As Ajzen (2005) defined, under volitional control means that people can easily perform these behaviors if they are so inclined, or refrain from performing them if they decide against it. Many behaviors in everyday life can be thought of as being largely under volitional control such as vote in political elections, watch the evening news on television, buy toothpaste at a drugstore, pray at a nearby church or donate blood to their local hospitals (Ajzen, 2005). Ajzen, in his official website, also gives some guidelines for measuring past behavior. As example, in order to measure the behavior of doing exercise for at least 20 minutes, three times per week, the author introduced a seven points bipolar adjective scale which raised the statement as “In the past three months, I have exercised for at least 20 minutes, three times per week” then the respondents have to select from one to seven in a seven points scale with two adjectives “false” and “true” stand at extremes. 2.1.2. Intention Intention to perform an action or behavior refers to the extent to which people want or don’t want to perform that action (Ajzen 1991). It is indication of how hard people are willing to try, of how much of an effort they are planning to exert, in order to 15 perform the behavior (Ajzen 1985). The stronger intention to perform a behavior is, the more likely that behavior will be done. Intention is not stable. We should not be surprised that sometime we remember that we supposed to do something but after that we forgot. It means our intention changed. The more time passes, the greater the likelihood that unforeseen events will produce changes in intentions (Ajzen 2005). In order to access the concept of intention, Ajzen also gave an example of measuring the respondents’ intention to perform the behavior of doing exercise for at least 20 minutes, three times per week for the next three months. The respondents will be given a statement that “I intend to exercise for at least 20 minutes, three times per week for the next three months” then have to select from one to seven in a seven points scale with two adjectives “Likely” and “Unlikely” stand at extremes. 2.1.3. Attitude Attitude refers solely to a person’s location on a bipolar evaluative dimension with respect to some object, action or event; represents a person’s general feeling of favorableness or un-favorableness toward some stimulus objects (Fishbein & Ajzen, 1975). Within the context of TRA, we examine that attitude toward an action or behavior, not an object. It is determined by the person’s evaluation of outcomes associated with the behavior and by the strength of these associations (Ajzen, 1985). In empirical extent, there are two ways to measure attitude: direct and indirect. Direct measurement uses seven points bipolar adjective scale. The question will raise a statement about a specific behavior and respondents will be asked to circle the number which best describes their opinion. As example mentioned by Aijzen to examine attitude of patients who have just undergone a major heart surgery 16 toward the behavior of doing exercise for at least 20 min, three times per week for the next three months, the statement raised that “My exercising for at least 20 minutes, three times per week for the next three months would be Bad:__ 1__:__ 2__:__ 3__:__ 4__:__ 5__:__ 6__:__ 7__: Good Pleasant:__ 1__:__2__:__3__:__4__:__5__:__ 6__:__ 7__: Unpleasant”. Respondents was asked to circle the number which best describe their thinking. Indirect measurement of attitude from belief will be described in next part. 2.1.4. Salient Belief Ajzen (2005) argued that we acquire many different beliefs about a variety of objects, actions or events and these beliefs may be formed by direct observation or by accepting information from outside sources as friends, television, newspapers, books, and so on. People can hold a great many beliefs about any given object, but they can attend to only a relatively small number, perhaps eight or nine, at any given moment (Miller, 1956 cited in Ajzen, 2005). Belief about an object provides the basis for formation of attitude toward the object and attitudes are usually measured by accessing a person’s belief (Fishbein & Ajzen 1975). According to TRA model, attitude toward a behavior is measured indirectly by measuring strength of believes about outcomes generated by performing that behavior (so called behavioral belief of salient belief) and evaluation about those outcomes. As explained in item 4 of this chapter, this research is just interested in measuring of salient belief strength. In order to measure strength of the belief of patients, who have just undergone a major heart surgery, that doing exercise will help them to faster recover from surgery, Ajzen introduced an example scale which raised the statement that “My exercising for at least 20 min, three times per week for the 17 next three months will result in my having a faster recovery from my surgery” and respondents will score from one to seven in a bipolar seven points scale with two adjective “likely” and “unlikely”. 2.1.5. Subjective Norm Subjective norm refers to person’s perception of social pressure to perform or not perform the behavior under consideration (Ajzen, 2005). Social pressure can come from several sources, but they must be perceived by receivers to be called subjective norm. In the same above example, Ajzen used seven points bipolar adjective scale again to measure subjective norm perceived by persons upon the action of doing exercise for at least 20 minutes, three times per week for next three months. Respondents will be requested to select the point which likely express their thought after seeing two statements like “most people who are important to me approve of my exercising for at least 20 minutes, three times per week for the next three months” and “most people like me exercised for at least 20 minutes, three times per week in the three months. 2.1.6. Normative Belief Ajzen (2005) defined that normative belief refers to person’s beliefs that specific individuals or groups approve or disapprove of performing the behavior. It is distinct from the term “subjective norm” which refers to person’s belief about social pressure forcing him to perform or not perform a behavior. Back to the example of doing exercise, in order to measure normative belief from doctor, Ajzen mentioned an example sematic differential scale: “My doctor thinks that 18 I should:___ 1__:___ 2__:___ 3__:___ 4__:___ 5__:___ 6__:___ 7___: I should not exercise for at least 20 min, three times per week for the next three months”. The respondents have to select the point which well describes their opinion. 2.2. Relation among TRA’s Components 2.2.1. Behavior, Intention, Attitude and Subjective Norm The theory of reasoned action postulates that a person’s intention to perform (or not perform) a behavior is the immediate determinant of that action (Ajzen, 1985). This relation can be denoted by (B ≈ I). Intention is a function of two basis determinants. One is attitude toward that behavior. Other is subjective norm. According to Ajzen (1985), attitude reflecting one personal in nature is the individual’s positive or negative evaluation of performing the behavior while subjective norm reflecting social influence is the person’s perception of the social pressures put on him to perform or not perform the behavior in question. This relation can be described by below equation: B ≈ I = w1 A + w2 SN Where “B” stands for behavior of interest, “I” stands for intention, “A” stands for attitude toward performing the behavior and “SN” stands for subjective norm; w1” and w2 are regression weights collected from empirical data by regressing of intention on attitude and subjective norm. Generally speaking, people intend to perform a behavior when they evaluate it positively and when they believe that important others think they should perform it (Ajzen, 1985). A great number of studies have provided strong support for the 19 proposition that intentions to perform a behavior can be predicted from attitudes toward the behavior, subjective norms (Ajzen, 2005). 2.2.2. Attitude and Behavioral Belief Attitude toward a behavior is determined by accessible beliefs about the consequences of the behavior, termed behavioral beliefs (Ajzen, 2005). These salient [behavioral] beliefs must be elicited from the respondents themselves, or in pilot work from a sample of respondents that is representative of the research population (Ajzen 1991). In the same example mentioned above to examine exercise behavior, Ajzen guided some questions to be used in pilot study to explore salient beliefs such as what do respondents see as the advantages or disadvantages of doing exercise for at least 20 minutes, three times per week for the next three months. Once behavioral believes are isolated, attitude (A) can be determined by equation below: A = ∑bi ei Where bi stands for strength of salient belief that performing the behavior will lead to outcome i and ei is the evaluation of that person about outcome i. By multiplying belief strength and outcome evaluation, and summing the resulting products, we obtain an estimate of the attitude toward the behavior, an estimate based on the person’s accessible beliefs about the behavior (Ajzen, 2005). It can be seen that, generally speaking, a person who believes that performing a given behavior will lead to mostly positive outcomes will hold a favorable attitude toward performing the behavior, while a person who believes that performing the behavior will lead to mostly negative outcomes will hold an unfavorable attitude (Ajzen, 2005). 20 2.2.3. Subjective Norm and Normative Belief Ajzen (1985) defined that subjective norms are also assumed to be a function of beliefs, but beliefs of a different kind, namely the person’s beliefs that specific (important) individuals or groups think he should or should not perform the behavior. These beliefs underlying the subjective norm are termed normative beliefs. Similar to indirect measurement of attitude, subjective norm (SN) will be measured indirectly from its theoretical equation: SN = ∑bjmj Where bj stands for normative belief of a specific important individual or group, mj stands for person’s motivation to comply with normative belief of that important individual or group. A pilot study also will be requested to explore individuals or groups who affect person’s behavior and respondents’ motivation to comply with that individual or group. Let review above mentioned example about exercise behavior, Ajzen suggested some questions used to isolate effect individuals or groups such as “please list the individuals or groups who would approve (or disapprove) or think you should exercise for at least 20 minutes, three times per week for the next three months”. In order to measure motivation to comply the respondents have to point out which score best describe their want to comply with that individual (group) within seven point Likert scale. 3. Previous Studies Theory of Reasoned Action (TRA) has been using in several fields to examine many kinds of behaviors. 21 McCarthy, Boer, O’Reilly & Cotter (2002) adopted TRA to examine factors influencing consumers’ behavior on consuming beef in Irish market. The authors isolated six behavioral beliefs and four influence groups. Behavioral beliefs include including health, safety, environmental, animal welfare, eating enjoyment and price. Influence group consists of friend, family, dietician and doctor. The survey was done with three hundred respondents. Partial multiple regression technique was employed to analyze the data. Analysis result showed that attitude and subjective norm significantly affect intention with regression weights of .77 and .09, respectively. Among six behavioral beliefs, health, safety and eating enjoyment significantly affect attitude with regression coefficients of .44, .36 and -.32, respectively. Among four influence groups, doctor and dietician are proved to significantly affect subjective norm with weights of .21 and .19, respectively. “The findings support the usefulness of this model in understanding behavior towards beef (McCathy at al, 2003). Muhamad, Jarita & Mohd (2011) adopted TRA to investigate factors that influence depositors’ withdrawal behavior in Islamic banks, particularly in Malaysia. A sample of 385 respondents was selected, yielding 365 qualified questionnaires which can be used for analysis. Five components of TRA were measured. Intention and subjective norm were measured by three items scale for each. Attitude and behavioral belief were reflected by seven items scale each and four items were designated to normative beliefs. All five scales achieved good reliability (minimum Cronbach’s alpha = .843). Structural equation modeling technique was used to test the fit of research model. The result showed that almost goodness of fit values are 22 acceptable (NFI=.908, TLI=.926, CFI=.935, RMSEA=.078). The research model fit quite well with data. Bond, Kriesemer, Emborg & Chadha (2007) employed Theory of Planned Behavior (TPB) – an extended version of TRA to examine farmers’ behavior of using pesticides in Jharkhand, India. Questionnaire was designed to measure components of TPB such as intention, attitude, subjective norm and perceived behavioral control. Data was collected from 86 respondents. The multiple regression of intention onto its determinants as attitude, subjective norm and perceived behavioral control was conducted. The result revealed that attitude, subjective norm and perceived behavioral control significantly affect intention with regression weights of .43, .23 and .37, respectively. Attitude affected consumption intention stronger than subjective norm and perceived behavioral control. 4. Research Model and Hypothesis The research model is built based on TRA theory and modifying some parts so that it can describe well situation of UZ45 market. “Evaluation” concept as specified by TRA model was not included in the research model. Following TRA, determinants of attitude are belief-evaluation scores which can be computed by multiplying belief scores by evaluation scores. Then, the linear multiple regression of attitude on those belief-evaluation scores will show which belief-evaluation has significant effect to attitude. Within the context of this research, structural equation model technique will be applied to test the relation among variables instead of multiple regression technique so that multiply of two observed variables is not appropriate because of causing deforming of distribution of observed variables. In this research model, attitude will be affected by salient 23 believes (SBi) denoted by single arrows pointing to it. Each SBi is a latent construct and will be reflected by indicators. Each indicator is an observed variable used as an item in the scale designed to measure that latent construct. Content of items will be showed in next parts. Number of salient believes will be decided after pilot study. Similarly, “motivation to comply” concept also disappears from research model. Subjective norm will be affected directly by normative believes (NBi) from important others and pilot study will indicate the number of normative beliefs. Moreover, according to TRA, attitude of one toward a behavior combines with subjective norm he perceives from social will form intention to perform that behavior. Once behavioral intention is formed, one can perform or not perform that behavior depend on each person. In the case of UZ45, the product has been sold in the market for some years so consumers now is repeating purchase behavior rather than transforming from behavioral intention to real behavior. Thus, examining intention is not needful. The research will examine direct effect of subjective norm and attitude to behavior of consuming UZ45 instead of via Intention. Obtained from pilot (qualitative) study (see chapter four), the research isolated two salient believes and two important others, respectively: “good quality”, “reasonable cost”, “retailer” and “colleague”. As a result, research model is proposed as Figure 9 below. 24 H2 (+) H1 (+) H5 (+) H8 (+) H9 (+) H7 (+) H3 (+) H10 (+) H6 (+) H4 (+) Figure 9: Proposed Research Model As showed in the proposed research model, consumption level, attitude toward consuming the product, subjective norm, salient belief about quality, salient belief about cost, normative belief from retailer, normative belief from colleague are latent variables. Salient beliefs of consumers (farmers) about quality and cost of the product (UZ45) will affect their attitude toward consuming the product, described by two single arrows out from variable “salient belief about quality” and “salient belief about cost” into variable “attitude toward consuming”. Normative believes from retailer and colleague will affect subjective norm perceived by consumers regarding to consumption of UZ45, denoted by the single arrow from variable “normative belief from retailer” and “normative belief from colleague” into variable “subjective norm”. The single arrows pointing to variable “consumption level” from variable “subjective norm” and “attitude toward consuming” stand for the direct effect of attitude of consumers toward consuming UZ45 and their perceived subjective norm 25 regarding to consumption of UZ45 to consumption behavior of UZ45. These relations are adopted from TRA model. Additionally, based on specific extent of UZ45 market, it is proposed additional relations to better describe market situation. Normative belief from colleague is proposed to affect consumers’ belief about quality and cost of UZ45, denoted by single arrows from variable “normative belief from colleague” to variable “salient belief about quality” and “salient belief about cost”. Normative belief from retailer is proposed to affect consumers’ belief about quality and cost of UZ45 indirectly via normative belief from colleague, described by the double arrow connecting variable “normative belief from retailer” and variable “normative belief from colleague”. Finally, perceived pressure from important others (retailer, colleague, family member, marketing staff…) regarding to consume UZ45 is proposed to affect directly consumers’ attitude toward consuming UZ45, as shown by single arrow out from variable “subjective norm” to variable “attitude”. Hypothesis  H1: Consumers’ salient belief about quality of UZ45 (SB_Quality) positively affects their attitude toward consuming UZ45 (AT).  H2: Consumers’ salient belief about cost of UZ45 (SB_Cost) positively affects their attitude toward consuming UZ45 (AT).  H3: Normative belief from colleague (NB_Colleague) positively affects subjective norm (SN).  H4: Normative belief from retailer (NB_Retailer) positively affects subjective norm (SN). 26  H5: Consumers’ attitude toward consuming UZ45 (AT) positively affects their consumption level of UZ45 (CL).  H6: Subjective norm (SN) positively affects consumers’ consumption level of UZ45 (CL).  H7: Normative belief from colleague (NB_Colleague) positively affects consumers’ belief about quality of UZ45 (SB_Quality).  H8: Normative belief from colleague (NB_Colleague) positively affects consumers’ belief about cost of UZ45 (SB_Cost).  H9: Subjective norm (SN) positively affects consumers’ attitude toward consuming UZ45 (AT).  H10: The correlation between normative belief from retailer (NB_retailer) and normative belief from colleague (NB_Colleague) is positive. 27 CHAPTER THREE – RESEARCH DESIGN This chapter describes the way how to test the research model and hypotheses raised in chapter two. It includes description of qualitative research, quantitative research and how to analyze data collected from the survey. 1. Qualitative Research In-depth interview is employed to explore salient behavioral beliefs and normative beliefs of farmers regarding to the behavior of consuming UZ45. The participants are 10 farmers, 3 retailers and 2 marketing specialist in this field. Findings explored from this study are used to develop questionnaire for survey study. See Appendix A for questions protocol in pilot study. 2. Quantitative Research 2.1. Sampling Procedure The sample is drawn from three districts of Lam Dong province: Dalat, Don Duong and Duc Trong, where account for 90% of potato and tomato in Lam Dong province. Cluster and convenient method are applied. For each district, random retailing shops are selected. In each retailing shop, interviewers select randomly consumers (famers) who come to buy pesticides for interviewing. Only farmers who have demand of mancozeb fungicides and have some information about UZ45 are selected, no matter what brand he/she is familiar with. 2.2. Measurement Scales Procedure for developing measurement scale, specifically the first four steps proposed by Churchill (1979) is adopted to develop measurement scale for constructs under research model. Those four steps include: specify domain of construct (what is included and excluded in the construct’s definition); generate sample of items (to 28 develop items which capture the domain as specified); data collection and purify measure. Researchers doing applied work and practitioners could at least be expected to complete the process through step 4 (Churchill, 1979). Regarding to content, all scales used in this research based on guidelines to conduct questionnaire for measuring components of Theory of Planned Behavior model, instructed by Ajzen, in his official website. From the guidelines combined with ideas of consumers and retailers from qualitative survey, 4 items will be developed for each construct then reliability will be tested by Cronbach’s alpha, obtained from data of 74 responds of pilot survey. Coefficient alpha absolutely should be the first measure one calculates to assess the quality of the instrument (Churchill, 1979). Those scales obtaining the alpha higher than .70 should be used for main survey. The overall measurement scale is shown as Table 1 below. Initially, 7 points and 5 points Likert scales both were applied in trial survey, but they faced the problem is that respondents (farmers) can’t be familiar with 7 or 5 points scale and can’t give good answer. After that, we saw that farmers were familiar with 10 points scale as it is the scale they have used in school period. Therefore, 10 points Likert scale was employed for all items. Respondents will be given statements and they will score them from point 1 to 10. The questionnaire mentioned that “The righter the statement is, the higher grade you mark and opposite, the less right the statement is, the lower grade you mark”. It means with a given specific statement, respondents have 10 options to score, from 1 to 10, depend on the magnitude of “right” perceived by them. Consumption level construct is aimed to measure the level to which consumers consume UZ45. In other words, it measures the behavior of consuming 29 Table 1: Overall Measurement Scale Construct Item NB1C 1 Normative belief from colleague NB2C NB3C NB4C 2 Normative belief from retailer NB5R Pesticide retailer told you that you should buy UZ45 when need a mancozeb fungicide. NB6R Pesticide retailer implied that you should spray UZ45 on crops. NB7R Pesticide retailer encourages you to use UZ45. NB8R Pesticide retailer think that you should not buy UZ45. If you buy UZ45, important others like colleagues, retailers, family members…will approve. If you buy UZ45, important others like colleagues, retailers, family members…will agree. If you buy UZ45, important others like colleagues, retailers, family members…will like. If you buy UZ45, important others like colleagues, retailers, family members…will disagree. UZ45's quality is good. SN9 3 Subjective norm SN10 SN11 SN12 SB13Q 4 Salient belief about quality SB14Q SB18C UZ45 control well fungi on crops. UZ45 well dissolve in tank mix and stick well on leaves' surface after spraying. UZ45's quality can be considered as low. UZ45 has reasonable price in comparing with other mancozeb fungicides. UZ45's price is acceptable. SB19C UZ45's price matches its quality. SB20C AT22 UZ45's price is too expensive in comparing with your pocket. According to you, using UZ45 on crops generates very good result. According to you, purchase UZ45 is a right decision. AT23 According to you, spraying UZ45 on crops is a lucid work. AT24 According to you, buying UZ45 can be a wrong decision. In past 1 year, you bought much UZ45 when you need a Manczeb fungicide. In past 1 year, you spray UZ45 with a significant quantity on your crops. SB15Q SB16Q SB17C 5 Salient belief about cost AT21 6 Attitude CL25 CL26 7 Consumption Level Statement Some colleague told you that you should buy UZ45 when need a mancozeb fungicide. You saw some colleagues spray UZ45 on their crops. You knew that some colleagues bought and used UZ45 on their field. Some colleague advised you that you should not buy UZ45. CL27 In past 1 year, the amount of money spent to buy UZ45 was more than amount spent to buy other mancozeb fungicides. CL28 In past 1 year, you used another Mancozeb brand instead of UZ45 when you need a Mancozeb fungicide. 30 UZ45. The pilot study showed that farmers, retailers and marketers are using some different words to illustrate the act of consuming UZ45 such as buying, using and spraying. Based on that information, four items scale (CL25, CL26, CL27 and CL28) is developed to measure the construct “consumption level” as shown in Table 1. Cronbach’s alpha statistic of this scale is .89. This scale is used for the main survey. Attitude construct measure the positive or negative feeling of respondents about the behavior of consuming UZ45. Four items (AT21, AT22, AT23 and AT24) is designed for attitude scale. Cronbach’s alpha for attitude scale obtained from data of 74 responds in pilot survey is .85. This scale is used for main survey. Similarly, salient belief about quality is measured by four items scale (SB13Q, SB14Q, SB15Q and SB16Q) with Cronbach’s alpha value of .84. Salient belief about cost is measured by four items scale (SB17C, SB18C, SB19C and SB20C) with Cronbach’s alpha value of .82. Four items (SN9, SN10, SN11 and SN12) constituted measurement scale of subjective norm with Cronbach’s alpha of .81. Four items (NB1C, NB2C, NB3C and NB4C) formed the measurement scale for normative belief form colleague with Cronbach’s alpha of .85. Finally, normative belief from retailer is measured by four items scale (NB5R, NB6R, NB7R and NB8R) with Cronbach’s alpha of .76. 2.3. Questionnaire The questionnaire includes measurement items as Table 1 and demographic questions. The order of items in the scale is sorted randomly to prevent respondent from bias which usually occurs if we show items of the same subscale continuously. Demographic questions such as consumers’ loyal brand, gender, age, crop, area, income… will be asked. See the questionnaire as appendix B. 31 2.4. Data Analysis Firstly, data collected from the survey is “screened” to detect multivariate normality and outliners. Once data is well prepared, the measurement model is specified accordingly to confirmatory factor analysis (CFA) procedure. Model’s identification is checked by examining number of data points, free estimated parameters and degree of freedom. Construct reliability will be tested by Cronbach’s alpha statistic. Construct convergent validity will be examined by standardize regression weight (factor loading) of items of the same subscales. Construct discriminant validity will be checked by correlation matrix of all latent variables. After all, the overall fitness of measurement model will be evaluated by a series of fit indices: CMIN/DF, GFI (goodness of fit index), TLI (Tucker-Lewis index), CFI (comparing fit index) and RMSEA (root mean square of approximation). Once measurement model is fit, structural model which was specified in the end of chapter two will be tested by structural equation modeling procedure. The overall fit of the model will be accessed by the same fit indices mentioned in measurement model testing part. All estimated regression weights and covariance will be carefully tested by bootstrap procedure to make sure that their standardized estimation interval below 1. Once structural model is fit, hypotheses which postulated in chapter two can be concluded (accepted or rejected) and application of those conclusions will be suggested. 32 CHAPTER FOUR – RESULT 1. Result of Qualitative Study Qualitative study was carried out to explored salient beliefs of consumers about consuming UZ45 and important people who affect consumers in consuming pesticide products like UZ45. Being asked about advantages and disadvantages of consuming UZ45, farmers, retailers and marketing staffs had some similar answers like “it can dissolve very well in water therefore easier for tank mix”; “it can stick steadily on leaf surface”; “it makes leaves greener and thicker”, “it contributes to fungi controlling” and “its price is reasonable”. This result infers that consumers told about two main concepts: quality and cost. Researcher tried to find other believes such as the trade name (reputation) of the company but not be supported because they usually can’t remember company name. It infers that consumers associated “good quality” and “reasonable cost” with UZ45. Those two believes will be specified in the proposed research model to examine their effect to attitude of consumers toward consuming UZ45. In addition, “retailer” and “colleague” are popular answers collected from the respondents by the question asked about people who most affect respondents’ decision on consuming pesticide products like UZ45. Other answers were seen like “family”, “relative” or “sales staffs of companies”, but retailer and colleague are the most. Colleague here refers to famers who live or work nearby the respondents’ place. The result implied that retailer and colleague are people who affect most significantly to consumers’ purchase decision. So, retailer and colleague will be specified to the research model to test their impact on subjective norm. 33 Moreover, famers, retailers and marketers usually illustrate the action of consuming UZ45 by some other words such as “buying” (buying it means consuming it), “using” (using UZ45 on their crops also means that famers consumed UZ45) and “spraying” (spraying UZ45 on crops means that farmers consumed it). 2. Sample Description The survey was carried out and 253 responds was collected in which, 243 questionnaires are valid and used for analysis. This portion describes demographic information of respondents in sample. 2.1. Age Young farmers whose age is lower than 30 account for 19.6% sample. Large portion of farmers are middle-aged who are older than 30 and younger than 51, account for 56% of sample. 2.2. Gender 63% of respondents are male and 37% are female. The rate of female is quite high because in the past, decision on purchasing pesticide was mainly done by male farmers. 2.3. Education Nearly 50% farmers studied in secondary school and 45% studied in high schools before stopping studying. There are 2 cases who graduated from a university in the sample and account for 1%. 2.4. Farm Area There is 85.4% famers in the sample possess lower than 1 ha of farming land and 48% of them have from 0.6 to 1 ha. 34 2.5. Farming Experience 43.5% farmers in sample have worked in farming from 11 to 20 years. The portion of farmer who had more than 6 years of experience is 90.3%. 2.6. Crop 70.5% of farmers from the sample are growing tomato and nearly 17% of them are growing potato. The rate for other crop like flower, vegetable is not much in this sample. 2.7. Selling Price of Farming Products Nearly 83% farmers in the sample said that in this period, they sold farming products at an acceptable price, not low and not high. 13% of them claimed about low price of their products and only 4% thought that they sold their products at high price. 2.8. Payment Term of Purchasing Pesticides Product Nearly 68% farmers are buying pesticide products from retailers by both cash and credit. When they have money, they can buy by cash to get cheap price. It can be in the harvest period when they have product to sell. But sometime in harvest time, but selling price of products is so low, they can’t get money and they have to buy by credit. Of course, they will be charged higher price. But someone, who can well manage their finance, can only buy by cash and always get good price. This group accounts for 13% of sample. And the rest 19% of famers only buy by credit. They will take pesticides from retailer by credit from begin to the end of crop season. They will pay to retailer at the end of season when they sold farming products. They never bought by cash even they are rich farmers. 35 2.9. Weather 48% farmers said that in recent years, the weather is unfavorable because it encourages harmful pests’ and fungi’ development. It means that they have to pay more for pesticide products. 51% said that the weather in recent years is acceptable, not favorable and not unfavorable. 2.10. Income Farmers in the sample are so humble, especially when we ask them about income. 44% of them reported that their income is lower than 100 million VND per year. 27.4% reported the income from 101 to 200 million per year and 3.6% of them said that their income exceed 500 million VND per year. 3. Data Screening 3.1. Multivariate normality Multivariate normality of all observed variables is a standard distribution assumption in many structural equation modeling and factor analysis applications (James, 2011). But it can be difficult to assess all aspects of multivariate normality. Fortunately, many instances of multivariate non-normality are detectable through inspection of univariate distributions (Kline, 2005). Kline (2005) also argued that skew and kurtosis are two ways that a distribution can be non-normal and the best known standardized measures of skew and kurtosis that permit the comparison of different distributions to the normal curve are calculated as follows: S3 Skew index = (S2)3/2 S4 Kurtosis index = (S2)2 36 Where S2, S3, S4 are second, third and fourth moments about the mean: S2 = Σ(XM)2/N, S3 = Σ(X-M)3/N and S4 = Σ(X-M)4/N. Variables with absolute values of the skew index greater than 3.0 seem to be described as “extremely” skewed and the absolute kurtosis index values from about 8.0 to over 20.0 of this index have been described as indicating “extreme” kurtosis. Table 2: Skew and Kurtosis Index Observed Variable NB1C NB2C NB3C NB4C_R NB5R NB6R NB7R NB8R_R SN9 SN10 SN11 SN12_R SB13Q SB14Q SB15Q SB16Q_R SB17C SB18C SB19C SB20C_R AT21 AT22 AT23 AT24_R CL25 CL26 CL27 CL28_R (Source: Data Analysis, N=243) Skew 0.06 -0.11 -0.07 -0.91 -0.41 -0.23 -0.39 -1.15 0.01 0.04 0.20 -0.83 -0.01 0.03 -0.08 -0.55 0.28 0.36 0.19 -0.29 -0.23 -0.12 0.00 -0.84 -0.14 -0.41 -0.15 -0.24 Kurtosis 2.61 2.47 2.28 3.83 2.73 2.10 2.36 3.82 2.23 2.40 2.38 3.03 2.69 2.41 2.67 2.45 2.35 2.16 2.41 2.94 2.14 2.56 2.23 3.01 2.38 2.05 2.01 2.16 37 Adopted from Kline (2005) argument described above, skew and kurtosis indices of all observed variables was computed and showed in Table 2 right above. All absolute skew indices are below 3.0 and all absolute kurtosis indices below 8.0. It means that no evidence of non-normality was detected. 3.2. Outliner 3.2.1. Univariate outliner. Outliners are cases which its values are extreme different from others. As rule of thumb, if a case contains value which lies out of the interval established by the mean minus three times of standard deviation ( ̅ –3sd) and the mean plus three times of standard deviation ̅ +3sd), it can be considered as an outliner. After reviewing all observed variables, there are five variables containing some potential outliners as Table 3. Some values which is smaller than ( ̅–3sd) were detected. Table 3: Observed Variables Contain Potential Outliners Variable Min Value x-3sd Case Normative belief-Colleague 1 1 1.5 107 Normative belief-Colleague 2 2 2.1 103 Normative belief-Colleague 4_R 1 2.1 169 Normative belief-Retail 5 1 1.7 103, 107, 163 Normative belief-Retail 6 2 2.3 163, 225 Normative belief-Retail 7 1 2.7 163 1 1.5 161, 163, 199 Normative belief-Retail 8_R (Source: Data Analysis. N=243.) Going to each variable in detail to examine the outliners, researcher decided appropriate adjustments. For variable “Normative belief-Colleage 1”, the detected 38 potential outliner is case number 107 which received value 1 as a typing mistake then has been corrected. In variable “Normative belief-Colleage 2”, the detected outliner is case number 103 which contained value 2. This respondent said that some colleagues encouraged him to use UZ45 but he has never seen anyone used the product. Thus the respond is “true” with him and his case can be decided as an outliner and will be deleted. For variable “Normative belief-Colleague 4_R”, case number 169 containing value 2 was detected. But it should be kept because his responds are quite consistent in other questions. Case number 163 will be delected because it faced problem with several variables. The rest cases will be kept similarly to case number 169. Totally, two cases are deleted (103 and 163) after univariate outliner accessing. 3.2.2. Multivariate Outliner In order to detect multivariate outliner, Kline (2005) also suggested Mahalanobis distance (D2) statistic, which indicates the distance between a set of scores for an individual case and the sample means for all variables (centroids). A value of D2 with a relatively low p-value (p < .001) may lead to rejection of the null hypothesis that the case comes from the same population as the rest. Through multiple linear regression procedure, Mahalanobis D2 score will be saved as a numeric variable. Its cumulative probability (P) will be computed from cumulative density function (CDF.CHISQ). Our main concern is the p-value can be computed by the formula: 1 – P. Finally, we detect some cases which obtained the small p-value (< .001) as showed in Table 4 below. 39 Table 4: Mutivariate Outliners Case Number D Square P-Value 1 17 85.05 0.000 2 105 78.73 0.000 3 107 83.62 0.000 4 112 78.19 0.000 5 119 66.62 0.000 6 154 71.16 0.000 7 161 79.84 0.000 8 213 71.49 0.000 Source: Data Analysis It means the null hypothesis that those cases come from the same population with the rest cases has been rejected. They should be treated as outliners and be deleted. After multivariate outliner accessing, more 8 cases was deleted: 17, 105, 107, 112, 119, 154, and 213. The sample remains 243 cases after accessing outliner. 4. Measurement Model Measurement model was tested by confirmatory factor analysis procedure. 4.1. Specification The specification of a measurement model reflects hypotheses about the correspondence among observed variables (indicators), factors [latent variables] that represent hypothetical constructs, and measurement errors (Kline, 2005). As showed in Figure 10, the measurement model consists of 7 latent variables (construct) with associated indicators and error terms. 40 Figure 10: Measurement Model Latent variables, so called constructs are denoted by ellipses. Those constructs were adopted from TRA model and the relation among them was hypothesized in proposed model in chapter two. But in this measurement model, we don’t care about that relation among constructs. Therefore, all constructs are specified to correlate together, denoted by double headed arrow connecting between each two constructs. 41 Rectangles in graph stand for observed variables, so called “indicator”, which were collected via survey. They are designated to reflect the construct which has a single headed arrow pointing to them. It is expected that large portion of variance observed variables will be explained by its construct. The portion of variance which can’t be explained by the construct will be supposed to be explained by others source and called error terms, denoted by circles which has an arrow pointing to observed variables. Initially, researcher design 4 indicators to reflect each construct. But after re-specifying, only construct “consumption level” remains three indicators while the rest constructs remain two for each. 4.2. Re-specifying Record The measurement scale developed in chapter three was used to measure all constructs. 243 responds were received. First step, all sub-scales were tested reliability again, although when developing the measurement scale, we tested it by the data of 74 responds collected from the pilot survey. Those items which had low item-total correlation should be deleted. It included: NB4C_Recoded (.30), NB8R_Recoded (.44), SN12_Recoded (.18), SB16Q_Recoded (.35), SB20C_R (.39) and AT24_Recoded (.43). Then, the remained items were assigned to their construct in the measurement model then tested according to CFA procedure. Follow CFA procedure, some items were eliminated to fit the model. To satisfy discriminant validity requirement, correlation between constructs should be lower than .85 (Kline 2005). SB15Q was eliminated to reduce correlation between AT and SB_Quality from 1.01 to .90. Then, AT21 was eliminated to reduce correlation between AT and SB_Quality from .90 to 42 .82. At this point, all correlation estimates were lower than .85 and their bootstrap (500 samples) 95% confident intervals were lower than 1. Next, in order to achieve goodness of fit statistics of measurement model, some items recommended by modification indices will be eliminated. CL28_R was eliminated to make GFI of measurement model increase from .89 to .90. Up to this point, measurement model was fit well with data (CMIN = 254, CMIN/DF = 1.9, GFI = .90, TLI = .94, CFI = .95, RMSEA = .06). After that, remained items were assigned to their construct in structural model. Modification indices were based on to fit the model again. SB17C was eliminated to increase GFI of structural model from .872 to .877. SN 11 was eliminated to increase GFI of structural model from .877 to .883. NB6R was eliminated to increase GFI of structural model from .883 to .893. Finally, NB2C was eliminated d to increase GFI of structural model from .893 to .905. Structural model was fit well with data (CMIN = 187, CMIN/DF = 2.34, GFI = .91, TLI = .94, CFI = .95, RMSEA = .07). 4.3. Identification According to the measurement model in Figure 10, we have 15 observed variables. So, we will have 120 data points (15*16/2). There are totally 51 free parameters to be estimated consist of 15 variance of error terms, 7 variance of constructs, 8 regression weights of indicators on its construct, not include indicators which has been assigned constraint variance and 21 covariance among 7 constructs. It mean we have 69 degree of freedom (120 – 51). Additionally, every construct has a scale and each scale has at least two indicators. Thus, both necessary and sufficient condition for identification of the CFA model is met. The model is over-identified. 43 4.4. Construct Reliability and Validity 4.4.1. Reliability The most commonly reported estimate of reliability is Cronbach’s coefficient alpha which measures internal consistency reliability, the degree to which responses are consistent across the items within a single measure. Generally, reliability coefficients around .90 are considered “excellent,” values around .80 are “very good” and value around .70 are “adequate.” (Kline, 2005). Table 5 below shows reliability of individual scales as Cronbach’s alpha statistics. All alpha values are higher than .70. It means that scores collected from items on the same scale have internal consistent and the all individual scales are reliable. Table 5: Reliability of Scales Items Cronbach’s Alpha 1 Normative belief from colleague 2 .75 2 Normative belief from retailer 2 .79 3 Subjective Norm 2 .88 4 Salient belief about quality 2 .72 5 Salient belief about cost 2 .84 6 Attitude 2 .73 7 Consumption level 3 .89 Construct Source: Data analysis, N = 243. 4.4.2. Validity Construct validity has two main facets including convergent validity and discriminant validity. Each construct must be reflected by a set of indicators. For each set of indicators, the standardized factor loadings are all relatively high, which 44 suggests convergent validity and poor discriminant validity as evidenced by very high factor correlations (Kline, 2005). According to common rules of thumb, 0.60 is considered as relatively high for factor loading and 0.85 is thought as too high for factors’ correlation. Table 6 below shows estimated factor loading of all indicators on their constructs. All of them are higher than .60. Bootstrap (500 samples) 95% interval confidence of these factor loadings showed that almost of them are higher than 0.6 and lower than 1. Thus, convergent validity is supported. Table 6: Standardized Regression Weights (Factor Loadings) Parameter Estimate Lower Upper NB3C [...]... Fishbein and Aijen in 1975 is adopted for examining factors influencing farmers purchase behavior in case of UZ45 in Lam Dong province, because of following reasons Firstly, theory of reasoned action is based on the assumption that human being usually behave in sensible manner; that they take account of information and implicitly or explicitly consider the implications of their actions (Ajzen, 1985) In. .. structural marketing plan so that they can allocate accurately marketing resources to maintain and increase the 5 product’s sales A deep insight of consumer purchase behavior accounts for a crucial role in this mission 3 Research Objective The research is aimed to find out main factors influencing farmers purchase behavior in the case of product UZ45 in Lam Dong province and suggest marketing strategy... Research Findings from this research can be generalized to explain purchase behavior of famers in Lam Dong province and in case of UZ45 product only 6 CHAPTER TWO – LITERATURE REVIEW This chapter is assigned to present available theories about consumer behavior and the reasons for selecting the most appropriate one to adopt for forming theoretical framework of this research The background of selected theory... control behavior The theory of reasoned action is designed to predict volitional behaviors of this kind (Ajzen, 1985) Therefore, the selection of TRA is appropriate Thirdly, TRA is more appropriate than Theory of Planned Behavior (TPB) The co-author of the theory, Ajzen, stated that “TRA is a special case of the TPB The only difference between the two theories is that the TPB includes perceived behavioral... develop the product’s sales 4 Significance of the Research Knowing main factors influencing consumers’ purchase behavior will help managers to allocate rightly marketing efforts to maintain and grow sales of the product For instance, if finally the research finds that consumers’ buying behavior is affected more strongly by normative belief from important others rather than his own beliefs, the company’s... additional determinant of intentions and behavior In the development of the TRA it was assumed that people have volitional control over the behavior of interest Under these conditions, perceived behavioral control becomes irrelevant and the theory of planned behavior reduces to the theory of reasoned action” Fourthly, as described above, the behavior under interest is a very simple action (consuming an agrochemical... marketing communication campaign should set his important others as targeted audiences instead of himself 5 Research Scope This research is carried out within Lam Dong province where accounts for 80% sales of UZ45 in Vietnam market The population is defined as a set of farmers who have demand of Mancozeb products, within Lam Dong province and no matter what brand he/she is familiar with 6 Limitation of the. .. explain volitional behaviors and its scope excludes wide range of behavior such as those that are spontaneous, impulsive, habitual…because their performance might not be voluntary To expand the range of behaviors encompassed by the TRA, Ajzen (1985) proposed the theory of planned behavior (Jerold et al, 2002) 1.2 Theory of Planned Behavior The theory of planned behavior (TPB) is an extension of the theory... TRA Model 2.1.1 Behavior A key assumption in TRA is that behavior concept, which is the main interest in the model, is under volitional control As Ajzen (2005) defined, under volitional control means that people can easily perform these behaviors if they are so inclined, or refrain from performing them if they decide against it Many behaviors in everyday life can be thought of as being largely under... with that individual (group) within seven point Likert scale 3 Previous Studies Theory of Reasoned Action (TRA) has been using in several fields to examine many kinds of behaviors 21 McCarthy, Boer, O’Reilly & Cotter (2002) adopted TRA to examine factors influencing consumers’ behavior on consuming beef in Irish market The authors isolated six behavioral beliefs and four influence groups Behavioral .. .FACTORS INFLUENCING FARMERS' PURCHASE BEHAVIOR – THE CASE OF UZ45 IN LAM DONG PROVINCE In Partial Fulfillment of the Requirements of the Degree of MASTER OF BUSINESS ADMINISTRATION In Marketing... adopted for examining factors influencing farmers’ purchase behavior in case of UZ45 in Lam Dong province, because of following reasons Firstly, theory of reasoned action is based on the assumption... research is aimed to find out main factors influencing farmers’ purchase behavior in the case of product UZ45 in Lam Dong province and suggest marketing strategy to managers to develop the product’s