Persuasive Power of Compliments:
Importance of compliments` relatedness to the desired behavior
Barbora Netolická, 10602089
Master Thesis University of Amsterdam Graduate School of Communication Master`s programme Communication Science
Supervised by: Annemarie Wennekers June 27th, 2014
Abstract
This paper`s purpose was to extend Cialdini`s (2004) theory of the persuasiveness of compliments by determining whether relatedness of the compliment to the desired behavior affects compliance with the behavior request, and whether this effect is moderated by the importance attributed to the complimented domain. To investigate these issues, 301
participants were gathered to participate in an online experimental study. Participants started with a filler test and were then asked a favor; to write a movie trailer review for a follow-up study. They could comply or deny this request. Before the request, participants received a compliment that was either highly related or unrelated to the requested behavior, or they did not receive a compliment. Level of compliance was assessed by whether they agreed to write the movie trailer review, how many words they used, and how much time they spent on it. Moreover, the level of importance they assigned to the complimented domain (good taste in movies/design) was measured. Results showed that a compliment`s relatedness by itself did not significant affect compliance, yet when the level of importance assigned to the
complimented domain was added to the model, significant effects were obtained. Specifically, it was discovered that the more importance one attributed to the complimented domain, the more one’s compliance is positively affected by the compliment relatedness. Thus, to use compliments effectively, it is important to consider the importance the target assigns to the complimented domain. To persuade people who assign high importance, compliments highly related to the desired behavior are recommended. Moreover, it was predicted that effects of compliments were mediated by self-efficacy and self-esteem, yet no evidence was found to confirm or reject this prediction.
One major motivation in people`s lives is feeling good about the self (Monahan, 1995). This human motivation may have given rise to the old saying “flattery will get you anywhere”. The truth is, people are persuadable by praise (Cialdini & Sagarin, 2005). It is a natural human ability to feel the power of compliments and intuitively use it. For example, compliments have become very effective and commonly used techniques in contexts of hiring, promotions and salary decisions (Knapp, Hopper& Bell,1984).
Despite how popular it is for people to use this technique, and its applicability in many fields and professions (Knapp et al., 1984), little research has been done on the persuasive effects of compliments. Consequently, there is important information about effective
application of compliments missing. That could lead to the inefficient usage of compliments, what would be an unfortunate waste of compliments` potential. Revealing this missing information would not only contribute to current scientific knowledge of compliments` effectiveness, but also set concrete directions in compliments` usage. One of the issues that remain unclear and yet could determine how to apply a compliment in the most effective way is whether the degree of a compliment’s relatedness to the desired behavior influences a compliment’s effectiveness and what are the underlying mechanisms of compliments` effect on people`s willingness to comply.
To demonstrate the issue, consider the effects of the following phrase: dear reader, the fact that you are reading this paper means that you have an excellent taste in academic
literature. I hope you enjoy reading this study and value it highly. Now consider effects of this phrase: dear reader, the fact that you are evaluating this paper, means that you have an
excellent ability to organize your time. I hope you enjoy this paper and value it highly.
Assuming the phrases intent to persuade the reader to give this paper a positive review, which compliment has a higher impact, the one that targets taste for academic literature or the one that targets time-organizing ability?
This study aims to clarify the best ways to apply compliments in order to fill the aforementioned knowledge gap. Furthermore, it aims to determine whether a compliment is more effective in triggering compliance if directly related to the requested behavior
comparing to when not related to the requested behavior, and it examines whether this
relationship is moderated by the importance the audience assigns to the chosen complimented domain. These questions will be answered through an experiment, which will also help to determine which underlying processes make compliments effective in affecting compliance. Outcomes of this study could be used in politics, business, sales, public health campaigns, doctor-patient communication, as well as in education (Blase & Kirby, 1992).
Literature Overview Persuasive Potential of Compliments
Most generally, we know that compliments - positive remarks about another person’s
trait, attitude, or performance (Cialdini, 2001, p. 74) - are commonly used persuasive technique we are confronted with in everyday life (Knapp, Hopper, & Bell, 1984). Compliments are commonly used for a range of purposes, from personal needs to voting behavior (LeBourveau, Dwyer, & Kernan, 1988), sales (Cialdini & Goldstein, 2004), client`s brand fidelity generation (Wolford, Cox and Culp, 2001), corporate management, or
education (Blase & Kirby, 1992), and for all these purposes compliments can be persuasive if used properly (Grant, Fabrigar & Lim, 2010; Simons, Morreale & Gronbeck, 2001). For instance, Weinstein, Tosolin, Ghilardi & Zanardelli (1996) established that proper compliments can increase patients` compliance with the doctor`s recommendations. This persuasive potential of compliments lies in their ability to manipulate judgments and opinions, two components that shape people`s attitudes (Simons et al., 2001). Compliments seem to work through automatic or heuristic bases, which allow them to influence both explicit and implicit attitudes (Chan & Sengupta, 2008). Yet it is not clear which mechanisms
are essential for the compliments` effectiveness (Knapp, Hopper, & Bell, 1984). Mechanisms Explaining the Persuasive Power of Compliments
Perspective of Cialdini`s principles of persuasion. Several theories about possible underlying mechanisms of compliments have been constructed. Currently the most popular theory is Cialdini`s six principles of persuasion (Cialdini & Sagarin, 2005). Cialdini`s
principles stand for aspects of persuasive communication that reinforce persuasion and simply make it harder for a request receiver to say no (Cialdini & Sagarin, 2005; Simons, 2001). These principles are thought to function thought heuristics; shortcuts for effective decision-making (Cialdini & Goldstein, 2004). The two principles that may underlie effects of compliments are liking of the persuasion agent and reciprocity.
Principle of liking the persuasion agent. That compliments work through liking of
the persuasion agent is the most commonly accepted point of view within the perspective of Cialdini`s principles of persuasion (Cialdini, 2001; Pratkanis, 2007). Cialdini called this
notion “the principle of liking”; he argues that we tend to be persuaded by and more willing to comply with people we like, compared to people we do not like or people towards whom we have a neutral attitude (Cialdini & Goldstein, 2004; Martin, 2008; Suls, Martin & Wheeler, 2000). Cialdini further argues that compliments are one of the ways to achieve being liked and thus being more persuasive (Byme, Rasche, & Kelly, 1974). Cialdini concludes that liking of the persuasion agent mediates the process of compliments increasing compliance.
However, new findings oppose this idea. Specifically, it was found that being flattered indeed increases liking of the source, but that is not responsible for increased compliance levels (Grant, Fabrigar & Lim, 2010). These results suggest that effects of compliments are mediated by a different mechanism, possibly another Cialdini`s principle – reciprocity.
Reciprocity. The principle of reciprocity is similar to the mechanism of gift giving (Turner & Edgley, 1974). To illustrate, one participant of a social interaction could offer a
compliment as a gesture of kindness. In reaction to receiving kindness, the other participant feels urged to a reciprocal offering (Falk & Fischbacher, 2006). The decision of the receiver to offer the reciprocal kindness depends on his/her evaluation of whether the reason why the sender offered the original kind gesture is genuine (Falk & Fischbacher, 2006). Similarly, the level of reciprocity urge and hence the amount of reciprocated kindness depends on the level of kindness expressed at the first place (Turner & Edgley, 1974) and the evaluation of what the outcomes of the sender`s kindness are. If these are evaluated as low, only a low level of kindness in return may be offered, but if these are evaluated as high (Falk & Fischbacher, 2006), a greater level of kindness may be offered (Cialdini & Goldstein, 2004; Knapp, Hopper, & Bell, 1984).
Perspective of the theory of positive reinforcement. Another perspective that used to explain the persuasive power of compliments is the principle of positive reinforcement. According to this theory, there is positive and negative kind of reinforcement. Let’s use training animals as an example: in animal training, positive reinforcers consist of presenting stimuli as food, water or sexual contact (Skinner, 1965), while negative reinforcers consist of presenting unpleasant stimuli (Michael, 1975). The effect of positively reinforcing a behavior increases the probability of the occurrence of behavior in the future (Michael, 1975).
According to this perspective, a compliment functions as an unconditioned stimulus that increases likelihood of the receiver having certain response in the future (Michael, 1975). There are two phenomena that could facilitate the process of compliments as positive
reinforcers eliciting a desired behavior: first, perceived self-efficacy and second, self-esteem.
Self-efficacy is a belief that one is capable of performing the actions needed to succeed at a task (Bandura, 1997). It seems to be a strong predictor of achievement; previous research showed that in e.g. mathematics, self-efficacy predicts achievement better than general mental ability (Siegle & Mccoach, 2007). Increasing one`s self-efficacy appears to
motivate to work harder and to empower ability to go through the whole process. In turn, compliments seem to be effective tools to increase one`s perceived self-efficacy, since they draw the target`s attention to the past achievements (Siegle & Mccoach, 2007), major source of feeling that the task can be achieved in the future (Bandura, 1993). Therefore, we could argue that compliments lead to compliance because they increase the receiver`s perceived self-efficacy, which increases the chances of the receiver`s motivation to go through the requested task (Siegle & Mccoach, 2007).
An alternative facilitator of reinforcements producing a desired behavior is enhancing the self-esteem. According to previous research, ego-enhancing is essential in the process of praise increasing compliance. To explain, even a superficial praise is ego enhancing
(LeBourveau, Dwyer, & Kernan, 1988) and logically, ego-enhancement leads people to feel good about themselves. Isen and Levin (1972) found that people feeling good about
themselves are more helpful towards others compared to those not feeling particularly well about themselves. Further, people feeling good about themselves show increased benevolence towards others and thus a higher tendency to comply. Therefore, self-esteem enhancing messages, such as compliments, can be persuasive (Monahan, 1995).
How to Use Compliments in the Most Effective Way?
While there is evidence proving compliments` effectiveness in increasing compliance, this evidence is inconclusive on what the boundaries of compliments` effectiveness are and on how to use compliments in the most effective way. This study will address that by answering question: how should we structure a compliment to reach its maximal impact on someone`s decision to perform a certain behavior?
The importance of choice of complimented domain. It is already determined that
compliments should target domains (characteristics) that can be worth praise (Davidson, 2008). Yet it is unclear how to choose the best domain to reach the most desired effect. For
instance, does the domain`s relatedness to the desired behavior affect the compliment`s effectiveness? Does targeting something directly related to the desired behavior increase the power of a compliment?
There are reasons to believe that the choice of domain we compliment on does matter and that the domain`s role is influenced by ascribed personal weight. It was discovered that praising different characteristics might produce different behaviors; e.g. Mueller and Dweck (1998) studied how to bring children to improve school results by praising their intelligence or their effort and discovered that children increased their effort in schoolwork (achievement, motivation and task persistence) after being praised for their effort, but not for intelligence. It suggests the choice of the complimented domain may alter the outcome of the compliment. Based on Mueller and Dweck`s (1998) results it can be assumed that if the desired outcome of praise is the target audience`s effort in schoolwork, it is more effective to praise the target audience on effort. Unfortunately, Mueller and Dweck (1998) did not test for the cause of these differences in outcomes. There is one likely explanation: the role of the complimented domain`s relatedness to the desired behavior.
It seems to be reasonable to believe that the extent to which the complimented characteristic relates to the desired behavior may alter the compliment`s effectiveness in leading to the desired behavior (Hill & Buckley, 1968). According to the theory of reinforcement, reinforcing behavior A increases chances of behavior A occurring in the future. So we can argue that if aiming to produce behavior A, we should reinforce
(compliment on) behavior A rather than reinforcing (compliment on) a different behavior, B, that would rather increase chances of producing behavior B. Following this logic, it is possible to conclude that closely relating a compliment to the desired behavior may be more persuasive than not relating it to the desired behavior. This leads to the following hypotheses:
H1a: A compliment will be associated with greater compliance in a form of accepting a request if the compliment is about a trait that is related to the desired behavior as compared
to a trait that is not related to the desired behavior.
H1b: Similarly, for those who accept the request, a highly related compliment will be associated with greater compliance in a form of effort invested into performing the requested
behavior as compared to a low relatedness compliment.
However, this argument is valid assuming that compliments are underlined by
principles of reinforcement, not Cialdini`s principles of persuasion. To explain, according to Cialdini, relatedness of the complimented domain should be less relevant than according to reinforcement principles. Liking of the flatterer and tendency to reciprocate kindness should not be conditioned by the compliment`s relatedness because any compliment is supposed to increase liking of the persuasion agent and any kind gesture should increase reciprocity urges. On the other hand, the theory of reinforcement argues that only the reinforced behaviors have a higher chance of occurrence in the future, and so domain`s relatedness to the desired
outcome matters. Therefore, evidence that relatedness does matter would suggest that the underlying mechanism of effects of compliments is the principle of reinforcement, not the perspective of Cialdini`s principles of persuasion, and vice versa. Hence, we can extend H1:
H1c: The effect of compliment relatedness to the desired behavior on compliance is mediated by increased levels of self-efficacy and self-esteem rather than Cialdini`s principles of
persuasion, reciprocity and liking.
Other speculating theories. An alternative theory would counterargue:
complimenting a closely related domain may produce suspicion of a persuasive attempt, leading the target to activate persuasion knowledge, consequently limiting persuasive effects of the compliment (Campbell & Kirmani, 2000). This theory is supported by a discovery that when the persuasion target is aware of an ulterior motive behind the flattery, the compliment is rejected and corrected for (Campbell & Kirmani 2000; Main, Dahl & Darke 2007; Chan & Sengupta, 2008). Based on this theory, it could be argued that complimenting a domain non-related to the desired behavior may be more persuasive; it would raise less suspicions of a
persuasive attempt activating less reactance (Fiedler & Mata, 2013) preventing the persuasive effect (Visser, 2010). However, research has also demonstrated that perception of the
compliment`s sincerity is irrelevant to the compliment`s effectiveness. There is evidence showing that virtually all compliments are interpreted as sincere to some degree (Turner & Edgley, 1974) and even in cases when the degree to which compliments are considered sincere is low, compliments seem to lead to the desired effect. In short, compliments seem to be persuasive even when the receiver is aware of them not being genuine (Turner & Edgley, 1974; Cialdini, 2005).
This discrepancy in findings may be explained by differences in measurement of explicit versus implicit attitudes. To explain, explicit and implicit attitudes are two separate measures of compliment effectiveness and interchanging them is likely to produce different results. Explicit attitudes comparing to implicit attitudes may be more vulnerable to lack of a compliment`s sincerity. A compliment`s ability to create explicit attitudes is thus dependent on the target`s conviction that the compliment is genuine. A compliment`s ability to produce implicit or unconscious attitudes is, however, unconditional; positive implicit attitudes or opinions result from being flattered even when the target is aware of the compliments having ulterior motive (Chan & Sengupta, 2008; Cialdini, 2005). Further, as it seems that implicit attitudes are stronger and resist negative information better than explicit attitudes, we can argue that due to this function, compliments are especially persuasive (Chan & Sengupta, 2008). This revelation disproves the speculating theory of compliments being less effective if closely related to the desired behavior for their tendency to create suspicion of their
persuasive purpose. Knowing that implicit attitudes are more reliable measures of
compliments` effectiveness, this study will measure more automatic indicators of the desired behavior as the dependent variables. Consequently, it is expected that using a compliment
directly related to the desired behavior might trigger greater compliance than using a compliment not related to the desired behavior.
Moderation by Perceived Importance of the Complimented Domain
There is a reason to believe the hypothesized effects are moderated by the amount of personal importance the target assigns to the complimented domain. As aforementioned, the power of compliments may be mediated by the ability to increase the target’s self esteem. Because people feel a particularly strong desire to increase their self-esteem in domains they assign high importance to (Monahan, 1995), it is likely that compliments targeting a domain highly important to the target may be more persuasive than compliments directed at a
moderately or not important domain. Thus, higher relatedness of the complimented domain to the desired behavior may depend on the level of importance the target assigns to the highly related domain. In other words, this variable may be an important moderator of the effects of compliment relatedness on compliance. Hence it is argued (see Figure 1, for conceptual model):
H2a: Personal importance assigned to the complimented domain will make the effect of compliment relatedness on compliance in the form of rejecting or agreeing to a request
stronger.
H2b: Similarly, personal importance assigned to the complimented domain will make the effect of compliment relatedness on compliance in the form of investing more effort (time and
Method
In order to test the effects of compliments and their relatedness to the desired behavior, on compliance hypothesized above, an online experiment was conducted. During this
experiment, participants were exposed to two types of compliments, related and unrelated to the desired behavior, or to no compliment at all, and their compliance with a request leading to the desired behavior was examined.
Pretest
The desired behavior for this experiment was decided to be writing a movie trailer review, a behavior including a moderately demanding activity, writing a review, yet not requiring other kinds of effort such as reading or tryouts. Simply, the requested behavior required only watching a two minutes video and writing down an evaluation. Importantly, the relationship of the predicted mediators, self-efficacy and self-esteem, to the desired behavior was pretested. For this purpose, a separate sample of 25 participants was questioned about a) to what extent on a 7-point scale without anchors did they think that feeling highly about themselves would have increased their motivation to perform the desired behavior, and b) to what extent on a 7-point scale without anchors did they think that their increased perceived capacities would have encouraged them to try to perform the behavior. The results showed that both variables were moderately related to writing a movie trailer review (MSelf-efficacy =
4.00, SD = 1.58, MSelf-esteem = 3.24, SD = 1.74).
Based on the choice of the desired behavior, the choice of compliments was made and pretested on a different sample of 25 participants. For this purpose, three compliments from each condition were developed and presented to the participants, who were asked to indicate on a 4-point Likert scale: how flattering the compliments were, whether the participants thought they could be true in their case, and how related the complimented domain was to the
desired behavior. First, the related compliment was selected: "You have a great taste in movies!" This compliment was seen as closely related to the desired behavior (writing a movie trailer review): on a scale from 1 (not at all) to 4 (very) (M = 3.40, SD = 0.65). Based on this choice, the unrelated compliment, “You have a great taste in design!”, was selected, being seen as similarly flattering (M = 2.72, SD = .18), t(1.20) = -1.00, p = .327, as the “movies” compliment (M = 2.48), SD = 0.17), yet appropriately unrelated to writing a movie trailer: On a scale from 1 (not at all related) to 4 (very related) M = 1.74, SD = 0.75.
Thirdly, on the same sample it was pretested whether the complimented domains showed comparable variance in the importance different people assigned to them. For this purpose, how much importance the participants assigned to each of the six compliments was measured on the same 4-point Likert scale. Results showed visible variance in importance assigned to having a great taste in movies: (M = 1.88, SD = 0.73), variance = 0.53, skewness = 0.19, kurtosis = -0.97, range = 1-3). Importance of having a great taste in design showed a similar variance: (M = 2.08, SD = 0.812) variance = 0.66, skewness = 0.35, kurtosis = -0.21, range = 1- 4). A dependent t-test and a Wilcoxon Signed-ranks test indicated that there was no significant difference between how important people find having taste in design (Mdn = 8.21) and movies (Mdn = 9.55), t(24) = -0.93, p = .364; Z = -0.95, p = .342. The two compliments, “You have a great taste in movies!” and “You have a great taste in design!” were thus used to manipulate the independent variables in the main study.
The Main Study
Participants and design. 301 participants were recruited and randomly allocated to one of the 3 (compliment: related vs. unrelated vs. no compliment control group) between-subjects conditions. The potentially moderating variable (importance of the compliment domain) was measured for the two groups that received a compliment. The sample included males (27.2%) and females (72.8%) ranging in age from 18 to 78 years (M=25.61, SD=7.69)
of 50 different nationalities. No participants were excluded and none showed any missing values. Participation in this study was voluntary, anonymous and did not include any reward. Participants had a right to request any information at any time, or even withdraw at any stage of the study. All of them were informed about all these conditions in a consent form.
Procedure and measurements. To make sure the outcomes of the compliments were realistic, the compliments had to be perceived as potentially genuine. For this purpose, the goal of this research was disguised and this study was divided into three phases, respectively the cover story, manipulation and measurement phase: 1) filling out an exhaustive survey composed of questions seemingly related to personality testing, 2) receiving a compliment related/unrelated to the desired behavior (or no compliment) followed by a request, and 3) complying or not with the desired behavior.
Phase 1. The first phase served as a tool to seemingly gather sufficient information about the respondent to draw conclusions about his/her personality and hence creating
opportunity to present a seemingly trustworthy and relevant compliment. For this purpose, the cover story that the purpose of the survey was to try out a personality test newly developed by a graduate student of psychology, Taylor Fenn, was used. This name was purposely chosen to be ambiguous to avoid indication of the imaginary student`s gender and nationality, and thus avoid effects of similarity and prejudice on the target`s perception of the persuasion agent.
First, people in the social network of the researcher were contacted via email or social media and were asked to fill out a survey plus send it further to their own social circle. Next, participants received the following materials: a sheet with instructions, consent form, and a survey composed of questions and tasks representing a personality test. The 20 questions composing this phase were inspired by real online personality tests. For instance, participants were asked to characterize themselves, questioned on their preference, opinions and feelings about art, or had to go through creativity and simple logic tasks in forms of
riddles and puzzles. In total the personality test took approximately 10 minutes, which time was estimated as sufficient for the test to seem real and for the participant to feel as s/he invested a considerable effort. These data were not used in the further analysis. In the end, all participants were debriefed, at which point the real approximate purpose of the research was disclosed, and so was the real identity of the researcher. Also, the extent to which the
participants had been able to recognize the purpose of the study was assessed by an open-ended question “what do you think this experiment was about?”
Phase 2. In the second phase of the research participants were divided into three conditions: highly related compliment to the desired behavior (further related to as high relatedness condition), unrelated compliment to the desired behavior (no relatedness
condition) and the control group, and complimented with a type of compliment corresponding to the condition they were assigned to. Further, they were presented with a request to
participate in an ostensibly unrelated study, which would require writing a review of a movie trailer. The compliment-giving was performed as follows: participants in all conditions were presented with the (same) result of the personality test that was purposely composed in a way to fit to as many people as possible and did not include any flattery. The last page of these results contained only the information that the results can also report specific skills and that in the case of the specific participant, they showed that the given participant has “a great taste in movies!” (high relatedness condition) or “a great taste in design!” (no relatedness condition). In case of the control group no compliment was given.
Phase 3. All groups were equally presented with a request in the following item: “Thank you for participating in this research! Now you have completed this research, I have one extra favor to ask you: For upcoming research I need good reviews about movie trailers. Would you please write one? You would watch one movie trailer and then you would be given space to express your thoughts of it.” In reaction to this request,
participants either indicated “Yes, why not! Continue to the movie trailers” (compliance), or “No, rather not! Continue to the end of the survey” (no compliance). In case they accepted, they were presented with a trailer of the movie “Coffee and Cigarettes” and asked by a simple yes or no question whether they have had seen this movie. Next, they were provided a large space to write their review. In the final part of the study, compliance with the request was measured by the following aspects of the respondent`s behavior: First, DV1 was evaluated by how many participants from each group agreed to write the movie review and thus comply and participate in the second study at all. Secondly, DV2, the number of words they used was measured and thirdly, DV3, the number of seconds the respondents spend on the task, was assessed.
In addition, regardless of whether they complied or not, in the end of the survey participants` levels of hypothesized mediating and moderating variables, self-efficacy and self-esteem, were measured. Specifically, to measure the mediating variables participants were asked to answer on a seven-point scale from 1 to 7 without anchors the following questions: for self-efficacy, “to what extent do you feel you could help the researcher out and write a good review about movie trailers?”, for self-esteem, “to what extent do you feel good about yourself now?”, for reciprocal tendencies, “after receiving your personality test results, to what extent did you feel like doing something for the researcher?”, and for liking of Taylor Fenn, “to what extent do you think the researcher, Taylor Fenn, did a good job with this personality test?”. To measure the moderating variable of importance that people assigned to the complimented domain, participants had to evaluate on a seven-point scale without anchors how much importance did they assign to seven chosen characteristics. Besides the two
characteristics that were complimented in the study, taste in movies and taste in design, these seven characteristics included a number of extra distracting items to cover the goal of the study. These cover characteristics included: taste in fashion, creative thinking, taste in music,
being good at sports, and being a good dancer. Since their purpose was not to be measured but support the cover story, it was only important for them to have approximately comparable weight to taste in movies and design to avoid effecting perceived importance of these.
Results Data Preparation and Checks
First of all, the data were checked for distribution normality and it was concluded that the normality is abnormal. Consequently, further following analyses were chosen accordingly. Then, the compliance levels were checked: 42.5% of all participants accepted the request and complied. Among those who complied, the average word count was 25.51 words, (SD = 2.5) and average writing time was 108.28 seconds (SD = 12.6). In regards of the assigned
importance, taste in movies scored on average 4.61 (SD =1.70) and taste in design 4.10 (SD = 1.70) and they did not significantly differ, t(300) = 1.658, p = .062. Results also showed that 89.1% of the participants did not see the movie “Coffee and Cigarettes” before (M = 0.11, SD = 0.31). It must be noted that the study had an extraordinarily high drop out rate: in total, 1237 participants opened the survey, yet only 301 finished it. This extensive dropout may have been partially caused by technical problems such as dysfunctions of the video in certain Internet browsers such as Safari. Because of these dysfunctions some participants were not exposed to the video and in reaction may have decided to end the session. However, the largest number of participants dropped out during Phase 1 or after Phase 2 of the survey. Main Effect of Compliments on Primary Measures of Compliance (H1a)
Compliance was measured by acceptance or rejection of the request (dichotomous) to write a movie trailer review in the first place (DV1), number of words used in case the request was accepted (DV2), and the number of seconds spent on writing it (DV3). The first variable can be referred to simply as a primary measure of compliance and the later two can be identified as measures of effort, secondary measures of compliance.
First, Cialdini`s assumption of compliment`s increasing compliance was tested by a Chi-square test. For this purpose, the factor variable (compliment vs. no compliment) was recoded: the complimented conditions were merged and coded as 1 as opposed to the control group coded as 0. The outcome variable, whether people do or do not agree to write a movie trailer review (DV1), was left dichotomous. Results of this Chi-square test confirmed Cialdini`s conclusions and showed a weak association between exposure to compliments and whether or not people agreed to write a movie trailer review indicating that complimenting significantly increased compliance level compared to no
complimenting. The odds that the participant complied was 1.67 times higher if they were complimented than if they were not complimented Χ2(1, N = 301) = 4.04, p = .048. Considering these results, a Chi-square test of differences in whether people do or do not agree to write a movie trailer review between all the three conditions (high relatedness condition n = 104, no-relatedness condition n = 98, the control group n = 99) followed. This analysis indicated a marginally significant effect of relatedness of the compliment on compliance. Interestingly, compliance was the highest for the non-relatedness condition, Χ2
(2, N = 301) = 5.63, p = .061 (see Table 1, for Chi-square test scores).
Table 1
Results of Chi-square test: all conditions x yes/no compliance
Note: Χ 2(2, N = 301) = 5.63, p = .061.
Next, only the two compliment-relatedness conditions were compared in their
Category Compliance yes Compliance no Total N High-relatedness compliment (movies) 44 60 104 No-relatedness compliment (design) 50 48 1.863 Control group 34 65 1.042
effect on compliance by a Chi-square test. For this purpose, the no-relatedness condition was recoded into dummy variable 0. The two groups did not significantly differ in compliance, Χ 2(2, N = 301) = 1.54, p = .259. H1a was thus not confirmed; the related compliment did not lead to higher compliance than the non-related compliment condition. Interaction Effect of Importance of Complimented Domain and Compliment
Relatedness on Compliance, H2a
A logistic regression was run to test the effect of the compliment`s relatedness on whether or not participants complied to write the review (DV1), and to test for the
potential moderating role of the complimented domain`s perceived importance. For the purpose of this testing, values within the variable measuring importance assigned to the complimented domain were first recoded: participants belonging to high-relatedness condition were complimented on their taste in movies, and hence only the value indicating importance they assigned to taste in movies was used as the final value of the moderating variable (M = 4.70, SD = 0.17). The same procedure was done on the participants
belonging to no-relatedness condition (M = 4.23, SD = 0.18). Participants belonging to the control group were, simply assigned a missing system value. Next, the values were
standardized. This analysis employed only the two compliment-relatedness conditions. Next, an Omnibus Test for Model Coefficients was performed to assess the fit of three models to determine which model is the best at predicting compliance (see Table 2, for omnibus test scores): Model 1, the main effect of compliment relatedness on
compliance (H1a), Model 2, the main effect of importance of the complimented domain on compliance, Model 3, the main effect moderated by importance of complimented domain (H2a). Next, only the most fitting model, model 3, was used to analyze the effects of the hypothesized interaction (Field, 2009).
Table 2
Values used in Omnibus Coefficients Tests used to examine fit of the models
Predictor Model (df) Block (df)
Relatedness (model 1) 1.54 (1) 1.54 (1) Importance (model 2) 10.45 (2)* 8.91 (1)* Relatedness*Importance (model 3) 21.60 (2)* 20.06 (1)* Note: p* < .05.
Based on this final analysis, it was discovered that there was a positive interaction effect of compliment relatedness and the importance assigned to the complimented domain significantly predicting compliance, -2LL = 257.46, Nagelkerke R2 = .14, Cox & Snell R2 = .10, Χ2(2) = 20.06, p < .001: as relatedness and importance increased,
compliance increased as well (see Table 3, for logistic regression test scores). For not related compliment each extra point on the scale of importance increased the predicted probability of compliance by 0.09% starting on 50.75%. Interestingly, for the highly related compliments importance up to value 5 increasingly predicted no compliance (value 1 predicted rejection of the request by 6.12%, value 5 by 44.02%). Starting by value 6, however, importance predicted acceptance of the request (value 6 = 59.43%, value 7 - 73.19% probability of compliance).1 In conclusion, higher levels of importance were associated with greater persuasive effects of compliment relatedness on compliance. H2a was thus confirmed. For participants with higher levels of importance attributed to the complimented domain the related compliment was associated with greater compliance
11 The size of the odds ratio and confidence intervals was examined and it was concluded that we could be fairly
confident that the direction of the relationship we observed is valid for the population. Next, the -2 Log
Likelihood values of the models were examined and it was concluded that model including the interaction effect is a better predictor of compliance than the constant (Field, 2009). Finally, the residuals for the logistic
regression run to test the final model (3) were diagnosed and it was concluded and that this model is a valid predictor of compliance. Specifically: Cook`s distance values < 1, leverage value < twice the size of the average leverage value, less than 5% of the standardized residuals` values lie outside +/-1.96 and less than 1% of them lie outside +/-2.58, and values of DFB0_1 < 1. Yet judging from examining the classification plot, this model is far from fitting the data perfectly; there is a lot of cases too close to value 5 meaning that a lot of cases are predicted almost by chance.
than the non-related compliment. For participants with moderate or low levels of compliance, the unrelated compliment was associated with greater compliance than the related one. In fact, for those participants the related compliment was associated with no compliance.
Table 3
Results of the Logistic regression on the final model (3)
95% Cl for Odds Ratio Predictor B (SE) Lower Odds Ratio Upper Relatedness -3.38* (.99) 0.005 0.034 0.239 Importance 0.00 (.11) 0.805 1.003 1.251 Relatedness*importance 0.62* (.20) 1.263 1.857 2.732 Constant 0.03 (.518) x 1.027 x
Note: R2 = 7.66 (Hosmer & Lemershow), 0.10 (Cox & Snell ), 0.135 (Nagelkerke). Model Χ2
(3) = 21.60*, -2 Log Likelihood = 257.46, Initial -2 Log likelihood = 279.060. p* < .05.
Interestingly, no significant main effect of the relatedness by itself was found when the interaction effect was removed from the model, -2LL = 277.52, Nagelkerke R2 = .01, Cox & Snell R2 = .01, Χ2(2) = 1.54, p = .215, revealing that there is no effect of compliment relatedness by itself on compliance. These results are in line with the findings of the Chi-square analysis testing main effects of the two compliment relatedness conditions on primary measures of compliance; the compliment relatedness without enhancement of high levels of importance assigned to the complimented domain may decrease compliance instead of increasing it.
Table 4
Results of Logistic regression on model 1 (not included in the final analysis)
Predictor B(SE) 95% Cl for Odds Ratio Lower Odds Ratio Upper Relatedness -0.35(.28) 0.404 0.704 1.227 Constant 0.04(.20) x 1.042 x
Note: R2 = .00 (Hosmer & Lemershow), 0.01 (Cox & Snell), 0.01 (Nagelkerke). Model Χ2(1) = 1.54, -2 Log Likelihood = 277.52, Initial -2 Log likelihood = 279.060. p* < .05.
Main Effects and Interaction Effects on Effort, H1b and H2b
Next, the effect of the secondary measures of compliance, word count and writing time, were analyzed. First, the differences in the main effect between the control group and the complimented groups in word count were tested by an independent sample Mann-Whitney test in regards of the sample`s abnormal distribution. This analysis showed that the two complimented groups showed significantly higher effort in the form of time invested in writing a review (M = 158.21) comparing to the control group (M = 136.28), U = 8726.50, z = -2.03, p = .022, r = -.01. Encouraged by these findings, we proceeded with another Mann-Whitney test analyzing effects of two levels of relatedness on word count, yet the high relatedness group (M = 103.28) and no relatedness group (M = 99.61) did not significantly differ, U = 4911.00, Z = -0.49, r = -.01, p = 0.304.
Adding the moderator to the picture, a multiple regression was run to test the effects of compliment relatedness IV (including the two compliment relatedness levels under moderation of the importance assigned to the complimented domain as factor variable) on word count DV2 as outcome variable. A new variable was created for the interaction effect of the two predictors. Interestingly, the analysis revealed no significant main effect of compliment relatedness by itself (model 1), F(1, 200) = 2.72, p = .101, but
a significant positive effect of importance by itself (Model 2), F(2, 199) = 5.87, p = .003, on the review`s word count, (R2 = .06), b = 5.02, t = 2.99, p = .003, 95% CI [1.705, 8.339]. The model with the interaction effect of the compliment relatedness and importance (Model 3) was, however, a better predictor of review`s word count, F(3, 198) = 7.10, p < .001, b = 9.95, t = 3.02, p = .003, 95% CI [3.444, 16.455] (see Table 5, for multiple regression test scores). The revealed effect indicated that the effect of importance differs per different compliment condition. Overall, the strength of the prediction of the last model was weak: (R2 = .10). Moreover, similarly as in the results of the logistic
regression, the analysis showed that within this final model, the relatedness by itself seems to decrease compliance rather than increasing it.
For the purpose of running a two-factor ANOVA post hoc test, the values of the importance were split up by the median (Mdn = 5, 0 < 5, n = 86, 1 ≥ 5, n = 116). The test indicated that in the low importance level there was no significant difference between the two relatedness conditions. However, in the high importance level, the related compliment triggered greater word count (Mno importance = 45.05, SD = 51.72) than the unrelated compliment (Mhigh importance = 20.87 words, SD = 34.18), Mdifference = 24.18, p = .002.
Table 5
Regression models to predict word count (N = 202)
Word count Model 2 Model 3 Predictor b b Constant 1.54 22.06* Relatedness 7.53 -36.99* Importance 5.02** 0.18 Relatedness*Importance 9.95** R2 0.05 0.10 F 5.87** 7.10*** ΔR2 0.04 ΔF 9.10** Note. * p <.05. ** p <.01. *** p <.001.
Similarly, a Mann-Whitney test was run to analyze differences between the complimented groups (M = 158.21) and the control group (M = 136.28) in effort invested to writing the movie review in the form of writing time (DV3) discovering that exposure to compliments led to longer writing time, U = 8542.00, z = -2.28, r = - 0.01, p = .011. In reaction, a multiple regression was run to examine effects of compliment relatedness (including the two relatedness conditions under moderation of importance as factor variables) on writing time. This analysis revealed no significant main effects of the compliment relatedness (Model 1), F(1, 200) = 0.01, p = .911, or importance (Model 2), F(2, 199) = 2.20, p = .114, but discovered a significant positive interaction effect of the compliment relatedness and importance (Model 3) identifying it as a predictor of review`s writing time, F(3, 198) = 3.70, p = .013, b = 46.23, t = 2.56, p = .011, 95% CI [10.684, 81.774] (see Table 6, for multiple regression test scores). These results indicated that the effect of importance differed per different compliment conditions. Overall, the strength of the prediction of this model is weak, (R2 = .05). In contradiction to the findings for the word count, the results of a two-factor ANOVA post hoc test indicated that the two compliment conditions differed only in the low importance level, where the unrelated compliment was associated with a longer review writing time (M = 145.32 seconds, SD = 212.15) than the related compliment (M = 42.02 seconds, SD = 107.15), Mdifference = 103.30, p = .034. For the participants indicating high importance, the compliment conditions differed only marginally, Mdifference = 76.43, p = .069. However, the
importance did not significantly affect the effects of the unrelated compliment, Mdifference
= 43.89, p = .334, while it did significantly increase the effects of the highly related compliment, Mdifference = 135.84, p = .003, (Mhigh importance = 101.43 seconds, SD = 245.56, Mhigh importance = 177.86 seconds, SD = 264.08).
Table 6
Regression models to predict writing time (N = 301).
Word count Model 3 Predictor b Constant 136.41* Relatedness -212.20* Importance -3.40 Relatedness*Importance 46.23* R2 .05 F 3.70* ΔR2 0.03 ΔF 6.58* Note. * p <.05. ** p <.01. *** p <.001.
The findings for both measures of effort are in line with the results of the logistic regression, testing the interaction effect of the compliment relatedness and importance, and the Chi-square test assessing the main effects of the compliment relatedness alone. It should be, however, noted that for both variables the regression might have been
compromised by a lack of homoscedasticity. In conclusion, H1b is not confirmed; the compliment relatedness alone did not increase compliance in the form of invested effort. H2b is confirmed; high levels of importance attributed to the complimented domain increased the persuasive effects of the related compliment.
Mediation Effect of Self-efficacy and Self-esteem, H1c
In addition, mediation effects were tested by bootstrap analyses run through Hayes` PROCESS (2014). Four mediators were tested: two for Cialdini`s principles of persuasion currently believed to mediate effects of compliments, liking and reciprocity, and two residing from the theory of positive reinforcement, self-efficacy and self-esteem. All mediators were tested at the same time in the PROCESS model 5 (Hayes, 2014) in their ability to mediate the effect of compliment relatedness (high vs. low) on compliance moderated by perceived importance of the complimented domain. The analysis, however,
showed that there was no significant indirect effect of the compliment relatedness on compliance through all the predicted mediators: self-efficacy b = -.02, SE = .05, 95% Cl [-.17; .04], self-esteem b = -.04, SE = .05, 95% Cl [-.22; .02], reciprocity b = .05, SE = .07, 95% Cl [-.06; .27], and liking b = .03, SE = .05, 95% Cl [-.02; .23]. Hypothesis 1c was thus not confirmed.
Further analysis of the mediation effect in PROCESS similarly showed that none of the mediating variables were significantly predicted by the independent variable, compliment relatedness: Self-efficacy, F(1, 200) = 1.39, p = .239, self-esteem, F(1, 200) = .46, p = .500, reciprocity, F(1, 200) = .80, p = .373, Liking, F(1, 200) = 1.29, p = .257. Yet interestingly, a multiple regression run on all the two conditions of compliment relatedness under moderation of importance showed that interaction effect of compliment relatedness and importance significantly predict increase in self-efficacy, F(2, 199)= 17.23, p < .001, b = 0.53, t = 5.73, p < .001, 95% CI [0.347, 0.710], and reciprocity, F(2, 199) = 3.357, p = 0.037, b = 0.24, t = 2.43, p = 0.016, 95% CI [0.045, 0.436], yet not self-esteem, F(2, 199) = 2.77, p = 0.065, or liking, F(2, 199) = 1.99, p = .139. The predictive ability of the effect is fairly strong, (R2 = .15). Additional two-factor ANOVA post hoc test revealed that the related compliment in combination with high importance increased self-efficacy significantly more (Mlow importance
= 2.20, SD = 1.34, Mhigh importance = 3.78, SD = 1.76) than the unrelated compliment (Mlow importance = 2.72, SD = 1.40, Mhigh importance = 3.04, SD = 1.84), Mdifference = 0.74, p = .016.
In low importance levels the compliment conditions did not differ, Mdifference = 0.52, p = .144, and only the high related compliment was enhanced by importance, Mdifference, 1.58, p < .000. In contrast, in affecting reciprocity, the compliment conditions did not significantly differ in ether level of importance. Increase in importance, however, significantly enhanced the power of the highly related compliment to increase reciprocity, (Mlow importance = 3.53,
multiple regression test scores). This effect is weak, (R2 = .03). Importantly, the distribution of the sample is abnormal and so these regressions may be compromised by a lack of homoscedasticity.
Table 7
Regression models to predict word count (N = 301).
Interaction effect: relatedness*importance predicting:
Self-efficacy Reciprocity Predictor b b Constant 1.54 22.06* Relatedness 7.53 -36.99* Importance 5.02** 0.18 Relatedness*Importance 9.95** R2 0.15 0.03 F 17.23*** 3.36* ΔR2 0.14 0.03 ΔF 32.84*** 5.90* Note. * p <.05. ** p <.01. *** p <.001.
Also, Spearman`s correlation test showed that self-efficacy is significantly correlated with dichotomous primary measures of compliance (yes/no), r = .24, p < .001, and so are reciprocity, r = .23, p < .001 and liking, r = .17, p = .004, while the same is not true for self-esteem, r = .07, p = .196. Hence, self-efficacy shows considerable potential for mediating the interaction effect of the compliment relatedness and importance assigned to the complimented domain. Reciprocity and liking show a weak potential, and self-esteem does not show any potential at all.
Discussion and Conclusion
In this paper Cialdini`s assumption that compliments increase compliance was adopted and elaborated on by addition of a new factor, compliment relatedness to the desired behavior in interaction with the complimented domain`s assigned importance. The following research questions were raised: when applying praise as a persuasive technique to produce a certain
behavior in a person, is a compliment more effective in producing compliance if directly related to the requested behavior comparing to when not related to the desired behavior? Is this relationship moderated by perceived importance of the chosen complimented domain?
First, it was expected that a compliment would produce a greater tendency to comply with a request if the compliment was about a trait related to the requested behavior as
compared to a trait unrelated to the requested behavior. Furthermore, it was expected that this effect would be stronger if the importance assigned to the complimented trait was high. These predictions were partially confirmed by the results of the present experiment. To explain, it seems that, just as Cialdini concluded, compliments do increase compliance in contrast with usage of no compliments. Specifically, complimenting people did increase their willingness to write a movie trailer review, and that was true for their acceptance of the request and effort invested in the task. Yet, relatedness of the compliment to the desired behavior by itself influenced compliance only marginally significantly, plus in the direction opposite to the hypothesis: non-related compliments seemed more persuasive. Confirmation of these results by further research would suggest that the counter-theory claiming that related compliments should be less effective for they awake persuasion knowledge and hence resistance (Campbell & Kirmani, 2000) was correct.
Nevertheless, it was found that inclusion of the importance assigned to the
complimented domain reversed the findings described above dramatically. It was revealed that, in line with H2a, for acceptance of writing a movie trailer review, in combination with high importance related compliments seemed to perform significantly better than unrelated ones. For effort, the findings were in line with H2b, yet they were contradictory in whether the related compliment enhanced by the importance was more persuasive. For word count, the effects were identical to the primary measures of compliance, yet for the writing time,
persuasiveness confirming H2b, it did not result in the related compliment`s greater
persuasiveness. We could thus conclude that compliment receivers react to the compliments` relatedness in relation to the importance they assign to the complimented area. Specifically, in case they assign high importance to the complimented trait, they are more likely to accept a request when exposed to related compliments than unrelated compliments. However, in contradiction with hypotheses H1a and H1b, if they assign moderate, low or no importance to the complimented domain, they seem to be more persuaded by unrelated compliments.
No conclusive evidence was found to support the same conclusion for effort invested in performing the requested behavior. Nevertheless, the measurement of the effort invested in the review writing showed potential weaknesses conditioned by the design of this research. Specifically, no attention was paid to the quality of the written text, only to the quantity of it. Consequently, as the experiment was conducted online, there was no guarantee that the text was indeed written by the participant and not copied from the Internet, or that time taken to compose the text was indeed invested in writing and not other activities. In addition, the results in regards to effort were obtained violating the assumption of equal variances in the population and the group sizes were not absolutely equal, and so the conclusions in this area cannot be generalized to the population. Thus, further research should readdress the issue using a design allowing for more reliable measurement of effort.
Finally, it was hypothesized that the effect of compliment relatedness to the desired behavior on levels of compliance was mediated by increased levels of efficacy and self-esteem rather than Cialdini`s principles of persuasion, reciprocity and liking. This was predicted because compliments were hypothesized to be based on positive reinforcement of a behavior instead of liking of the persuasion agent, and hence compliments` relatedness was predicted to increase a compliment`s effectiveness. However, no conclusive evidence supporting the theory of self-efficacy and self-esteem was found, yet neither was evidence
suggesting that liking or reciprocity were mediating the effects. However, it should be noted that self-efficacy, as the only tested mediation variable, was shown to be increased by the discovered interaction effect, and at the same time was significantly associated with compliance. Hence, although there was no significant indirect effect of this variable
discovered, there are reasons to believe that self-efficacy could play a role in the tested model, a role that was perhaps missed because of the weak measurement of the mediating variables. Specifically, the levels of these variables were only assessed after exposure to the compliment and were estimated by explicit measures instead of implicit measures, which would have been more accurate. Further confirmation of self-efficacy underlying the interaction effects of the compliment relatedness and importance assigned to the complimented domain would be in line with the findings of the present study. To explain, self-efficacy mediating the discovered interaction effects would support the theory of positive reinforcement standing behind the hypotheses formulated and confirmed in the present study.
Further research should also focus on more elementary directions of compliments usage. For instance, during the pretest a lot of other issues in regards of how compliments functioned were discovered. For instance, it seems that there is incomplete evidence that could explain the extent, to which compliments are found flattering. For instance, could compliments targeting domains related to opinions of others (such as respect or the
community, looks, or friendship) be seen more flattering than compliments targeting domains invisible to others? Just like answering questions posed by this study, clarification of this issue could be useful for methodology of further research on compliments in general, but also in usage of compliments in various types of professional and social environment. Efficient application of compliments` relatedness to the desired behavior can be used for instance in persuasive doctor-patient communication in the form of encouraging the patients to mind their health conditions more carefully (Martin, 2008). Similarly, it could be used in
teacher-student communication in the form of persuading teacher-students to invest their bets effort in schoolwork (Siegle and Mccoach, 2007). At the same time it can be effective in direct marketing and personal selling.
In conclusion, we can be confident to claim that the use of compliments indeed increases compliance. In addition, strong evidence of the interaction effect of compliments` relatedness and the importance assigned to the complimented domain increasing acceptance of a request was found suggesting that the more importance one assigns the complimented trait, the more it is persuaded by highly related compliments. Moreover, fair evidence was found suggesting that this interaction effect may increase effort invested in performing the requested behavior in the same way. Nevertheless, this conclusion must be confirmed by further research. Interestingly, a possible opposite effect of compliment relatedness was discovered when the importance was excluded form the model; without assigning high
importance to the complimented domain, related compliments may even be less effective than unrelated ones. Thus the answer to the research question is: compliments highly related to the desired behavior may only be more effective in producing compliance than unrelated
compliments for those who assign high importance to the complimented trait. When using compliments to persuade, it is thus essential to consider the importance the target assigns to the complimented domain. If the target shows high importance, it is recommended to apply a compliment highly related to the desired behavior. Otherwise, using compliments highly related to the complimented domain is discouraged. However, further research using better measures of effort invested in the requested behavior needs to be done to confirm these conclusions. Unfortunately, no conclusive evidence of self-efficacy and self-esteem was found. Nevertheless there are reasons to believe that this lack of evidence may have been caused by weaknesses of the present study`s research design, and thus further research should address this question again focusing especially on self-efficacy. Further evidence confirming
self-efficacy underlying the effects discovered by the present study would support not only the effects discovered in the present study, but also the theoretical rational leading to this study`s hypotheses: the theory of positive reinforcement.
References
Bandura, A. (1993). Perceived self-efficacy in cognitive development and functioning. Educational Psychologist, 28(2), 117–148. doi: 10.1207/s15326985ep2802_3 Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W. H. Freeman.
Blase, J. & Kirby, P. C., (1992). The Power of Praise—A Strategy For Effective Principals. NASSP Bulletin, 76(548), 69-77. doi: 10.1177/019263659207654809.
Byme, D., Rasche, L., & Kelly, K. (1974). “When I like you” indicates disagreement an experimental differentiation of information and affect. Journal of Research in Personality, 8(3), 207-217. doi:10.1016/0092-6566(74)90032-4
Campbell M. C., & Kirmani A. (2000). Consumers’ Use of Persuasion Knowledge: The Effects of Accessibility and Cognitive Capacity on Perceptions of an Influence Agent. Journal of Consumer Research, Inc., 27(1), 69-83. doi: 10.1086/314309
Cialdini, R. B. (2001). Harnessing the Science of Persuasion. Harward Business Review,
79(9), 72-79. Doi: 10.1016/0092-6566(74)90032-4
Cialdini, R. B. & Goldstein, N. J., (2004). Social influence: compliance and conformity. Annual Review of Psychology, 55, 591(31).
Cialdini, R. B., & Sagarin, B. J. (2005). Principles of interpersonal influence. In T. C. Brock & M. C. Green (Eds.), Persuasion: Psychological insights and perspectives, 2, 143-169. Thousand Oaks, CA: Sage.
Culp, K., (2001). Effective Motivators for Master Volunteer Program Development. Journal of Extension, 39(2 ). Retrieved from:
http://www.joe.org/joe/2001april/rb4.php.
Davidson, T. W. (2008). Six Principles of Persuasion You Can Use to Influence Others. Physician executive, 34(5), 20-24.
Falk, A. & Fischbacher, U., (2006). A theory of reciprocity. Games and Economic Behavior, 54(2), 293–315. Retrieved from:
http://dx.doi.org.proxy.uba.uva.nl:2048/10.1016/j.geb.2005.03.001
Farber, P. D., Khavari, K. A., & Douglass, F. M. (1980). A factor analytic study of reasons for drinking: Empirical validation of positive and negative
reinforcement dimensions. Journal of Consulting and Clinical Psychology, 48(6), 780-781.
Fiedler K., & Mata A. (2013). The art of exerting verbal influence through powerful lexical stimul. In Forgas J. P., Vincze O., & Janos L. (Eds.), Social Cognition and
Communication. London, UK: Taylor & Francis. Retrieved from
http://www.sydneysymposium.unsw.edu.au/2012/chapters/FiedlerEASP2012.pdf Fishbein, M., & Cappella, J. N. (2006). The role of theory in developing effective health
communications. Journal of Communication, 56, S1-S17.
Hayes, A. F., (2014). Some Frequently Asked Questions. Retrieved from: http://www.afhayes.com/macrofaq.html.
Hayes, A. F., (2014). Supplementary PROCESS documentation. Retrieved from http://www.afhayes.com.
Hill, M., Budkley, W., & Buckley, N. K. (1968). The use of positive reinforcement
in conditioning and attending behavior. Journal of Applied Behavior Analysis, 1(3), 245–250. doi: 10.1901/jaba.1968.1-245
Isen, A. M., & Levin P. F. (1972). Effect of feeling good on helping: cookies and kindness. Journal of Personality and Social Psychology, 21(3), 384-388. Knapp,M. L., Hopper, R., & Bell, R. A., (1984). Compliments: A Descriptive Taxonomy.
Journal of Communication, 34(4), 12–31. doi: 10.1111/j.1460-2466.1984.tb02185.x LeBourveau, C. A., Dwyer, R. F., & Kernan, J. B. (1988). Compliance strategies in
direct response advertising. Journal of Direct Marketing, 2(3), 25-34. Doi: 10.1002/dir.4000020306
Martin, S. (2008). The science of compliance: the first of six articles on social influence will examine the concept of social proof. (communication skills). Practice Nurse, 35(1), 38-40.
Michael, J. (1975). Positive and Negative Reinforcement, a Distinction That Is No Longer Necessary; Or a Better Way to Talk about Bad Things. Behaviorism, 3(1), 33-44. doi: 10.1300/J075v24n01_15
Monahan, J. L. (1995). Thinking positively: Using positive affect when designing health messages. In E. W. Maibach & R. L. Parrott (Eds.), Designing health
messages: Approaches from communication theory and public health practice (pp.81-98). Thousand Oaks, CA: Sage.
Mueller, C. M., & Dweck, C. S. (1998). Praise for Intelligence Can Undermine Children's Motivation and Performance. Journal of Personality and Social Psychology, 75(1), 33-52. doi: 10.1037/0022-3514.75.1.33
Pratkanis, A. R., (2007). A Social influence analysis; an index of tactics. In Pratkanis, A.R. (Ed). The science of Social Influence. Advances and Future Process. 30-40.
Schwartz S. H. (1999). A Theory of Cultural Values and Some Implications for Work. Applied psychology: an International Review, 48(1), 23–47. Doi: 10.1111/j.1464-0597.1999.tb00047.x
Siegle, D, & Mccoach, D. B., (2007). Increasing Student Mathematics Self-Efficacy through Teacher Training. Journal of Advanced Academics, 18(2), p.278-312. Simons, H. W., Morreale, J., & Gronbeck, B. E. (2001). Persuasion in society. Thousand
Oaks, CA: Sage.
campaigns. American Journal of Public Health, 78(2), 163-167.
Suls, J., Martin, R., & Wheeler, L. (2000). Three kinds of opinion comparison: the triadic model. Personality Social Psychological Review, 4, 219-37. Doi: http://dx.doi.org/10.1207/S15327957PSPR0403_2
Visser, C.F. (2012). The Solution-Focused Mindset: An Empirical Test of Solution-Focused
Assumptions. Retrieved from
http://www.m-cc.nl/The%20Solution- Focused%20Mindset%20-%20an%20empirical%20test%20of%20solution-focused%20assumptions.pdf
Weinstein, R., Tosolin, F., Ghilardi, L.,, & Zanardelli, E., (1996). Psychological intervention in patients with poor compliance. Journal of Clinical Periodontology, 23(3), 283–288. doi: 10.1111/j.1600-051X.1996.tb02090.x. Wolford, M., Cox, K., &
Appendix
Table 8
Results of Logistic regression on the interaction effect of two levels of
relatedness*importance on the primary measures of compliance for model 2 (not included in the final analysis)
Predictor B (SE) 95% Cl for Odds Ratio Lower Odds Ratio Upper Importance -1.02(.42) 0.346 0.616 1.097 Constant 0.25(.09) 1.084 1.284 1.521
Note: R2 = 18.29 (Hosmer & Lemershow), 0.05 (Cox & Snell ), 0.07 (Nagelkerke). Model X2 (2) = 10.45*, -2 Log Likelihood = 268.61, Initial -2 Log likelihood = 279.060. p* < .05.
Survey used in the experiment Thesis project: personality test - moderator
Q1 Thank you for participating in this survey!
My name is Taylor Fenn, I am a psychology student at the University of Amsterdam, and this survey is part of my thesis research project called Social Psychology Research. The purpose of this research is to pretest a personality test I developed. I greatly appreciate your help with this project!
Before you proceed to the experiment itself, please go through the following information about the experiment you are about to engage in and the informed consent form:
Q2 Informed consent for participation in Thesis Research Project. The experimental research project that is being carried out under the auspices of the ASCoR research institute, a part of the University of Amsterdam. ASCoR conducts scientific research into media and communications in society. You will receive no material reward In
recognition of your assistance.As this research is being carried out under the
responsibility of the ASCoR, University of Amsterdam, we can guarantee that:1. Your anonymity will be safeguarded, and that your answers or data will not be passed on to third parties under any conditions, unless you first give your express permission for this.2. You can refuse to participate in the research or cut short your participation without having to give a reason for doing so. You also have up to 24 hours after
participating to withdraw your permission to allow your answers or data to be used in the research.3. Participating in the research will not entail your being subjected to any appreciable risk or discomfort, the researchers will not deliberately mislead you, and you will not be exposed to any explicitly offensive material.4. No later than five months after the conclusion of the research, we will be able to provide you with a research report that explains the general results of the research.For more information about the research and the invitation to participate, you are welcome to contact the project supervisor at any time: Annemarie Wennekers PhD, ASCoR, University of Amsterdam, Kloveniersburgwal 48, 1012 CX Amsterdam; A.M.Wennekers@uva.nl. Should you have any complaints or comments about the course of the research and the procedures it involves as a consequence of your participation in this research, you can contact the designated member of the Ethics Committee representing ASCoR, at the following address: ASCoR secretariat, Ethics Committee, University of Amsterdam,
Kloveniersburgwal 48, 1012 CX Amsterdam; 020-525 3680; ascor-secr-fmg@uva.nl. Any complaints or comments will be treated in the strictest confidence.
If you agree with these conditions, please indicate so on the next page and proceed to the experiment:
