主觀知識和客觀知識如何影響判斷修正量

全文

(1). 國立臺灣師範大學管理學院管理研究所碩士論文 Graduate Institute of Management College of Management National Taiwan Normal University Master Thesis. 主觀知識和客觀知識如何影響判斷修正量 How should Subjective Knowledge and Objective Knowledge Influence the Amount of Judgmental Correction. 林小鳳, Puntarik Panasawatwong 指導教授：蕭中強博士 Advisor：Chung-Chiang Hsiao, Ph.D. 中華民國107年7月July, 2018.

(2) . ABSTRACT. The Elaboration Likelihood Model (ELM) has been introduced in marketing researches for describing advertising persuasion since the 1980s. ELM defined that there are variables that resolve people’s emotion and have the effect on his or her evaluation. This study will apply the flexible correction to examine how the contrast degrees between subjective knowledge and objective knowledge affect attitude confidence and judgmental correction. In the present research, we define high involvement in all our participants and manipulate participants into two major groups. Half of the participants will be served as high subjective knowledge with low objective knowledge, and the remaining participants will be served as low subjective knowledge with high objective knowledge. The participants in this experiment are Thai people, and we collected the data from an online survey. The results of these analyses showed that participants with low subjective knowledge have more correction on attractive endorser than participants with high subjective knowledge. The attractive endorser has more effect when participants are served as a high subjective knowledge. Key words: Elaboration Likelihood Model, Flexible Correction Model, Bias Correction, Subjective Knowledge, Objective Knowledge, judgmental attitude, evaluation. . I.

(3) . Acknowledgment The day I decided to do my Master Degree in Taiwan, I knew it is not easy for me. Many “Can I … or not” questions in my mind: “Can I do it or not, Can I pass it or not, or Can I graduate on time or not.” These two years are the most important and challenging step in my life. It is not easy to get it. “If you think you can, you can. And if you think you can’t, you’re right.” – Henry Ford. Now, I can answer all the questions in my mind that “Yes, I can!” I have to thank for the fate that made me choose this route and brought me here to accomplish my dream in Taiwan. This achievement is not only owned to me, but it is also yours – all the persons who always supported me go through this process. I would like to express my heartfelt appreciation to my thesis advisor, Assoc. Prof. Chung-Chiang, Hsiao, for his support, advice and invaluable assistance during this thesis process and during my two years in this institute. I would not have achieved my goal, and my thesis would not have been completed if without all the support that I received from him. I thank my committee members very sincerely for your suggestion on my study. Besides, I wish to express my appreciation to my two previous chiefs, Asst. Prof. Thanawut Naigowit and Mr. Robert Wu for your advice during my thesis process and before I did my Master Degree. My friends here in Taiwan and those in Thailand, thanks for listening to me and encouraging me. Lastly, I most gratefully thank my family for their love and support. It is not an easy journey, but with your love, your faith and your support that helped me go through this part. Thanks to my three brothers who always encouraged me and believed in this sister. All of you. II.

(4) have always been there for me, encouraging me and pushing me not to give up. Words cannot express how proud I am of being your daughter and your sister, I love you and thank you so much!. Puntarik Panasawatwong. III.

(5) . Table of Contents CHAPTER 1 INTRODUCTION…………………………………………………1 Motivations and Study Purpose……………………………………………...1 CHAPTER 2 LITERATURE REVIEWS…………………...…….………….....4 Elaboration Likelihood Model (ELM)………………………...………….....4 The Influence of Confidence………………………………...…...….………5 Self – Validation Hypothesis……………………….………………...6 Correction Models………………………………………..………….………7 Set – Reset Model……………………………………………...……..7 Inclusion – Exclusion Model…………………………………...…….8 Flexible Correction Model (FCM)……………………………...…….8 Objective knowledge and Subjective knowledge……………………...…….9 Knowledge miscalibration………………………………………..…..9 CHAPTER 3 PROPOSED THEORY AND HYPOTHESES…................……11 Proposed Theory……………………………………………………...……11 Hypotheses…………………………………………………………………12 CHAPTER 4 RESEARCH METHODOLOGY……………………………….13 Overview………………………………………………………………..….13. IV.

(6) . Main Experiment…………………………………………………….……..13 Participants and Design…………………………………………...…13 Experiment Procedure……………………………………………….14 Independent Variables……………………………………………….16 Dependent Variables……………………………...…………………17 CHAPTHER 5 RESULTS…………………………………………..…………..19 Manipulation Checks………………………………………….……………19 Confidence…………………………………………………………..19. 20 Involvement…………………………………………………………19 Endorser……………………………………………………………..20 Argument Quality………………………………………….………..21 Objective Knowledge………………………………………………..22 Dependent Measure…………………………………………………...……22 Attitude…………………………………………………………...….22 CHAPTHER 6 GENERAL DISCUSSION………………………...…………..26 Conclusion………………………………………………………………….26. 27 Academic Contribution and Managerial Implication………………………37 Limitations and Future Research……………………………………….…..28 REFERENCE…………………………………………………….………………29. V.

(7) . List of Tables. Table 1. Experimental Design ……………………………………………………..……13 Table 2. ANOVA Output on participants’ Rating of Confidence………………....19 Table 3. ANOVA Output on participants’ Rating of Involvement………………..…20 Table 4. ANOVA Output on participants’ Rating of Endorser Attractive………….21 Table 5. ANOVA Output on participants’ Rating of Argument Quality………..…..22 Table 6. ANOVA Output on participants’ Rating of Objective Knowledge………..22 Table 7. ANOVA Output on participants’ Rating of Attitude………………………..23 Table 8. Syntax Output of simple effect test……………………………………………25. VI.

(8) . List of Figures Figure 1. Mean Rating of Target Attitude (Low subjective knowledge)…...…….24 Figure 2.Mean Rating of Target Attitude (High subjective knowledge) …….…..25. VII.

(9) . CHAPTER 1 INTRODUCTION. Motivations and Study Purpose Social psychologists have studied the changes in behavior for a long time, and the influence of this study has been the main idea for many applied psychological fields. In classical social psychology study, Brinol and Petty (2009) explained “the persuasion is possible when a person or a group of people (i.e., the recipient) receives an intervention (e.g., a persuasive message) from another individual or group (i.e., the source) in a particular setting (i.e., the context)” (p. 71). The persuasion will succeed while an attitude influences to recipient’s thought direction. Persuasion has been a fundamental factor in human behaviors. Many studies developed the theories of persuasions. The elaboration likelihood model has described many ways that persuasion could happen when people assume the target with attentively or thoughtlessly. Besides, this process and result are also not the same, it depends on the variables, for instance, message, source, context or recipient that effect in each situation. The elaboration likelihood model (ELM) also indicated that a variable could determine a person’s related emotions and influence his or her evaluative judgment (Brinol & Petty, 2009). An individual's level of knowledge can be one of a variable that influences one's judgment or evaluation. Previous researchers (see Johnson & Russo, 1984) explained that when consumers (i.e., recipients) have more knowledge about the product, they would have more efficiently to search the information about that product while decreasing their cognitive resources to get that related information (Alba & Hutchinson, 1987).. 1.

(10) Conversely, this process will lead consumers to have more confident in their ability to make an appropriate judgment on that product (Brucks, 1985). In consumption research, consumers’ actual knowledge is separated into two categories based on their knowledge’s evaluation (e.g., Alba & Hutchinson 1987, 2000; Brucks, 1985). Objective knowledge and Subjective knowledge have a clear distinction in using its relevant constructs. Objective knowledge related to ability and special skills that each person had learned from their experience, while Subjective knowledge refers to the consumers’ confidence in their understanding for their judgments. (Alba & Hutchinson, 1987; Bearden, Hardesty, & Rose, 2001) The phenomena of correction have been studied in many research fields containing context and priming effects, judgment, impression structure and attribution. However, there are only a few researches that focus on the specific causes of correction. Most of the researches had focused on the influence of the unrelated contexts in social judgments. There are many variable biasing factors had been found when studying about bias correction. For instance, the effects of specific situations such as native culture, media, and social norms, or the individual effect such as motivation, emotion, and confidence but there is only some researches’ attention that tries to find or study about the processes when the biases have an attempt toward the target. In other words, many pieces of research do not focus on the process how the bias has the effect or when the people have a bias, how and when they try to eliminate this bias form their perception and evaluation (Wegener & Petty, 1997).. 2.

(11) Briefly, the corrections are the outcome of people perspective’s result by using their native theories. That is, there is a more flexible form of corrections because the corrections are depended on people’s native theories. Recently, many experimental researches studied the bias corrections that impact of unrelated contexts on social judgments (e.g., Martin, 1986). However, there are a few studies focused on the influence of bias corrections toward the product’s evaluation or judgment. Many relevant researchers had studied how objective knowledge and subjective knowledge affect in different consequence toward decision-making or information search (e.g., Hadar, Sood, & Fox, 2013). Also, the previous studies (e.g., Hadar, Sood, & Fox, 2013) researched only one kind of the consumer knowledge or focus on the impact of knowledge on searching behavior. This study will focus on the contrast level between objective knowledge and subjective knowledge that affect attitude confidence and judgmental correction.. 3.

(12) . CHAPTER 2 LITERATURE REVIEW. Elaboration Likelihood Model (ELM) The idea of "two routes to persuasion" was introduced several years ago, in the role of the Elaboration Likelihood Model or ELM. The principle concept of ELM is about the processes that focus on the attitude changes toward the perceptions of objects; for every variable will have the influence to the processes in a different level of perceptions resulting and evaluative judgments (Petty & Cacioppo, 1986a). The Elaboration Likelihood Model, it determined that there are two routes of the perception process – central route and peripheral route that depended on how does effortful processing about the products. Central route to persuasion refers to attitude changes with high-effort cognitive to process product information. Under the central route, the information process activity related extensively and effortful to try to assess and scrutinize the central merits of the supportable situation. With the low-effort of attitude changed is presented by the peripheral route to persuasion. Under a peripheral route, the process depended on a capability that less cognitive method than the central route. In the conceptualization of ELM suggests that people probably accept the information that based on their existing knowledge and evaluates them superficially when the elaboration likelihood decreases. The resultant of this process is lacked the careful consideration of relevant information; it is formed on (a) the result that affiliated with positive or negative cues and unrelated to the attitude stimulus or (b) the implication based on different cues in persuasion context. Contrarily, in case that the elaboration. 4.

(13) likelihood is high, the acceptable or rejectable from the cognitive method is predicted (a) to be relative with the issue-relevant thinking and more likely to related to their earlier experience and knowledge about that object; (b) the recipients will have more confidence and willingness to process over their attitudes. (Cacioppo & Petty, 1984). In Elaboration Likelihood Model also described the elaboration continuum that is defined by how motivation and ability estimate the central merits of attitude object as shown in figure 1. There are not only two points – central route and peripheral route - in the model, but there are also the continuums from each point. At the upper end of the central route continuum, the attitude change and the elaboration get higher than the continuum of a peripheral route. It means that the recipients, who have a cognitive method through central route, they will have more effect from attitude change and have more responsibility than another who process through peripheral route. The influence of confidence Confidence is a critical mainly factor for people to convert their thinking into evaluations or judgments. It can be representative of a sense of one's beliefs and thinking. (Petty, Brinol, &Tormala, 2002; Wegener, 2007). Moreover, the confidence in people's thoughts also can be the primary factor that impacts people's persuasion. (Petty, Brinol, & Tormala, 2002). Brucks (1985) described that the self- confidence could measure the level of subjective knowledge that is it high or low. Besides, when people receive some information, self-confidence may influence their information perceiving and their evaluation. For instance, compare with the actual knowledge level, people with a. 5.

(14) shortage of confidence may have more motivated to search the related information that people with higher of confidence. Self-validation Hypothesis. Petty, Brinol, & Tormala (2002) suggested that there are only thoughts that people have the confidence to accurate and form their attitude. The more holding of confident though, the more effect on their judgments. The main idea of Self-validation Hypothesis is that thought confidence is a relevant cause of thought predicts attitudes, while attitude confidence is a relevant cause of attitudes predicts behavior (Petty, Brinol, &Tormala, 2002; Wegener, 2007). Gross, Holtz, and Miller (1995) recommended that this self-validation hypothesis concept between thought confidence and attitude have related to the concept of attitude confidence and behavior. Petty, Brinol, &Tormala (2002) explained the attitude confidence as "a subjective sense of conviction or validity regarding one's attitude" (Page 560). In meta-cognitive also mentioned that in one's thoughts are composed of the degree of confidence for estimate their thought and the degree of confidence will be ranged from the highest assuredly to the lowest (e.g., highest doubt). Hence, the more confidence in that thought, the more effect on that judgment compare in the same perception. (Brinol & Petty, 2009). In additional, Self-validation Hypothesis indicated that generating thoughts is not enough to affect their judgments, but they need to have the confidence in that perception. Moreover, the amount and the valence of thought are also the critical factors to determine thought confidence (Petty et al., 2002).. 6.

(15) Correction Model Set-Reset Model. Initially suggested by Martin in 1986 and later on developed by a number of the researcher, the Set-Reset model becoming one of the fundamental models for correction cognition as well as consumer research studies. Martin claimed that both negative and positive element might add in a representation of the target. In the model, Martin (1986) argues that augmented processing volume can affect consumers' efficiency to assess the correctness of contextual message. Furthermore, following to the Set-Reset model, it described that when the users (consumers) have the comment about that understanding, they will assess both of their motivation and that factors ("setting"). On the other hand, if the person recognizes that context is as the influence to affect their reception and they have the confidence on it, it will have a bias toward their motivation and will reset their decision from that factor ("reset") (Martin, 1986). When the individual resets his/her judgment, there are some factors that they believe that it can be identifiable in that context. There are many opportunities occurred that the evaluator could not evaluate the impact of contextual consciously, so they generated the contrast response. Forehand and Perkins (2005) explained set/reset model in their study with an example that is easier to understand as follows "if an evaluator believes that pleasant weather (a contextual cue) is biasing their evaluation of a new acquaintance, they may estimate how the weather prompts much positive effect and adjust their evaluation of the acquaintance accordingly. If the evaluator overestimates the degree to which positive effect is due solely to the weather, the net effect of resetting is a less positive evaluation. 7.

(16) of the acquaintance than is objectively warranted" (p. 436). Correspondingly, the overestimate of the negative influences the result of the contextual cue and affect more favorable evaluation than justified. Inclusion-Exclusion Model. Schwarz and Bless presented the InclusionExclusion Model (IEM) in 1992 and become one of the prominent models in context effects to explain assimilation (positive relationship) and contrast (negative) (Bless and Schwarz, 2010). The model claims that individuals have to form two mental representations to assess the target's evaluation: the first one refers to the target stimulus and the other representation of a standard of comparison to evaluate the target stimulus. In other words, when the person is in a precise moment, and the information arises to concentration in precisely that moment appeals special consideration. Reliantly on how individuals classify the information, the same access information can outcome in assimilation or contrast effects. The assimilation effects are likely to occur when the accessible information builds the representation of the target, conversely, the contrast effects likely to occur when accessing information used to construct the standard of comparison. Flexible Correction Model (FCM). The model introduced by Petty and Wegener (1993) called Flexible correction model, focuses more on how the correction works and addressing more into correction view. This model argues that if there are alternatives and adjustments of targets, and if the judgments are motivated to change, flexible correction occurs that depends on their naïve theories of how the context influences the judgment of the target. Briefly, the flexible correction model (FCM) depends on individuals' use of their own biases' understanding to effort in a hypothetical judgment situation.. 8.

(17) Basically, the FCM explained that corrections are intended to remove the bias of the individual considers are related to the factors. The FMC assumes that when people believe that there is a bias making their judgments of an object too similar to the context (aware of assimilation effect), they will try to remove it, that is, less like the context. On the contrary, when people convince of some biasing factors rendering their judgments away from the context (aware of contrast effect), they would revise their judgments to be more like the context. Objective knowledge and Subjective knowledge In the consumer context, based on individual knowledge's assessment, consumers' knowledge can be separated into two categories (e.g., Alba & Hutchinson 1987, 2000; Brucks, 1985). Objective knowledge refers to the product-related information, which is stored in long-term memory and can be retrieved with accurately and validly. Many researchers indicated that objective knowledge affects to consumer's motivation and the ability for searching (Brucks, 1985), processing and evaluating the related information. On the other hand, subjective knowledge refers to consumers' confidence about product-related information that they keep in their memory (Brucks, 1985). Subjective knowledge also influences service quality assessment and searching method for the product. Knowledge miscalibration. Knowledge miscalibration is a concept that presents the argument between subjective and objective knowledge. It shows how the error occurred on subjective knowledge. The knowledge miscalibration can be formed into two types: overconfidence and underconfidence. Alba and Hutchinson (2000) noted that a consumer with overconfidence tends to have the less suitable effort for searching the. 9.

(18) information; they also think that they knew it already and did not have the motivation to search more information. Conversely, underconfidence consumer has over an effort to collect the information. So they think their existent knowledge is not enough for processing this action.. 10.

(19) . CHAPTER 3 PROPOSED THEORY AND HYPOTHESE. Proposed theory According to the Flexible Correction Model concept, when people attempt to “correct” or make a judgment, they often use their own naïve theories of bias to generate it (Chien, Wegener, Hsiao, & Petty, 2014). However, the critical idea of FCM is that the corrections are driven for generating the corrected or uncorrected of biases that people believe that they have collaborated with other factors. That is, people may behave not enough ability to evaluate or adjust their perceptions because of their confidence or their knowledge. To attempt to present the Flexible Correction Model into a framework, we determined that if the levels of subjective knowledge and objective knowledge influence one’s attitude confidence and, in turn, that confidence may affect the judgmental correction. We wonder what the people with high subjective knowledge and low objective knowledge (i.e., subjective expert) may have more confidence than the people with low subjective knowledge and high objective knowledge (i.e., subjective novice). With their higher confidence will tend to affect their judgment and may tend to perceive the judgmental bias correction. In this study, we focus on one’s subjective knowledge and objective knowledge with their attitude confidence that influence their evaluation. That is, people correct their perceived bias with their own naïve theories (i.e., subjective knowledge and objective knowledge). Their consideration creates the level of confidence. 11.

(20) and their adjusted judgments to the target. After that, we want to examine that the condition of correction (bias of attitudinal confidence) whether or not affect the bias of judgmental correction. Hypothesis In order to investigate how much subjective knowledge and objective knowledge affect consumers’ attitudinal judgment. We designed two hypotheses for our experimental as below. Hypothesis 1: People with high objective knowledge and low subjective knowledge are more likely to correct for bias. In this hypothesis, we expect that people with high objective knowledge and low subjective knowledge are not confident on their product judgments, and therefore, they would be more likely to correct for the identified bias. Hypothesis 2: People with low objective knowledge and high subjective knowledge are less likely to correct for bias. In this hypothesis, we assume that people with low objective knowledge and high subjective knowledge are confident on their product judgments, and therefore, they would be less likely to correct for the identified bias.. 12.

(21) . CHAPTER 4 RESEARCH METHODOLOGY. Overview This study was managed by an experimental design to examine whether or not Subjective knowledge influences the consumers’ evaluation. Besides, how the subjective knowledge impacts the consumers’ confidence in assessing the product. To examine these hypotheses, we conducted the questionnaire into three parts. First, Bias reminder to remind the participants to regard to the bias in the target product. Secondly, Subjective knowledge to manipulate the participants to get the high subjective knowledge with low objective knowledge or low subjective knowledge with high objective knowledge, after this part the participant will be conducted to adjust their confident level toward the target product. Lastly, Endorse to examine participants’ attitude toward the endorser and how was the effect on consumers’ evaluation. In the experiment, we collected the data randomly by sharing the link of the online survey platform to the participants. There will be eight different questionnaires in our experiment to examine the different variables. The participant will be asked to select the questionnaire following on their birthday. We expected to collect the 15 participants for each group to do our analysis. Main Experiment Participants and Design. Since our experiment designed to examine Thai people, all of our data will be collected by the online platform. The Jotform.com website was used to create our online questionnaires.. 13.

(22) We expected to have at least 180 Thai males and females to participant in this experiment. There were eight different questionnaires and participants were asked to do for each questionnaire by randomly using participants’ birthday. For each questionnaire, we designed each questionnaire by following to the cells in a 2 (Bias Reminder: bias reminder or without bias reminder) x 2 (Subjective knowledge: high or low) x 2 (Endorser: attractive or average). Our questionnaires’ design is presented in Table 1.. Table 1. Experimental Design Variables. Level. Bias Reminder. Have. Without. Subjective knowledge. High. Low. Endorser. Attractive. Average. Experimental Procedure. In our experiment, after the participant selected the questionnaire according to their birthday, the first page of the questionnaire will be the survey introduction explained that our study researched on consumers’ cognitive capability and reading comprehension on the general media information. Also, there was the notification that the participants cannot turn back to the previous page so the participants should read the articles and do the questions on each page carefully. The next page was Bias reminder advertisement. Half participants discovered the Bias reminder advertisement while the other participants discovered the advertisement. 14.

(23) without Bias reminder. After this page, there were five questions regarding the product quality, celebrity endorsement and purchase intention to evaluate the target product. In this experimentation, we would like to examine that how the biases affect participants’ evaluation. In the Bias reminder advertisement, there were some articles about the bias (e.g., costing for endorser and packaging) to remind the participants. Participants encountered with Bias reminder advertisement were recalled about the biases in these articles, and they would have more attention on the product while the other would not. After Bias reminder’s experiment stage, the participants would read the article about the daily product. The half of participants got the article related to our knowledge test on the next page to serve as high objective knowledge. After the knowledge test, there was a message informed participants that they got only two correct answers from 9 questions for manipulating our participants to have a low subjective knowledge. On the other hand, the reminding participants would get the article unrelated to following knowledge test to serve as low objective knowledge and got the message reported that they got seven correct answers from 9 questions for controlling our participants to get a high subjective knowledge. For this process, we would like to manipulate the half participants to serve as a high objective knowledge with a low subjective knowledge and the others serve as a low objective knowledge with high subjective knowledge. Then, there was the advertisement to lead the participants to have a high involvement with our target product, after that the endorser for the target product’s advertisement will be introduced. The half of participants encountered with attractive endorser while the others encountered with the average endorser. Finally, participants will be asked to answer the questions about the target product. In this last experiment. 15.

(24) questionnaire, we would like to examine participants’ evaluation of our product target with different endorser and different attitude on subjective knowledge and objective knowledge. Independent Variables Endorser. In our experiment, there were two endorsers in the target advertisement, first endorse is a famous actress in Thailand Toey - Jarinporn and the second endorse is an unknown college student. Both of these two endorses will appear in the same kind of picture with the target product and some description sentence to promote the target product. Involvement. To serve as the high involvement, the article in the target advertisement that introduced the brand of target product had the plan to expand its market to South East Asian, and it selected Thailand to be the first market to sell this product in next two weeks. Bias Reminder. There were two kinds of target advertisement in this part of our experiment. The first one was the advertisement of PP supermarket with the description that the products in this store have a low price because the prices are not included the celebrity endorsement cost or the packaging cost to remind the participants to regard to these biases. The second advertisement was also the advertisement for PP supermarket, but there was not any bias reminder on the advertisement. Subjective knowledge. Half of the participants will be manipulated to have high subjective knowledge but low objective knowledge, they will be asked to read the daily article that does not relate to the nine questions on the next page. On the other hand, the other participants will be controlled to have low subjective knowledge but high objective. 16.

(25) knowledge, they will be encountered in the daily articles that related to the nine questions on the next page. Dependent Variable Attitude. The questions regard to the attitude toward the target product were asked after all the participants encountered the target advertisement. Five 7-scale items from 1 (very poor quality/ very bad/ very negative / dislike very much/ low purchase intention) to 7 (very good quality/ very good/ very positive/ like very much/ high purchase intention) were used to measure participants’ attitude toward the target. The average of these five items served as the index to indicate the participants’ rating on attitude (Cronbach’s Alpha = .91). Confidence. Participants’ confident levels were assessed by using three 7-scale items from 1 (very unconfident/ very uncertain/ very unsure) to 7 (very confident/ very certain/ very sure). The coefficient of Cronbach’s Alpha = .92 showed high consistency. These three items were average to serve as the index of participants’ rating on confidence. Involvement. Three 7-scales questions from 1 ( not involved at all/ not very serious/ very inattentive) to 7 (highly involved/ very serious/ very attentive) were applied to check participants’ involvement toward the target. The rating on there three questions were highly consistency ( Cronbach’s Alpha = .93). Thus, the average of these three questions served as the index to indicate the participants’ rating on involvement. Endorser. Attitude toward the endorse was measured by two 7-scales items from 1 (very negative/ dislike very much) to 7 ( very positive/ like very much). The average of. 17.

(26) these two items served as the index to indicate participants’ rating on attitude toward this celebrity (Cronbach’s Alpha = .85). Argument Quality. Participants’ perception about the product was assessed by using two 7-scales items from 1 (very useless/ not persuasive) to 7 (very useful/ very persuasive). The coefficient of Cronbach’s Alpha = .87 showed high consistency. These two items were average to serve as the index of participants’ rating on argument quality.. 18.

(27) . CHAPTER 5 RESULTS. Manipulation Checks Confidence. In the 2 by 2 by 2 ANOVA on participants ‘ ratings of confidence, there were a significant main effect of Endorser and a partially significant three-way interaction. The other effects were not significant as shown in Table 2. Participants encountering the celebrity endorser in the target advertisement reported higher confidence (M = 4.47) than participants encountering the average endorser in the target advertisement (M = 4.11), F (1,159) = 5.67, p < .05. The three-way interaction approached significant, F (1,159) = 3.92, p = .05. . Table 2. ANOVA Output on participants’ Rating of confidence Source Endorer Bias_reminder* Subjective_knowledge* Endorser Error. Sum of Squares 6.379 4.393. df. F. Sig.. 1 1. Mean Square 6.379 4.393. 5.688 3.917. .018 .050. 178.308. 159. 1.121. The manipulation of subjective knowledge did not influence confidence ratings. It was likely that immediately after the manipulation of subjective knowledge, participants’ confidence on dairy knowledge did differ as expected, but after the rating of attitude (i.e., after the correction), participants’ confidence ratings. 19.

(28) might be enhanced. It is therefore suggested that confidence rating can be measured immediately after the manipulation of subjective knowledge Involvement. To check participants’ involvement, there were main effect of Endorser and two-way interactions were significant, as shown in Table 3. The celebrity endorser in the target advertisement led the participants to higher involvement (M = 4.21) than the average endorser in the target advertisement (M= 3.89, F (1, 159) = 4.10, p < .05). The significant Bias Reminder x Endorser interaction, F (1, 159) = 5.52, p < .05 showed that celebrity endorser heightened participants’ involvement only when the bias reminder was not presented in the PP supermarket advertisement (M = 4.33 vs. 3.51), but not so when the bias reminder was presented (M = 4.09 vs. 4.15). In addition, two-ways interaction (Bias Reminder x Subjective knowledge), F (1,159) = 3.40, p = .07 and three-ways interaction, F (1,159) = 2.93, p = .09 was partially significant. . Table 3. ANOVA Output on participants’ Rating of Involvement Source Endorer Bias_reminder* Endorser Error. Sum of Squares 6.243 8.398. df. F. Sig.. 1 1. Mean Square 6.243 8.398. 4.103 5.519. .044 .020. 241.941. 159. 1.522. Endorser. The manipulation check on endorser attractiveness showed only the significant main effect of Endorser, while the other effects were not significant, as shown in Table 4. Participants perceived the celebrity endorser in the target. 20.

(29) advertisement more attractive (M = 4.76) than the average endorser in the target advertisement (M = 3.95), F (1,159) = 29.84, p < .001. . Table 4. ANOVA Output on participants’ Rating of Endorser Attractiveness Source Endorer Error. Sum of Squares 27.085 144.341. df 1 159. Mean Square 27.085 .908. F. Sig.. 29.835. .000. Argument Quality. Following the 2 by 2 by 2 AVOVA on participants’ rating of argument quality, there were a significant two-ways interaction (Bias Reminder x Endorser, F (1,159) = 7.31, p <. 05) and three-ways interaction, F (1,159) = 6.03, p< .05 respectively as shown in Table 5. The significant Bias Reminder by Endorser interaction showed celebrity endorser intensified the argument persuasiveness when the bias reminder was not presented (M = 4.39 vs. 3.80), but denounced the argument persuasiveness when the bias reminder was presented (M = 4.02 vs. 4.52). The significant three-way interaction indicated that the aforementioned Bias Reminder by Endorser interaction was only presented when participants’ subjective knowledge was manipulated relatively low rather than when participants’ subjective knowledge was manipulated relatively high. . 21.

(30) Table 5. ANOVA Output on participants’ Rating of Argument Quality Source Bias_reminder* Endorser Bias_reminder* Subjective_knowledge * Endorser Error. Sum of Squares 13.151. df. F. Sig.. 1. Mean Square 13.151. 7.313. .008. 6.034. 1. 10.851. 6.034. .015. 285.921. 159. 1.798. Objective knowledge. For checking the level of participants’ objective knowledge by using the 2 by 2 by 2 ANOVA, there was only a significant main effect of Subjective knowledge as shown in Table 6. The participants who read the article enhancing the objective knowledge received higher scores (M = 5.549) in the subsequent knowledge test than the participants who read the article unrelated to the following knowledge test (M = 4.74), F (1,159) = 5.83, p < .05. . Table 6. ANOVA Output on participants’ Rating of Objective knowledge Source Subjective_knowledge Error. Sum of Squares 22.555 615.022. df 1 159. Mean Square 22.555 3.868. F. Sig.. 5.831. .017. Dependent measure Attitude. The result of ANOVA displayed that there was a very significant main effect of Endorser and a partially significant main effect of Bias Reminder, as. 22.

(31) shown in Table 7. The celebrity endorser in the target advertisement led the participants to have higher attitude ratings (M = 4.63) than the average endorser in the target advertisement (M = 4.11, F (1,159) = 13.77, p < .0001). Additionally, the participants who came across without Bias Reminder in the target advertisement expressed higher attitude rating (M = 4.52) than the participants who came across with Bias Reminder in the target advertisement (M = 4.24, F (1,159) = 3.43, p = .067. . Table 1 ANOVA Output on participants’ Rating of Attitude Source Endorser Error. Sum of Squares 11.586 133.797. df 1 159. Mean Square 11.586 .841. F. Sig.. 13.786. .000. As shown in Figure and Figure 3, for participants low in subjective knowledge, bias reminder negatively influenced the target attitude only when the endorser was attractive (M = 4.32 vs. 5.20, F(1,159) = 8.97, p <.005 in the simple main effect test as shown in Table 8), but not so when the endorser was not as attractive (M= 4.16 vs. 4.18, F (1,159) = .01, p =.95 in the simple main effect test), consistent with Hypothesis 1, in that the correction on attractive endorser was more likely to occur in the condition of low subjective knowledge. On the other hand, for participants high in subjective knowledge, bias reminder did not differ the target attitude either when the attractive endorser was presented (M = 4.51 vs. 4.56, F (1,159) = .02, p = .89 in the simple main effect test) or when the average endorser was received in the target advertisement (M = 4.01 vs. 4.13, F (1,159) =. 18, p = .68. 23.

(32) in the simple main effect test), consistent with Hypothesis 2 in that the correction on attractive endorser was less likely to occur in the condition of high subjective knowledge. . Figure 1. Mean Ratings of Target Attitude (Low subjective knowledge). 24.

(33) . Figure 2. Mean Ratings of Target Attitude (High subjective knowledge). Table 2. Syntax Output of simple main effect test. 25.

(34) . CHAPTER 6 GENERAL DISCUSSION. Conclusion. This current study has illustrated that subjective knowledge can be the. primary effect on consumers’ evaluation of the product. The result of our experiment presented participants with low subjective knowledge; the attractive endorser does not have much effect on their evaluation. On the other hand, participants with high subjective knowledge, the attractive endorse influences their evaluation. Even though the participants in the condition of high objective knowledge answered more questions correctly than in the condition of low objective knowledge (M =5.49 vs. 4.74), the participants high in objective knowledge answered .75 correct answer (less than one correct answer) on average more than the participants low in objective knowledge. The difference between the two conditions was not pretty huge. Such a significant but not huge difference might support the manipulation of subjective knowledge immediately after the knowledge test. That is, it was more likely for the participants high in objective knowledge with the average correct answer of 5.49 to believe that they only made two correct answers. It would have been even more unlikely for the participants high in objective knowledge making 8 or 9 correct answers to believe that they only made two correct answers shown on the page of subjective knowledge manipulation. In. 26.

(35) the same vein, it was more likely for the participants low in objective knowledge with the average correct answer of 4.74 to accept that they turned out making seven correct answers in the manipulation of high subjective knowledge. It would have been even more unlikely for the participants low in objective knowledge actually making 1 or 2 correct answers to believe that they eventually made seven correct answers. The success of subjective knowledge manipulation depends on the ambiguity of participants' prior perceptions of their own subjective knowledge. Additionally, our study also recommends that subjective knowledge may impact consumers’ confidence in the evaluation. Academic Contribution and Managerial Implication The present study has examined the corrections of people with high subjective knowledge and low objective knowledge and people with low subjective knowledge and high subjective knowledge of the product’s evaluation. After our experiment, we discovered that all of our hypothesizes are certified. People with low subject knowledge have the correction with attractive endorser more likely than people with high subjective knowledge. In contrast, people with high subjective knowledge, bias reminder, and endorse have not distinct toward their attitude. Thus, doing the marketing plan, the marketing planner should be more accurate when using the attractive endorser. The attractive endorser constantly may not be effective in the advertisement. As the attractive endorsers possibly do not have sufficient influence to persuade the consumers, especially with consumers who have the high subjective knowledge. . 27.

(36) . Additionally, the questionnaires in this experiment were designed by based on Taiwanese experiment pattern. However, we collected the data and experimented with Thai people. Hence, this study presented that this pattern of questionnaires is practicable with Thai people. Limitations and Future Research There are multiple effects have the influence on consumers’ evaluation and subjective knowledge is one of the main effects on evaluation. When we started to experiment and designed our questionnaires, we also expected that objective knowledge and subjective knowledge could have any effect on consumers’ confidence and this confidence could play an important role on consumers’ judgment toward that product. However, our results did not definitely verify of this expectation. Thus, we suggest that after manipulating participants’ objective knowledge and subjective knowledge, we should immediately exanimate participants’ confidence level and check participants’ subjective knowledge before the target advertisement was presented. Lastly, this study we collected the data by the online survey in Thailand, and Thai people do not get used to this kind of online questionnaire. Perchance there were some participants did not pay more attention to our questionnaire. Therefore, we suggest collecting the data by paper collection may be more suitable for this experiment. . 28.

(37) . REFERENCE. Alba, J., & Hutchinson, J. (1987). Dimensions of consumer expertise. Journal of Consumer Research, 13(4), 411. doi: 10.1086/209080 Alba, J., & Hutchinson, J. (2000). Knowledge calibration: What consumers know and what they think they know. Journal of Consumer Research, 27(2), 123-156. doi: 10.1086/314317 Bearden, W., Hardesty, D., & Rose, R. (2001). Consumer self-confidence: Refinements in conceptualization and measurement. Journal of Consumer Research, 28(1), 121134. doi: 10.1086/321951 Bless, H., & Schwarz, N. (2010). Mental construal and the emergence of assimilation and contrast effects: The inclusion/exclusion model. Experimental Social Psychology, 42, 319-373. Briñol, P., & Petty, R. (2009). Persuasion: Insights from the self-validation hypothesis. Advances in Experimental Social Psychology, 41, 69-118. Brucks, M. (1985). The effects of product class knowledge on information search behavior. Journal of Consumer Research, 12(1), 1. doi: 10.1086/209031 Cacioppo, J., & Petty, R. (1984). The elaboration likelihood model of persuasion. NA Advances in Consumer Research, Eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, 11, 673-675.. 29.

(38) Chien, Y., Wegener, D., Petty, R., & Hsiao, C. (2014). The flexible correction model: Bias correction guided by naïve theories of bias. Social and Personality Psychology Compass, 8(6), 275-286. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7(2), 117-140. doi: 10.1177/001872675400700202 Forehand, M., & Perkins, A. (2005). Implicit assimilation and explicit contrast: A set/reset model of response to celebrity voice-overs. Journal of Consumer Research, 32(3), 435-441. doi: 10.1086/497555 Gross, Sharon, Rolf, H., & Norma, M. (1995). Attitude certainty,” in attitude strength: Antecedents and consequences, ed. Richard E. Petty and John. A. Krosnick, Mahwah. NJ: Erlbaum, 215-245. Hadar, L., Sood, S., & Fox, C. (2013). Subjective knowledge in consumer financial decisions. Journal of Marketing Research, 50(3), 303-316. doi: 10.1509/jmr.10.0518 Johnson, E., & Russo, J. (1984). Product familiarity and learning new information. Journal of Consumer Research, 11(1), 542. doi: 10.1086/208990 Martin, L. (1986). Set/reset: Use and disuse of concepts in impression formation. Journal of Personality and Social Psychology, 51(3), 493-504. doi: 10.1037//00223514.51.3.493 Petty, R., & Cacioppo, J. (1986). Communication and persuasion: Central and peripheral routes to persuasion. New York: Springer - Verlag.. 30.

(39) Petty, R., & Wegener, D. (1999). The elaboration likelihood model: current status and controversies. New York: Guilford Press, 41-72. Petty, R., Briñol, P., & Tormala, Z. (2002). Thought confidence as a determinant of persuasion: The self-validation hypothesis. Journal of Personality and Social Psychology, 82(5), 722-741. doi: 10.1037//0022-3514.82.5.722 Petty, R., Briñol, P., Tormala, Z., & Wegener, D. (2007). The role of meta-cognition in social judgment. E. T. Higgins & A. Kruglanski (Eds.). Social Psychology: Handbook of Basic Principles (2Nd Ed.). New York: Guilford Press. Petty, R., Wegener, D., & White, P. (1998). Flexible correction processes in social judgment: Implications for persuasion. Social Cognition, 16(1), 93-113. doi: 10.1521/soco.1998.16.1.93 Razmdoost, K., Dimitriu, R., & Macdonald, E. (2015). The effect of overconfidence and underconfidence on consumer value. Psychology & Marketing, 32(4), 392-407. doi: 10.1002/mar.20787 Wegener, D., & Petty, R. (1995). Flexible correction processes in social judgment: The role of naive theories in corrections for perceived bias. Journal of Personality and Social Psychology, 68(1), 36-51. doi: 10.1037//0022-3514.68.1.36. 31.

(40)