E; )IHNIH“NW“NUIWIUIHHIHWHHNNIHHM 404 '-1 I .C” lllllltllli lllllllllllfllllllllllllllll Mass .____a_ r l LIBRARY Michigan State University This is to certify that the thesis entitled ATTCLFuféfi VM5EE“ / f‘ -' ' ll; ’ . ,' 4.’ 1’ r i: Coa'l ”'1- {\fitég‘i’g‘wc/‘b presented by rEfl~Lim/Zla/ has been accepted towards fulfillment of the requirements for M A Jegree in «BKC (L/JCJ VB \ / /1 “mi—'— ._, 7’4 (1/ x) // [/7; / I ‘ Major professor . /,~ ,,. Date ’/")4‘/7 3 0-7639 MS U is an Affirmative Action/Equal Opportunin Institution PLACE N RETURN BOX to remove We checkout from your record. TO AVOID FINES return on or More ode due. DATE DUE DATE DUE DATE DUE MSU leAn Mandi” Action/Ewe! Opportunity Instituion W ”1 A'ITRIBUTED POWER: THREAT vs. PERSUASION 3! Yen-lin Lin A THESIS Submitted to liohigen state Univeroity in pertiel fulfillment of the requirements for the degree of MASTER OF ARTS Department of Sociology 1995 ABSTRACT ATTRIBUTED POWER: THREAT V8. PERSUASION BY Yea-lin Lin How do people attribute power? An experimental study on a sample of college students examined the effect of the method of influence (threat vs. persuasion), structural power, and success known/unknown. major findings were that 1) people generally appreciate persuasion. The persuasion users would be perceived more likely to cause a desired outcome than the threat users, 2) while using threats successfully does not always lead to attribution of more power, threats may lead to perception of more power when structural power is equal, and 3) the threat users will be considered less likable and less intelligent than the persuasion users. The effect of both influence tactic in a democratic society and the responses from the observers has been discussed. ACKNOWLEDGEMENTS I am deeply indebted to Dr. Stan Kaplowitz for his guidance throughout my study and his continued efforts in helping me to produce a final draft. I also wish to acknowledge my committee members, Dr. Cliff Broman and Dr. Tom Conner for their support. Finally, I would like to express my sincerest gratitude to my parents and friends, Li-chung, Shu-fen, Lucy, Benny and William for many kind of words of encouragement. ii TABLE OF CONTENTS LIST OF TABLES THE TITLE OF A GRAPH INTRODUCTION METHOD Sample Design of the Experiment Description of the Situations The Manipulations of Independent Variables Manipulation Checks Dependent Variables RESULTS Manipulation Checks Hypotheses 1) Data Transformation 2) Hypothesis Tests DISCUSSION GRAPH 1 LIST OF REFERENCES iii Page iv 12 12 12 13 14 17 19 20 20 22 22 24 30 34 35 Table Table Table Table LIST OF TABLES Page Predicting Influencer Being Likely to 25 Compliance by Method of Influence and Structural Power: success unknown Rating of Ability to Overcome Resistance by 26 Method of Influence and Structural Power: success known Mean Amount of Liking of Agents by Method of 28 Influence and Structural Power: all subjects Mean Intelligence rating of Agents by Method 29 of Influence and Structural Power: all subjects iv INTRODUCTION One approach of social psychology is studying "power attribution”. How do people attribute power‘? Does the use of threats cause the attribution. of igreat. power, cause resentment toward the power user, or both? In this study, I attempt to answer two preceding questions as well as the following ones. If the outcome hasn't been known, how will people predict the probability of the successful influence? How do people evaluate the agent's likability and intelligence after they witness threatening process? Under what conditions will the agent be attributed with the greatest power, likability, and intelligence? Basically, our research design is similar to Layton and Moehle (1979) research. In their study, a sample of 233 male and female undergraduates volunteered for the experiment. The participants were asked to read a case history describing the process of an agent attempting to modify the behavior of a target. Then, they responded to some questions about the vignette. Schopler and Layton (1972) believed that a strong 1"Power is defined as the ability or potential to influence others" (Manz and Gioia, 1983, p.473). ”...power...,defined here as the inferred potential of one person(the agent) tolcause another person(the target) to act in accordance with the agent's wishes" (HinKin and Schriesheim, 1990, p.222). 1 2 intervention such as a threat should be more likely to produce compliance than a weak intervention, such as a suggestion. Then they inferred that when intervention strength increases, the probabilities for compliance also increase. Layton and Moehle (1979) found that a strong intervention caused change, i.e., compliance or countercompliance, rather than noncompliance. Threats were such strong interventions that they pushed the targets to respond or to make a change, regardless of the direction of that change. Although knowing that explicit threats may cause reactance and anger by the targets, and that the reactance may cause countercompliance, Layton insisted that the use of threat is an efficient method to obtain the agent's desired outcome. In contrast, Ridgeway (1987, Ridgeway and Diekema 1989) demonstrated the disadvantages of the use of stronger interventions. The research(1989) showed that bystanders consistently intervened against dominance behaviors. Bystanders showed negative response to dominance behavior”. Ridgeway and Diekema also found that dominance behaviors achieved no more influence over bystanders' decisions than did neutral confederate behaviors. ". . .bystanders were no more influenced in this way by dominant confederates than by neutral confederates" (Ridgeway and Diekema, 1989. p.88) . 2 According to Ridgeway and Diekema(1989) , the dominance behaviors include dismissal of another's arguments, addressing another in a contemptuous tone of voice, raised or angry voice, and leaning forward in an intrusive manner...and so on. 3 Briefly, bystanders preferred neutral confederate behaviors rather than dominance behaviors. I would link the use of threat and dominance behaviors together. A threat user usually dismisses another's argument, and addresses another in a dismissive or contemptuous tone of voice. A threat user always tries to dominate others. Thus, Ridgeway implied that the use of threat would not cause more compliance. From the view of bystanders, the dominance behaviors would not enhance the confederate's influence. In other words, Ridgeway did not predict that threats would lead to more success in the process of influence. On the other hand, although Layton did not directly compare threatening with persuasion, he did prefer the use of threat for achieving a desired outcome. Persuasion is the intervention which is stronger than a suggestion, but is weaker than a threat. Layton believed that the probabilities of compliance increase when the strength of intervention increases. Thus, Layton would predict the use of threat might cause more compliance. Then, we have two opposite hypotheses: l.1a if the success of influence is unknown, explicit threats will be considered by observers more likely to succeed than persuasion.(Schopler and Layton 1972) n.1b if the success of influence is unknown, explicit threats will be considered by observers less likely 4 to succeed than persuasion. (Ridgeway 1987; Ridgeway and Diekema 1989) We can test those two opposite hypotheses at the same time since ANOVA is a non-directional test. The direction of the hypothesis which has been supported can be identified. Now let us consider the attribution. ‘When the influence is snsssssfnl, the use of threat makes the power inequality salient. According to Kaplowitz's ( 1978) proposition, the use of threat implies that the target is very resistant to the influence. Thus, if the influence has succeeded, the very fact of having overcome great resistance would make more power attributed where an explicit threat was used. This implies that both agents and targets are aware of the unpleasant nature of threats. The agent avoids using threats except when running into great resistance. Thus, Kaplowitz (1978) predicts that more power will be attributed to the agent when the agent has used a threat if the influence has succeeded. Although Layton and Moehle ( 1979) had a proposition similar to Kaplowi‘tz's, their reasoning was different. At the beginning, Layton argued that threat should be more likely to produce compliance than would the weak intervention(Schopler and Layton 1972) . In other words, threats lead to more compliance than persuasion. However, Kaplowitz did not make this assumption. Layton implied the existence of the positive relationship between ”threat" and "compliance"; as 5 intervention strength increases, the probabilities for compliance also increase. After that, Layton concluded that "Thus, when the target complies, more influence should be attributed to the agent when the influence attempt is stronger" (1979, p.245). I would argue that Layton did not elaborate why the agent is credited with more influence over the target when the influence attempt is stronger. He did not sense that the reactance resulting from threats would cause the agent to avoid using a threat. He did not deal with the relationship between "the stronger intervention” and "the great resistance”. (Although Layton's proposition was similar to that of Kaplowitz, he did not express his logical thinking clearly. However, Kaplowitz(1978) made the more cautious proposition; only if the target has complied, more power would be attributed to the agent when a threat has been used. To test Kaplowitz and Layton's proposition, we have the hypothesis: 8.2 if A prefers outcome 8 and x occurs, then.more power over 8 will be attributed to A, if A has used a threat; and less power will be attributed to A if A has used persuasion. Let us now consider the role of structural power. Hegtvedt et al. (1993) mentioned that people may identify the level of the power-based social category to which each actor 6 belongs. People perceive power position positively related to structural power position. According to Hana and Gioia (1983) , French and Raven proposed five different types of power: 1) reward power; 2)coercive power; 3) legitimate power; 4) referent power; 5) expert power. Before I estate the hypotheses below, two assumptions have to be demonstrated here. First, I would like to simplify French and Raven's five types of power to two categories of power: structural power and personal power. structural power, including 'reward power', 'coercive power' and 'legitimate.power', results from the unequal status of power relationship. Structural power has been given from outside/social system. Personal power, including 'referent power', and 'expert power', is from an agent's inside/personality. Second, generally, when people attribute power to an agent, they do not separate those different types of power. They give an overall evaluation about power. Furthermore, I will predict that more power would be attributed to the threat user than to the persuasion user when the agent does not have a clear structural power advantage if the influence has succeeded. When the agent does not have structural advantage and dares to threaten, the more power would be attributed to this agent than to the agent using persuasion if the target has complied. The use of threats causes resentment, and the target may ask,”Who are you? How 7 dare you threaten me?” Without ' structural power' , the agent' s 'personal power' must have been big enough to overcome the reactance from the resentment and to lead to the target's compliance. Thus, successful threats are more likely to be useful for establishing the perception of power when there is not an obvious structural power. . However, in an unequal power condition, I will predict that the successful persuasion users would be credited with more power than the successful threat users. There are two possible reasons: 1) the agent might "speak softly but carry a big stick”. When status is obviously unequal, threats need not be used and compliance occurs without them; 2) people would attribute both 'structural power' and personal power' to the agent, using persuasion. The successful threat users would be considered to have less 'personal power', and need to use 'structural power' to achieve his/her way. Yet, someone may argue that threats might imply the agent can afford to he hated even when power is obvious. The agents might be proud of‘their structural power advantage and do not worry about the resentment which may cause countercompliance. Thus, we need to test the hypothesis: l.3 There should be an interaction effect between 'structural power' and 'means of influence' used on power attribution when the influence is successful. when structural power is equal, the use of threat may cause perception of more power than persuasion. In 8 the unequal power situation, threats may not convey more power . Now we move to how people give the attribution of likability to the agent. Manz and Gioia (1983) in ”The interrelationship of power and control” demonstrated that "the type of power used to exercise personal control will influence the controlled/controller relationship"(p.459). Coercive power, such as threats, used in the process of influence is likely to cause the target to be alienated from the agent. The use of referent power, which is often processing by persuasion, is more likely to lead to commitment. If the target feels comfortable and likes the agent through the exercise of influence, then the target.would like to commit to the agent's goal. Since people naturally hate being threatened, I will predict that the target would like the persuasion user rather than the threat users. Using raw threats will reduce the degree to which the influencing party is liked. A similar effect was found from Scholar and Layton's (1974) study that the use of coercive power was likely to decrease the attraction between the agent and the target. Ridgeway and Diekema(1989) also stated,"Dominant confederates were perceived as less likable. . .than neutral confederates" (p.90) . Then, we have the hypothesis: 3.4 a person who uses raw threats will be considered less 9 likable than one who uses persuasion. How do people attribute the intelligence to the agent? Schopler and Layton (1974) have suggested that the use of different types of power results in different attributions about the agent. From that study, the correlation between 'the use {of different types of power’ and 'different attributions about the agent' is confirmed. Yet this statement did not definitely confirm that people would attribute less intelligence to the threat users. Let me explain how people attribute intelligence to the agents in either equal or unequal power condition below. In a power equal condition, the agent would be considered less intelligent if s/he uses threats. There is no basis for the agent to use threats when both actors have equal power. People would not attribute more intelligence to the agent who lacks a structural power advantage and still tried to threaten. This kind of threat usually causes resentment toward the agent and seldom results in the successful influence. In an unequal power condition, the threat users would be evaluated as less intelligent than the persuasion users, too. ’Personal power' results from the respect for the intelligence of the agent. Whether ' personal power' exists depends on which kind of method of influence has been used. Anyway, people would not attribute much 'personal.power' to the threat 10 users. Since people would increase the attribution of the intelligence of the agent as the personal power increases, the threat users would not be evaluated as more intelligent. Finally, we concluded that the threat users would be evaluated as less intelligent in both an equal and an unequal power condition. Intelligence may be substituted for raw threats in interpersonal relations. There is a similar finding in Ridgeway and.Diekema’s research (1989). Those*with dominant behaviors were rated as lower in task capacity than those with neutral behaviors. Yet, there is another argument that intelligence is not necessary to link with 'personal power'. The threat users still could be credited with more intelligence. Sometimes, people comply or obey simply because of fear. Although the targets did not respect the agents' personal power, they do comply with-the threat users' demand. They comply because they fear the agent's intelligence. They do not want to resist the agent's intelligence. In this situation, the threat users would be attributed with more intelligence. Thus, we need to test the hypothesis: 8.5 the threat users will to be viewed as less intelligent than the persuasion users. In addition, there is another point I hope to remind--we are going to examine the responses from the observers. In the studies I mentioned above, the ratings or evaluations are all 11 from the persons who are bystanders or observers. We will see how people, not targets, attribute power to the threats users. METHOD Sample The sample consisted of 193 Americans (59 Males, and 133 Females). The sample was selected from students at Michigan State University. Data was gathered in Fall, 1993. Design of the Dxperiment Each subject was randomly assigned to one of the 32 separate conditions in a 2 x 2 x 2 x 2 x 2 complete factors design. The forms included three between-subjects variables and two ‘within-subjects variables. Each form. has four vignettes. The Ivignettes differed on five independent variables: 1) sex of actors (all male or all female); 2) whether the influence in all four vignettes of each form is successful or the success is unknown; 3) whether the relationship between actors is personal or impersonal’; 4) whether the actor attempting to influence the other used threats or persuasion; 5) whether the actor who was attempting influence had more structural power than the other or whether they had equal power. Each experimental form contained four vignettes. The 3While 'sex of actor' and 'personal vs. impersonal' were manipulated, this paper will not consider the effects of these variables. 12 13 last two independent variables were within subject variables. Subjects were assigned randomly to one of 32 forms .' However, to simplify the analysis, I decided to analyze only the data from the first story of each four. There are two reasons for using only the first story. First, this way I do not need to deal with the effect of the order in which the story occurred in the form. The second reason is that considering all four stories would have required more advanced statistical techniques, such as repeated measures MANOVA. Even just using data from the first story, I still have data of 193 cases randomly from all different conditions. Description of the Situations All vignettes had a dialogue which went like this: A : We are having difficulties here. I need you to work extra to fill in gaps left by the absence of another workers. B : If I do more, will I get paid more? A : I don't think that's possible. * * * * t 'k * * * 'k * "How realistic is the story?", somebody may ask. We designed a question to measure whether the subjects see the stories as "a real story". We found two significant F-values on main effects. "The method of influence" and "structural power" affected whether the subjects believed in the l4 reliability of the stories. The mean of the story reliability was 60.30 for the ’threat’ group; but 77.15 for ’persuasion’ group(F[1,189]=17.82,p<.001). The subjects tend to believe that the influence process with persuasion is more realistic. I would argue that persuasion happens more often than threats in their everyday life. People consider ’the use of threat’ as more unusual. Moreover, the subjects believe that Person A is more likely to influence Person B in the unequal power situation. The mean of the reliability was 72.03 for the ’unequal power’ group; but 60.67 for ’equal power’ group (F[1,189]=8.38,p<.005). None of the 2-way interactions is significantly different. Briefly, the manipulation of the story reliability was not failed. Most subjects believed that story could be happen in the real life. The average of the ratings of reliability was over 60 out of 100. The Manipulations of Independent Variables The independent variables were manipulated as follow: Sex of actors we manipulated the actors’ sex through giving A and B masculine or feminine first names or titles (Mr. vs. Ms.). Threat vs. Persuasion For gnzssg, after saying ”I don’t think.that’s possible...", A said "But.if you.know“what’s good for you, you’ll co-operate." For M, after saying, "I 15 don’t think that’s possible", A said, "But let me explain to you why this request is in the long range interest of the business and of all of us who work here...", the form then added "The explanation followed." Successful influence vs. Success unknown For cess, after the influence attempt, the sentence, ”After some thoughts, B did what A asked" was added at the end of the vignettes. For .sgggsss__gnkngyn, this sentence ‘was not included. Personal vs. Impersonal For a W, the actors were siblings, or parent and child ”in a family owned business". For impegsonal relationship, the actors worked "at a large company". In addition, in the small family owned business, we called people by their first name; in the impersonal large company we used people’s last names with Mr. or Ms. Structural. power (equality 'vs. inequality) In. the personal condition, the actors were "siblings" and co-workers when they had egga], pQweg. For M, A was the parent of B and owner of the business. Power resides in dependency relationship. The power of person A over person B was equal to the dependency of B upon A. A is both a parent of B and an owner of controlling resources over B. Those two relationships showed us their unequal power (Manz and Gioia, 1983; Emerson, 1962). In the impersonal condition, an "equal relationship" was 16 done in this way--”A and B were division managers at thg samg {s33 at a large company". 0n the other hand, "A and B were both managers at a large corporation but A was B’s superior" showed "unequal relationship". I will copy the vignettes in two totally oppOsite forms to show how we describe different situations‘. As the following form, the situation of this form is that all actors are female, success of the influence is unknown, personal relationship between actors is personal family business, the way of influence is persuasion, and structural power is equal. Here is a :male, success of influence unknown, personal relationship, persuasion, equal power vignette. Daniel and Robert were siblings and W in a family owned business. Daniel: We are having difficulties here. I need you to work extra to fill in gaps left by the absence of another worker. Robert: If I do more, will I get paid more? Daniel: I don’t think that’s possible. But let me explain to you why this proposal is in the long range interest of the business and of all of us who work here and the entire family. "Sex of actors’ is designed as a manipulative between variable in this study. Sex of influencer does affect how people attribute power to the agents(Hegtvedt et al. 1993; Molm 1986, 1991). In this paper, I do not intend to deal with the gender issue. 17 The explanation followed. * i i i * i * * t * * The example below is totally opposite form--female, success known, impersonal, threat, unequal power. Ms. Johnson and Ms. Davidson were both managers at a large corporation but WW- Ms. Johnson: Another division under me is having difficulties. They are understaffed and I need some people from your division to give them extra help. Ms. Davidson: If we do more than our regular jobs, will we get paid more? Ms. Johnson: I don’t think that’s possible but if you know what’s good for you, you’ll co-operate. After some thought, Ms. Davidson agreed to do what Ms. Johnson asked. * * * * * * * * * * * Manipulation Checks In this paper, I only focus on three independent variables--success of the influence, threats or persuasion, equal or unequal power. After collecting and entering the data, we checked these manipulations in each story to make sure that people did answer the questionnaires seriously. In question 1, the subjects were asked to compare the agent and 18 the target compare in their ability to reward or punish the other one» Thus, through this question, we can make sure that before answering the questionnaires, the subjects understand clearly whether those two actors’ structural power are equal or unequal. Question 1: 1) Based on what you know so far, how do A and B compare in their ability to reward or punish the other one? If A has all of this ability and B has none choose 100, if B has all and A has none, choose 0. If they have equal ability, choose 50. You may however choose any number between 0 and 100. B has all A has all of this A and B of this ability have equal ability and A none ability and B none 0 50 100 A has units of the ability to reward or punish. * * * * 'k 'k * * * * * In question 5, the subjects were asked to recall which approach the agent used.to try to influence the target--threat or persuasion. As a check on the manipulation of "the means of influence", the subjects were asked Question 5: Sometimes people attempt to influence others by using a thggat and sometimes they use W (attempting to convince people with ideas). Which approach did A use to try to influence B? Let 100 be entirely based on threat and 0 entirely based on persuasion and numbers in between indicating a mixture of the two. A’s attempt to influence B was based % on threat. 19 i i i * * * 'k * * * i To test the manipulation of success of influence, we asked: Question 6: How likely is it that S complied with what A requested? (Choose any number between 0iand 100, where 100 - certain he did agree, 0 = certain he didn’t ; 50 = he was equally likely to accept or not accept.) The probability that B agreed to A’s proposal is % i * * i * * * * * * * Dependent Variables In this paper, there are 4 dependent variables on the measures of "be likely to success", "ability to overcome the resistance", "likability", "intelligence". Question 6: ”How likely is it that B complied with A’s request?" (when success is unknown) is a :measure of predicting’ whether influence is believed successful or hots Question 10 : "A.has units of ability to get his/her way" (when A has already overcome the resistance) is a measure of perceived power. In addition, Question 14 and 16 are designed to measure the likeable evalhation of the agent, and Question 12 is a measure of intelligence of the agent. Those questions are below: Let moderate ability to get one’s way be 100. If you think someone has ggigg as much as moderate ability to get his/her give that person a score of 200. If you think he/she has half as much of this ability as moderate, give him/her a score of 50. If you think the person has no ability at all to get his/her way give him/her a score of zero. You may use any number you wish and there is no upper limit. 20 Question 14: If 100 is moderately likeable, A is units likeable. Question 16: If 100 is moderately considerate, A is units considerate. Question 12: If 100 is a moderate amount of intelligence, A has units of intelligence. RESULTS Manipulation Checks To see if we successfully manipulated structural power relations, we examined responses to Question 1 about "the distribution of ability to reward or punish". We did manipulation checks on Variable 1 (the ability for A to reward or punish B) and found that the mean of Variable 1 for ’unequal power’ group was 77.13, and Standard Deviation was 17.43; but the mean for ’equal power’ group 51.46, and standard Deviation 9.51. The t-test for the difference between means was highly significant t(1, 163.9) = 12.94, p < .001. That means, the subjects recognized the power status of the agent very well before doing the questionnaires. To see if we successfully manipulated the threat or persuasion condition, we examined Variable .5, the responses to "the methods of A attempting to influence B". We found that the mean of Variable 5 for ’threat’ group was 80.67, and Standard Deviation was 22.72; but the mean for ’persuasion’ group 22.64, and Standard Deviation 25.98. The result of significant t-test, t(1, 136.6) = 15.76, p < .001, indicated that the condition of threat or persuasion had been recognized well. Both the skewnesses of these two variable (Variable 1 and 5) are minimal. Thus, we don’t grossly violate the assumption 21 22 of normality, and we can use T-test to do manipulation checks. . To see if we successfully manipulated the ’success’ or ’success unknown’ condition, we examined the responses to Question 6 about ”the probability that B complied with A’s request”. However, because the skewness of Variable 6 with ”success” did not fall between -l.5 and +1.5, we did nonparametric Mann-Whitney test, instead of t-test, on variable 6(success is known or unknown). We took Variable 6 with ”success" as the independent variable to manipulate the successful situation. We found, the median of probability of compliance for success group equalled 95.00, but the median for success unknown group equalled 75.00. The significant 2 value ( z[1, 192] = -5.3612, 2-Tailed.p < .001) showed us that the subjects could tell whether the influence was successful or success unknown. Briefly, the manipulations were successful. Hypotheses 1) Data Transformation After checking manipulations, we checked the skewnesses of all dependent variables. Statistical procedures in the general linear model (eg. regression and ANOVA) requires that population residues be normal and homoscedastic. As needed, we did data transformations on half of variable 6 , and all other dependent variables of interest to achieve normality and homoscedasticitys The process is belowu To have the 23 skewnesses falling between -1.5 to +1.5, we used LG(Variable x + 25) instead of the original Variable X. So far, we have modified the data to approximate normality. Next, running factor analysis helps us to decide how many fastors from the dependent variables we have from this data set. Factor analysis helps us to choose which variables we will use to test the hypotheses. For example, originally, we measured the units of the agent’s likability, consideration, fairness, and rationale. Thesefour variables were expected to represent "how likeable the agent is". After running factor analysis, from reliability analysis we found better Alpha value (-.8665) could be obtained if two items have been deleted. Thus, we decided to delete the two items, the agent’s fairness and rationale. Through the similar process, a factor analysis showed that Variable 14 (the agent’s likeable units) and Variable 16 (the agent’s considerate units) loaded most highly on the 'Likeable” factor, and an index composed of these two items was used for the likable factor. We found that Variable 8 (units of ability to get the agent’s way) and Variable 10 (units of ability to overcome resistance) loaded.most highly on the power factor. However, an index composed of both variables 8 and 10 had less correlation with our dependent variables than did Variable 10 alone. Thus, we used Variable 10 (the agent’s ability to overcome the resistance) alone to measure perceived power. 24 2) Hypothesis Tests Hypotheses are based on ANOVA using as predictors of all five manipulated variables plus "gender of subjects". Hypothesis 1a predicts that if the success of influence is unknown, threats will be considered more likely to succeed than persuasion. Hypothesis 1b states that when success is unknown people tend to predict that using persuasion is more likely to result in successful influence than using threats. From Table 1, the mean of success for threat group is only 67.48 compared with the mean for persuasion group 79.57. A significant of (F[1, 93] = 5.87, P = 0.013 < .05, n’= .048) showed that Hypothesis 1b was supported. Thus, we can conclude that people in our population of college students believe the use of persuasion is more likely to succeed than the use of threats. The ‘ data analysis of the same variable showed that "structural power" and "the method of influence" had an interaction. The interaction was significant--F(~1,93)= 4. 031, p = .049, n3=.033 From the subjects’ prediction, the use of persuasion is more likely to succeed than the use of threat no matter people are equal power or not. Yet the difference of the successful probabilities between the use of persuasion and the use of threat in equal power condition is significantly higher than that in unequal power condition. 25 Table 1 Predicting Influencer Being Likely to Compliance by Method of Influence and Structural Power:success gnkgown (sample size) Structural Power A has more power equal power Method of Influence Threat 79.63 54.46 67.48 (30) (28) (58) Persuasion 84.00 73.75 79.57 (21) (16) (37) 81.43 61.48 (51) (44) **Note that the dependent variables had been logarithmically transformed. Hypothesis 2 predicts that if the influence is sgsssssjgl, explicit threats will be considered to show more power. than where persuasion was used. By examining the results constructed by ANOVA on Variable 10 (the agent’s ability to overcome resistance), hypothesis 2 was not supported. None of the F-value of the main effects was significant--for method of influence, F(1, 95) = .394, p=.532; for structural power, F(1, 95)=1.094, p=.30. That means, when the influence is successful, we did not find the significant difference on attributing the amount of power to the threat user or to the persuasion user. 26 Table 2 Rating of Ability to Overcome Resistance by Method of Influence and Structural Power: success knggn (sample size) Structural Power unequal power equal power Method of Influence Threat 2.26 2.28 2.27 (32) (29) (61) Persuasion 2.30 2.19 2.25 (20) (16) (35) 2.28 2.25 (52) (45) **Note that the dependent variables had been logarithmically transformed. As we predicted the Hypothesis 3--if the influence is successful, the agents using threats will be attributed to more ability to overcome resistance than those using persuasions when the actors’ status were equal-- was supported. From the graph 1, the data shows that when structural power is sggg1_and the success is known, the use of threat led to the perception of more power than would the use of persuasion; when structural power is W and the success is known, the result is opposite. The analysis of variance conducted on variable 10 showed a significant effect for 2-way interaction of "threat” and "structure” (F[l,95] = 4.664, p - .035 < .05, n3 a .053). Only this interaction is significant among all interactions. 27 We now consider Hypothesis 4, which says that a person who uses explicit threats will be considered less likable. After running factor analysis, we decided to combine Variablel4 (the agent’s likeable units) with Variable 16 (the agent’s considerate units) . as one dependent variable, "likability”, to test Hypothesis 4. According to the data analysis, we found the high significance on ”the methods Of influence'--F(1,188) 3 43.117, p<.001, n3- .184. This supported Hypothesis 4 and showed.us that most.people like the agents who use persuasion rather than threats. There is no significant 2-way interactions neither 3—way interactions. These two independent variables could be considered as affecting the dependent variable independently. The Variable of "success known" did not affect the other two independent variables, either. Mean Amount of Liking of Agents 28 Table 3 by Method of Influence and Structural Power: s11 subjects (sample size) Structural Power A has more power equal Method of Influence Threat 4.12 4.00 4.07 (63) (57) (120) Persuasion 4.40 4.35 4.38 (38) (32) (70) 4.23 4.13 (101) (89) **Note that the dependent variables had been logarithmically transformed. Finally, Hypothesis 5 holds that using raw threats will reduce the degree of intelligence for which the agent is credited. By Table 4, analysis of variance conducted on Variable 12 (the intelligent units) showed a significant effect on intelligence (F[l, 187] 811.86, p=.001, n1= .059). This indicates that the person using persuasion will be considered more intelligent than those using threats. Hypothesis 5 was supported. In addition, there was no significant effect of structural power and no significant interaction between the method of influence and the structural power. Also, there is no significant 2-way interactions neither three way interactions among those independent variables. We could make sure that whether the success is known will not affect other variables. Mean Intelligence Rating of Agents Table 4 by Method of Influence and Structural Power: s11 subjects (sample size) Method of Influence Threat 2.22 (62) Persuasion 2.29 (38) 2.25 (100) Structural Power A has more power equal power 2.20 (57) 2.26 (32) 2.22 (89) 2.21 (119) 2.28 (70) **Note that the dependent variables had been logarithmically transformed. DISCUSSION The results of the study clearly supported the most hypotheses, except for Hypothesis 1a and Hypothesis 2. Overall, we can see a tendency--people generally appreciate persuasion. From the Hypothesis 1b, we found that the use of persuasion to influence others would be perceived more likely to cause a desired outcome than would the use of threats. When success of influence is ‘unknown, the agents ‘using persuasion would be thought to have a higher probability of success influencing the targets than would the threat users. Without considering any other factors, the use of persuasion would be predicted to cause more influence than the use of threats. Especially, in the unequal power condition, the use of persuasion is thought to have a much higher probability of success than would the use of threats. Ridgeway’s proposition of the limits of threats has been supported in this study. There are two possible explanations why our results support Ridgeway’ s demonstration rather than Layton’ s demonstration of the advantage of threats. First, Layton’s ”weak interventions" referred to ”suggestions" rather than ”persuasion“. Or we can say, Layton’s ’suggestions’ are a weak form of persuasion. He did not directly consider whether strong persuasion would lead to more compliance than would threats. Second, Ridgeway’s research has been finished in more 30 31 recent years. It is based on more current data than Layton’s research. It is possible that Americans’ attitudes about the use of threats has changed from 1979 to 1989. Now, Americans are more used to persuasion, instead of threats, than they did before. As to Hypothesis 2, some social scientists (cf. , Schopler r. Layton, 1972; Layton and Moehle, 1979; Kaplowitz, 1978) indicated an advantage of threatening. They believed that the use of threats may convey more power if the influence has succeeded. Yet, their hypothesis was not supported in this research. Both Hypothesis 2 and Hypothesis 3 are under the situation in which the influence is successful. From the evidence. of supporting’ Hypothesis 3, ‘we agree ‘with. the necessity of using threats to show' more power in some situations. While using threats successfully does not always lead to attribution of more power, threats may lead to perception of more power when structural power is equal. Hypothesis 4 was finding that the threat users will be considered less likable than the persuasion users. Finally, the finding was consistent with Hypothesis 5 which says that the persuasion users are considered.more intelligent than the threat users. Without considering other factors, the persuasion users is considered to be credited with more intelligence, compared with the threat users. In addition, this study has been done in U.S. society, which is in a universe of all democratic societies. 32 which is in a universe of all democratic societies. Therefore, people would think and hope that threats will not succeed because explicit threats cause resentment and reactance. If however we sampled from the population in a communist society, which belongs to another universe, the observers may not predict that threats lead to less compliance because of hating the threats. However, in China, sometimes the power is so strong that threats need not be used, and compliance occurs without explicit threats. In fact, the influence of structural power may be so strong that it is not necessary to use explicit threats to be perceived as powerful. "Speak softly but carry a big stick.” In other words, the structural power could be so strong that the method of influence may have little or no effect on the success of influence. In my opinion, the people in a democratic society would be unlikely to comply with a lot of fear. The main reason is that people can get help from other groups against the threat users in a democratic society. The authority is not concentrated into one unique actor. In most situations, people can always ask for help from other powerful actors to avoid being threatened and complying. Then, is it possible that threats will cause more compliance in a non-democratic society than in a democratic society? Because they are used to being threaten, people in a democratic society may have less resentment toward threats than those in a democratic 33 society. On the other hand, as I mentioned, we are examining the response of an observer who was not threatened directly. However, the observers are outsiders, and there is no real reward/punishment to them. When they become the real targets, the possibilities of successful influence under threat condition might be different. It may be easier for an outsider to refuse the agent's request then for the person receiving the threats to do so. When the threat has been used to influence, the target would have much more fear and nastiness against the agent than would the observer. Thus, the probability of successful influence caused by the use of threat might be.higher than that the observers predict. In the future, there are two directions of research we can pursue. First, a study should be done in. a non-democratic society to test whether the use of threats will be considered more influential to the agent. Second, I suggest designing a . situation which‘ can put the subjects in the position of targets. We can test whether the ratings of the targets would be different from those of the observers. Will the targets show more compliance than the observers’s predictions in a threatening condition? 34 Graph 1 Ratina of Ability to Overcome Resistance by Method of Influence and Structural Power success known abllitv to Overcom 2.32"— 2.28 __ ;/mrem 2.24 -- T 2.20 7— ‘ ‘ ‘ . ‘ persuasion ' t i - > structural power unequal power eauai power IREJHERIDNCHES French, J., Jr.,& Raven, B. 1959. "The Bases of Social Power." in C. Cartwright(Ed.), Studies in Social Power. Ann Arbor: U. of Michigan Press. Hegtvedt, Karen A., Eliane A. Thompson, and Karen 8. Cook, 1993. "Power and Equity: What Counts in.Attributions for Exchange Outcomes?” Social Psychology Quarterly Vol. 56 No.2 Pp.lOO-ll9. HinKin, Timothy and Chester Schriesheim. 1990. "Relationship Between Subordinate Perceptions of Supervisor Influence Tactics and Attributed Bases of Supervisory Power. " Human Relations, Vol.43, No. 3, Pp.221-37. Kaplowitz, Stan. A. 1978. "Towards a systematic theory of power attribution." Social Psychology, 41:131-148. Layton, Bruce and Debra Moehle, 1979. "Attributed Influence: The Importance of Observing Change." Journal of EXperimental Social Psychology 16:243-252. Manz, Charles & Dennis Gioia. 1983. "The Interrelationship of Power and Control." Human Relations,Vol.36, No.5,Pp.459-476. Molm, Linda D. 1986. ”Gender, Power, and Legitimation: A Test of Three Theories." AJS Vol.91,No.6,Pp.1356-86. . 1991. ”Affect and Social Exchange: Satisfaction in Power-Dependence Relations . " American Sociological Review Vol.56,Pp.475-93. Ridgeway, Cecilia L. 1987. "Nonverbal Behavior, Dominance, and the Basis of Status in Task Groups."American Sociological Review 52: 683-94. and David Diekema, 1989. ”Dominance and Collective Hierarchy Formation in Male and Female Task Groups." American Sociological Review 54:79-93. Schopler, J. and Layton, 8.0. 1972. Attributions of interpersonal power and influence. Morristown, N.J.: General Learning Press. _ , 1974 "Attribution of interpersonal power" In J. Tedeschi(Ed.). Perspectives on Social Power Chicago: Aldine. 35 "Il'lllllllllilllllllli