Development and validation of implicit association test for propensity to be engaged in manipulative communication

Размещено на http://www.allbest.ru/

Running Head: DEVELOPMENT AND VALIDATION OF IMPLICIT ASSOCIATION TEST FOR MANIPULATION

FEDERAL STATE AUTONOMOUS EDUCATIONAL

INSTITUTION OF TERTIARY EDUCATION

NATIONAL RESEARCH UNIVERSITY «HIGHER SCHOOL OF ECONOMICS»

FACULTY OF SOCIAL SCIENCES

DEPARTMENT OF PSYCHOLOGY

MASTER'S PROGRAM «APPLIED SOCIAL PSYCHOLOGY»

Development and validation of Implicit Association Test (IAT) for propensity to be engaged in manipulative communication

Student

Kovalenko Elena

Scientific supervisor

Eliseenko A.S.

Moscow 2018



Table of contents

Introduction

Chapter 1. Theoretical background

1.1 Models of communication in social psychology. Review of the main communication strategies

Chapter 2. Development and validation of Implicit association test for propencsity to be engaged in manipulative communication

Refferences

Abstract

Appendix



Introduction

Understanding and predicting the behavior of human beings is a matter of vital importance in psychology. Due to the fact, that human behavior is highly mediated by situational factors, personal characteristics and presence of others, psychologists came up with many theories, which could potentially predict the outcome, knowing the predispositional factors. Attitudes (Allport, 1935), Balance theory (Heider, 1946), Theory of Reasoned Action (Ajzen, 1989), Theory of Planned Behavior (Ajzenand and Fishbein, 2005) give us a tool to predict the behavior in different situations. All of these theories have one common characteristic - they consider human behavior a conscious process, which is fully controlled by a person, and thus they focus on explicit processes and disregard implicit ones.

Implicit processes refer to unconscious influences such as attitudes, perception and memory that influence people`s behavior, even though they have no conscious awareness of those influences (Reingold and Eyal; Ray and Colleen, 2006). Studies have shown that implicit processes have a significant impact on addictive behavior (Payne, Govorun, Arbuckle, 2008), voting behavior (Friese, Smith, Plischke, Bluemke, Nosek, 2012; Raccuia, 2016) and investment behavior (Chassota, Klцcknerb, Wьstenhagena, 2015).

The same implicit processes are involved in how interact with one another (Wegner, Daniel; Vallacher, Robin, 1987). A large portion of interaction is interpersonal communication. The main goal of every communication process is to change behavior of another, that is why predicting behavior of an individual is impossible without taking into account it`s interaction with others. Communication involves informative (the process of information exchange), perceptive (how people perceive each other) and interactive (the organization of interaction) levels. The change of behavior of another could be reached only if communicator uses the right strategy at each level. Depending on the desirable outcome different strategies could be used: imperative (being an authoritarian and commanding), manipulation (using a hidden influence) and dialogue (using open and equal conversation) (Andreeva, 1980).

In this research we would like to focus on the propensities to be engaged in manipulation. We assume that a propensity to use a certain strategy of influencing is not fully controlled and could be demonstrated in situations, which are characterized by a lack of conscious control. Moreover, manipulation is a type of behavior, which is seen as negative and socially undesirable feature of a person. So, when people are asked to evaluate themselves on the scale of manipulation, they usually give dishonest answers. That is why we feel the need to create an instrument, which will minimize the probability of socially-desirable answers. The implicit-association test (IAT) is designed to measure the speed of automatic associating between mental representations, which indicate the strength of relationships between these concepts in memory. The usage of IAT allows to avoid social desirability (Greenwald, Anthony; McGhee, Debbie; Schwartz, Jordan, 1998) that is a matter of vital importance in testing negative characteristics, which people don`t want to expose. Thus this technique should be appropriate for our goal.

The aim of this research is to create and validate the IAT on propensity of people to be engaged in manipulative communication. We expect that this test will have a high power of prediction of actual people`s behavior. Moreover, it will help HR-specialists to control the promoting process - it will help to distinguish people who do better job because of their skills and those, who are manipulating others and ascribe their success to themselves.

Our research question is the following:

1. Will the IAT work better, than explicit testing in predicting manipulative behavior?

The novelty of this research is that there is no implicit association test on manipulative communication and that IATs have never been used for the task of personnel selection and assessment process. We expect useful and substantial results from IATs in this sphere because nowadays employers are very often interested in the behavior of an employee under stress- or critical conditions - the situations, where the level of conscious control is low. Thus the usage of IAT will allow employers to detect during the personnel selection process the propensities to manipulation, which are a significant factor a decision.



Chapter 1. Theoretical background

1.1 Models of communication in social psychology. Review of the main communication strategies

Communication is a process of interaction among people and their environment. People use language as an instrument of transmitting information, ideas and attitudes. Communication might be viewed as interpersonal process, which has several goals: provoke changes in behavior, express thoughts and feelings. Changing behavior could be performed by exercising control over actions of other people or by facilitating decision making through giving the specific information in the needed dosage. This kind of activities is usually classified as incentive or provoking behavior, which in certain circumstances might transform into manipulation. There are a lot of classifications of communication. In order to work with different types and processes in communication we need to consider different models built in psychology.

All communication systems, regardless of how simple or complicated they may be, operate on the same principle: signals transmit messages from a source to a destination with the help of a specific tool reaching certain effect. 5-factor Lasswell's model of communication emphasize the general principle of communication. He distinguished the main concepts of analysis: sender, receiver, message, channel and effectiveness. As Lasswell (1948) wrote:

“who” refers to the communicator who produces a message; “what” is the content of a message; “channel” indicates how the transmission was mediated; “whom” describes either an individual recipient or the audience of mass communication; “effect” is the outcome of the message» (p. 119).

During the act of communication, it is not just the movement of information that takes place, but the mutual transmission of coded information between two individuals, the subjects of communication. But people do not just exchange values, they try to work out a common meaning. For that information should be not only accepted, but also comprehended. Communicative interaction is possible only if the person sending the information (communicator) and the person receiving it (the recipient) have a similar system of codification and decoding of information. Thus, we have highlighted the main requirements for successful communication, as it is a mechanical process. Further we would elaborate on other classifications that could make communication less effective or even break it.

Second, we could specify levels of communication. Communication involves informative (the process of information exchange), perceptive (how people perceive each other) and interactive (the organization of interaction) levels (Andreeva, 1989). Each level provides a subject an ability to be understood and too influence another in a desirable way. The difference in the levels lead to communicational problems.

The third categorization is formal and informal communication. Informal communication is not defined accurately and often considered to be residual concept. According to this approach, informal communication is a set of actions, remained after eliminating the coordination of rules and hierarchies. Simply put, informal communication is spontaneous, without rules and norms, defined by the institute or any other formal organizations. Informal communication is characterized by the lack of pre-specification. Information is not prepared in advance and then sent; rather, information is often exchanged interactively, through meetings and conversations, and courses of action are worked out in the context of the circumstances into which the actions must fit. The role of informal communication in successful performance is proven in the organizational psychology (Kraut, Fish, Root, Chalfonte, 2002).

The last model of communication is transactional theory by E. Berne. He created the theory of transactional analysis as a way to explain human behavior (Berne, 1981). Berne developed the idea of social transactions that occur among three ego-states of an individual, which he called Parent, Adult, and Child. He then investigated communications between individuals based on the current state of each that were called transactions. This model elaborates on the principal of uneven communication that could take place in interactions and explains the problems that occur in this interaction. We believe that this approach could be the basis of understanding layer of communication.

In conclusion we should mention that described models of communication allow us to identify the main points that should be analyzed in every communication: sender, receiver, message, channel, effectiveness, level, degree of formality and transactional layer. In this paper we will focus on deceptive form of communication and we will analyze it thus conceptualizing this phenomenon.

Manipulation as a specific form of communication

The major focus of our work is communicative influence, which typically takes the form of argumentation, persuasion or manipulation. There is no clear boundaries among these concepts (Oswald, 2010), it is almost impossible to fully capture their distinctive features. For instance, some researchers posit that persuasion is a type of manipulation (Galasinski, 2000), while others (Blass, 2005) indicate the difference in intention of a subject. In fact, in most instances manipulation has been regarded as a form of deficient communication in some respect. Authors point out the ill-formed nature of manipulative utterances on various - not necessarily compatible - levels. Therefore, to work with this phenomenon, we should start with its definition.

The topic of manipulation is popular in all spheres of psychology. For example, Sergei Georgievich Kara-Murza speaks of manipulation as the programming of the opinions, aspirations, moods and mental state of the masses to provide such behavior to the people that the manipulator needs. Harriet Breaker, an American clinical psychologist, defines psychological manipulation as a type of social influence that is directed to change the perception or behavior of other people with the help of hidden, offensive or misleading methods. Roberts Chaldini also speaks of manipulation as a hidden influence on person or group of people to achieve their goals, with such an impact may have a complicated multi-stage structure, which further leads us away from understanding the motive of manipulation and the very fact that we are being manipulated. However, these approaches are not systematic, they don`t take into account all the necessary elements of manipulation, thus they don`t provide us with clear understanding, and what is more important - they don`t allow us to retrieve manipulation and analyze it as a phenomenon.

On the contrary, D. Maillat (2013) gives a model of manipulation which includes three parameters: manipulative arguments, tools and intention of a manipulator. Parret (1988) studied the manipulation as a form of a speech act, taking into consideration only the content of a communicative message. He postulated that manipulation could not be analyzed only by the speech act, because this type of communication will lose its function when identified (the discussion in Parret, 1988).

Finally, scholars underlined the imbalance between communicator`s goals and recipient`s retrieval of his intention (Attardo, 1998; Rigotti, 2005). Thus, a manipulator hides his true intentions and makes a recipient believe that he is acting on his will. However, we should not forget the influence of a manipulator`s personal traits on the process. It means that manipulation could not be achieved only by set of utterances independently of the speaker who produces them.

As a result, manipulation presupposes influence that is any behavior of one individual that makes changes in behavior, attitudes, and sensations of another with definite goals, known only to the subject. Also, manipulation involves creating the illusion of the independence of an object from the impact of outside influence, the illusion of freedom of decisions and actions. Moreover, we believe that we should take situational factors as influencing agents. Thus, the complete metaphor of psychological manipulation contains four most important features:

1. A subject exercise influence on the second party
2. In manipulation process two parties are unequal - manipulator considers himself higher and lead the situation

3. An illusion of independence of decisions and actions of the addressee of the impact. His influence is indirect and usually undetectable for the addressee. That is why a true manipulator always seem a nice person to others.

4. The manipulation is always negative ethically, because a manipulator is acting on his own motivation, in the expenses of another person; but it can have positive outcomes for the addressee if the motivation of manipulator and manipulated party are the same or similar

5. The manipulator possesses certain personality traits, attitudes and skills in performing the methods of action

6. The situational factors could mediate or/and moderate the process of manipulation.

7. Manipulation could be in a form of propensity - unconscious predisposition to use manipulation as a form of influence; or in a form of a skill - a conscious decision (which requires specific training to implement) to use manipulation tactics in order to achieve goals.

Manipulation is a complex phenomenon, which allows a subject receiving goals at the expenses of an object. Such situations are common in modern companies, where the level of competition is very high, and some people could be promoted because they ascribe somebody`s else work to themselves. However, we cannot posit whether they do it on purpose. For example, in situations of low behavioral and conscious control, such as stressful or pressured situations, people can behave not according to their beliefs, but according to predispositions or learned patterns, thus the implicit system will be activated. Thus, in our research we will focus on predispositional factors, which could enable to detect people who will manipulate others without conscious intention in a disposing situation.

Empirical research on manipulation

We could specify three major focuses of empirical research of manipulation: massive manipulation (propaganda), manipulation of consumer behavior and negative consequences of manipulative workers. Three of them try to suggest ways how to avoid being influenced by manipulators and describe the mechanisms involved. However, these studies do not accentuate the rules of distinguishing influence into manipulation, as manipulation is not defined the same in them.

The first and perhaps still the most influential part of empirical research concerning manipulation are dedicated to the studies of massive manipulation or so-called propaganda. But later these studies changed their focus to persuasion. We can see that change in shifting terms - after World War II the term “propaganda” changed to “persuasion”, but the main focus of those studies remained the same. Works such as Lasswell's demonstrated the effects of propaganda as an emotional fear. The stimulus-response model, derived from learning theory served a basis for Lasswell`s work. This approach posits that people respond to media in «uniform and immediate» ways (Hardt, 1989; Lang 1989; Sproule, 1991).

Another sphere of research is consumer behavior (Tanski, 2004; Bompard and Baranowski, 2007, Stokes, 2009; Danciu, 2014). In order to get more profit, companies wanted to manipulate consumer`s behavior. A great role in the consumers' manipulation had the manipulation by advertising due to the fact that it helped to promote new products and improve the sales` rates.

As for manipulation at a work place, there were number of research that investigated the role of Machiavellianism in carring out job responsibilities (e.g., Granitz, 2003; Gunnthorsdottir, McCabe, and Smith, 2002; Hegarty and Sims, 1979; Tang and Chen, 2007). For example, Jonason, Slomsky and Partyka (2011) have investigated the impact of Dark Triad (Machiavellianism, narcissism and psychopathy) on work-related behavior. They established that high levels of Machiavellianism and narcissism led to the choice of soft tactics of influence, while Psychopathy and Machiavellianism were correlated with hard ones. Thus, manipulation in a work-place can be expressed in both hard and soft skills of influential communication. Scherer et al. (2013) demonstrated that counterproductive behavior at work could be predicted not only with Dark Triad, but also with sub-clinical psychopathy, that characterized by high rates of manipulation usage.

Hereby, we can see that manipulation is a major interest in psychology and the research area of manipulation is quite wide. However, there is no clear evidence of measurements of personal predisposition or propensity to manipulate others. This test could be beneficial in HR sphere. To create the most appropriate measurement for personnel selection, we decided to use Implicit Association Test.

Implicit and Explicit psyche

The first milestone of discovering unknown, subconscious and uncontrollable by a subject processes was the study of Sigmund Freud. Although, he was criticized for being unscientific, later studies were influenced by his concept, which investigated the complex nature of factors, underlying human behavior, dividing them into two categories.

The common name for these theories is dual system models. In general, they include two autonomous, but interconnected systems that work together to provide a human behavior. The main difference between two systems is in their content - one system is fully reasonable, conscious and controlled by the individual, while the other system is unconscious and uncontrollable (Smith, DeCoster, 2000). The earliest models defined those systems as central and peripheral by analogy with neural system (Petty, Cacioppo, 1986); later they were distinguished to rational part, which uses laws and rules, and experiential, which works on the basis of experience and learning (Epstein, 1991).

Smith and DeCoster posited that all models are the same in the sense of their understanding of processes involved in the systems. The conscious part operates on the basis of rules, structured by language and logic. Operations could be remembered from the first usage and they are maintained in conscience entirely and incrementally. The unconscious part uses associative connections, formed in the process of learning by repeated trials of adjacent or similar elements, operations start automatically and they are inaccessible for a subject - he is not aware these processes. Thus, the dual system models benefit the general understanding of psyche and allow using more complex and absolute approach to human behavior.

In this work we will use reflective-impulsive model of Strack and Deutsch (Strack and Deutsch, 2004). We believe that this model is appropriate for our research, because it explains social behavior as a joint function of reflective and impulsive processes. It suggests the idea of how two models interact and the connection of the model with behavior. For that authors suggest understanding of systems according to information processing, information storage and how they trigger behavior.

Reflective system in this model is fully exposed to consciousness, it operates on rules and laws and behavior occurs from the process of decision making and forming intentions. The reflective system is capable of forming propositional representations by connecting one or more elements through the instantiation of relational schemes to which a truth value is attached. When known and perceived information are compared, the system activates the certain behavioral pattern. For example, identifying an object as a familiar one requires a subject to have knowledge about familiarity in general and correlating an object to this general concept. But in real life it is not the way it works - children can distinguish familiar objects from unfamiliar ones at a very early age. That is why we could assume that this process is executed by the impulsive system.

Impulsive system represents the set of elements with associational connections. After a perceptual input an individual is prone to behaving the fixed way that was formulated earlier. As described in James' (1890) ideo-motor principle, «a behavior may be elicited without the person's intention or goal» (Lotze, 1852). When a person faces any situation in his experience, the bonds between stimulus and reaction occur, creating a network of associations. In this network, elements are located in activation templates. According to the models of Smith (1998), the relationships between elements are stable once it appears and it does not change throught life without specific training. Moreover, every bond has different strength, meaning that some elements are tight together closer, than others. It follows in argument, that different stimuli activates closer elements first, resulting in a certain behavior and only then, if needed, a person can correct his behavior by activating consciously other reactions that he has in his associative storage.

The activation of elements on the network can change rapidly. Availability of content will be increased due to the frequency and duration of pre-activation. Higgings (1996) called it «activation potential», meaning that no additional activation is needed in order for extracting or further processing. In general, links are created or amplified if incentives are presented or activated in close temporal or spatial proximity. Thus, associative clusters are formed, in which information about frequent correlations between stimuli and simple motor reactions or even complex behavioral patterns are contained. Due to the fact, that those responses are held in impulse system, a person could not consciously reach them as they do not contain any cognitive logic - there are no reasons for such multimodal correlations.

In essence, the associative store of the impulsive system works as a simple memory system (Johnson and Hirst, 1991), which slowly forms stable, non-propositional representations of typical environmental properties in many learning processes (McClelland et al. , 1995, Smith and DeCoster, 2000). The main advantage of this system is that it is very fast and does not need additional recourses. Thus, it is primarily used when a person does not have enough time or resources, e. g. attentional to perform. As a disadvantage, impulse system has low flexibility. Moreover, a bias may occur when elements become connected due to reflective operations, resulting in connections between elements that are not connected in reality (Smith and DeCoster, 2000). Thus, impulsive system acquires semantic concepts through propositional categorizations. Those semantic concepts will reflect typical, almost reflex behavior of a respondent or his attitudes. In order to investigate them, Implicit Association test was developed.

Implicit Association Test

Traditionally attitudes were measured by self-report methods, usually by semantic differential (e. g. Osgood et al. , 1957). Sometimes scholars combined direct methods with indirect ones, for example projective tests. The revolutionary method was proposed by Greenwald, McGhee and Schwartz (1998). They developed Implicit Association Test, which was supposed to be less forgeable and more predictive compared to self-reports.

Implicit association tests (IATs) measure the ease of association between concept and its category. The amount of time needed to perform association is considered evidence for an implicitly-held attitude toward the concept. The IAT could be used to measure a variety of attitudes including race, gender and political constructs (Nosek, Greenwald, and Banaji, 2005). The IAT measures the strength of relationships between concepts (e. g. , Asian people, gay people) and evaluations (e. g. , good, bad) or stereotypes (e. g. , smart, fast). The main idea is that when categorization is faster, it represents the actual thinking of a respondent, while slow categorization is caused by other phenomena, such as Social Desirability (See “The effects of social desirability in measurements”).

The number of studies was conducted to explore the power of implicit testing. The IATs were tested and validated as a mean to contrast established groups, such as phobic patients versus control participants (Teachman, Gregg, and Woody, 2001). Moreover, IATs were proved to predict behavior, such as acts of discrimination (McConnell and Leibold, 2001). The scoring algorithm by Greenwald, Nosek, and Banaji (2003) complements the validity of this procedure.

Perugini (2005) has compared explicit and implicit attitudes and how well they can predict behavior, using IAT. He postulated that there were three main models of interpreting explicit-implicit relationships:

(a) additive (the two types of attitudes explain different portion of variance in the criterion), (b) double dissociation (implicit attitudes predict spontaneous whereas explicit attitudes predict deliberative behavior), and (c) multiplicative (implicit and explicit attitudes interact in influencing behavior) (p. 29)

He demonstrated that in different spheres all three models might be relevant. Thereby, it is a matter of vital importance to compare explicit and implicit measurements in pioneer research to reveal the mechanism in the particular situation.

In the process of validation it was found that the results of IAT weakly correlate with explicit measures. Correlations between implicit measures of the Big Five personality traits and similar explicit ones, as derived from the NEO-FFI reached from . 18 to . 41, as demonstrated by Grum and von Collani (2007). Other studies have also uncovered moderate correlations (e. g. , Steffens and Buchner, 2003). Because of this we assume that in our research the results of IAT will not or will weakly correlate with the result of the explicit test - MAC-scale.

What is more important is that the results of IATs are hardly distorted. Asendorpf, Banse, and Mucke (2002) showed that participants failed to disguise their personality traits, even if they were instructed to do so. The same result was obtained with attitudes (e. g. , Kim, 2003). Nevertheless, there are some doubts in this assumption (e. g. , Steffens, 2004; Fiedler and Bluemke, 2005; McDaniel, Beier, Perkins, Goggin, and Frankel, 2009).

Steffens (2004) showed the possibility of small distortion of results, if participants were asked to fake them. Later, participants were trained to change their results consciously, by being tested several times with short explanation of the results. After these trials, results were distorted more, but they still represented the initial categorizations. Thus, Steffens concluded that implicit measures are less susceptible to falsifications by respondents, than explicit ones, that went along with previous results of other scholars, positing that it is almost impossible for a subject to consciously change the result of IAT (e. g. , Greenwald, McGhee, and Schwartz, 1998; Egloff and Schmukle, 2002). Fiedler and Bluemke (2005) showed the actual possibility to improve results of certain desirable personality traits while using IAT. Their participants received instructions on how Implicit Association Test works and how subjects can manipulate their answers to receive a certain result. After the specific guidance, participants were able to change the result of Extraversion, but the level of Consciousness remained the same. When McDaniel, Beier, Perkins, Goggin, and Frankel (2009) replicated the study, they received the same results. Thus, some characteristics might be inflated with specific instructions, while others will remain the same even if a respondent knows how to manipulate his answers.

Considering all these facts, we believe that IAT could be the better predictor of participant`s behavior. Thus, the results of IAT should correlate with the results of 360° questionnaire, while the results of the MAC-scale will not correlate or correlate weaker than IAT`s.

General approaches to measure manipulation

Manipulative behavior in communication is quite a widespread phenomenon. Although there was a lot of research in this area, there is still no clear way to find out whether a certain person will manipulate others. There are quite a few explicit tests on the predispositional factors, but they are broader in their conceptual understanding of manipulation. Moreover, they all suffer from the problem of distorted answers, due to the fact that they are based on self-report and are vulnerable to social desirability. The most popular ones are MAC-scale and Bant`s scale.

Machiavellianism

Perhaps the first person who comes to mind when speaking about manipulation is Niccolo Machiavelli. A famous politician and writer who claimed the policy of influencing others to obtain your own goals, was later immortalized in psychology in the form of the MAC-scale test.

The first research on interpersonal manipulation was carried out in the late 20th century (Szasz, 1961; Shostram, 1967; Bursten, 1983). Richard Christie developed the idea of personality orientation that allows being more successful in manipulating others. Christie described his theoretical construct of Machiavellianism and published his scale development techniques in Studies in Machiavellianism (Christie and Geis, 1980). Later the outcomes of this trait were examined by a variety of studies (Braginsky, 1966; Hacker and Gaitz, 1980; Bochner and Tucker, 1981; Bochner, et. al. , 1982).

Christie scanned Machiavelli's The Prince and The Discourse for items which could be used to measure the construct. In addition, he added other items, related to the concept. As a result, he created a questionnaire with 71 items, the later analysis of which showed that only three items failed to distinguish between high and low scores. Nowadays, there is an adapted and validated Russian version of MAC-scale by Znakov, consisting of 20 questions (Znakov, 2000).

According to Christie and Geis, people who score low on scales of Machiavellianism have high scores on social desirability scales, meaning that correlation between the results of MAC-scale and Social Desirability scales has high negative coefficients (r > -.60) and is significant (p < .05). In the process of validation Znakov has used the Marlowe-Crowne Social Desirability Scale to reveal dishonest answers. He has demonstrated the negative correlation between his testing material and the Marlowe-Crowne`s scale (r = -.38, p < .05). However, the power of the coefficient is very low and there was no clear conclusion whether social desirability is eliminated from the results. Thus, we believe that the bias may still occur in self-report measurements of manipulation.

Machiavellianism scale is an explicit measure, with its disadvantages, inherent to all explicit measures of a phenomenon, viewed in society as a negative one. First of all, these tests are pretty obvious for a participant, who can fake the results - for example, Mach II has questions about honesty and moral principles that the majority of people answers in a socially desirable way. Moreover, such tests have a low predictive power of behavior, because they do not show true intentions of a respondent.

The Bant`s scale is another version of Mach scale, which is popular in Russia. It has the pole of 20 questions that measure tendency to manipulation. This test has even more questions - there is no data of its validity and reliability, together with the information how it was tested. Moreover, questions of this test have the same problems as Mach - they are obvious and forgeable.

Social Dominance

Social dominance theory was developed by Jim Sidanius and Felicia Pratto in 1999. The main idea of this theory is that every society tends to have structure that might be viewed as a set of groups arranged hierarchically. These groups are organized by different social categories, such as age, gender, social class, etc. and are formed due to the general organization of society. Those people, who belong to a powerful group dominate those, who are in different position of power. However, every individual possesses a characteristic that can influence his position in society, called social dominance orientation (SDO).

Social dominance orientation shows how an individual perceive hierarchical orientation, whether there is a «preference for unequal relationships among categories of people» (Sidanius and Pratto, 1999, p. 743). Generally it reflects the desire of an individual to have equal or hierarchical relationships within in-group. The higher the score of SDO is, the more a person will be centered on enhancing of hierarchical relationships. People with high SDO prefer jobs and activities that allow them to dominate others, while people with low SDO try to attenuate policies that dictate unequal relationships. Ligneul, Girard and Dreher (2017) have shown in their study that social dominance is highly correlated with the sensitivity to social ranks that anterior dorsolateral prefrontal cortex has.

Later studies have shown that SDO can predict attitudes toward certain groups (Heaven, 1999; Bassett, 2010; Whitley and Жgisdьttir, 2000; Sanders and Mahalingam, 2012). In this sense social dominance scale is similar to implicit association test. However, implicit measures deal with processes that are not visible for a subject directly, that might be activated only when explicit ones are not able to be active or when they are not needed. Therefore, social dominance orientation might be used combined with implicit association test, but not instead.

In our research the main concept is manipulation. Even though the social dominance theory might seem as a similar construct and there are a lot of studies investigating the connection of dominance (e. g. Dominance and Prestige Scale) with dark triad, there is no significant research on the topic of Social Dominance, to the author's knowledge. Their connection is further to be investigated. Therefore, we will use the approved test of Machiavellianism, which is often used to measure manipulation.

The effects of social desirability in measurements

Social desirability (SD) is a common response bias that represents the tendency of respondents to answer questionnaires in the most socially favorable way. Due to the desire of self-protecting, self-esteem maintaining and making positive impression on others, subjects misreport their data. The phenomenon of Social desirability was found predominantly in self-report measures (e. g. Levy 1981; Peltier and Walsh 1990; Robinette 1991; Simon and Simon 1985; Zerbe and Paulhus 1987). The major influence this bias has on the scales, that measure negatively perceived concepts, e. g. manipulation, because it is not “correct”, according to social norms to expose this characteristic (Maecoby and Maccoby 1954). Therefore, any questionnaire or assessment-method should be aware of this phenomenon.

Social desirability undermines the vital criterion of a questionnaire quality - validity (Huang et al 1998). An instrument is valid if it accurately measures what it aims to measure (Beanland et al 1999). According to Nederhof (1985) «between 10% and 75% of the variance in participants' responses can be explained by SDR which can confound relationships among the variables of interest by suppressing or obscuring relationships among variables or producing artificial relationships between variables» (King and Brunner 2000 p. 81).

As well as other response biases, it could be approached with specific measures. For example, some tests include a scale (Social Desirability Scale) that allows eliminating participants who tend to give dishonest answers (Paulhus, 1991; Stoeber, 2001). Marlowe-Crowne developed the 33-item Marlowe-Crowne Social Desirability Scale (MCSDS) (King and Brunner 2000; Crown and Marlowe 1960). There are also shorter forms of this scale (Loo and Thorpe 2000; Fischer and Fick 1993; Ballard 1992; Zook and Sipps 1985; Silverstein 1983; Reynolds 1982; Strahan and Gerbasi 1982).

The major problem of using scales for social desirability was discussed by its creator - Crowne. He postulated that SD is on the one hand a personality characteristic, which cannot be eliminated to reveal the true intentions of a subject. On the other hand, eliminating people who have high levels of SD creates representation bias. Therefore, we can conclude that using scales of social desirability might not be the best way to handle the problem.

According to Fisher (1993), indirect questions might be the way to obviate the SD bias. In his research of consumer behavior he compared the results of direct and indirect questions and demonstrated that indirect, projective questions mitigated social desirability in issues that are influenced by social norms and did not affect neutral ones. However, these results were obtained for consumer behavior, which does not have such an impact on personal image, as manipulative traits that generally affect interactions with others. Moreover, this research was conducted before the development of IAT, which was proved to avoid the problems of social desirability.

Due to the fact that SD provides the possibility of creating positive image and preserving positive social identity, we can conclude that people when asked about such a negative characteristic in explicit testing will try to deceive and forge the results. Thus, we assume that the majority of the participants will have high scores of IAT-testing and low scores of MAC-scale. Thereby, we come to the conclusion that the existing measures of manipulative behavior are not predictive, easily forgeable and they do neither achieve required goals in science, nor meet expectations of HR-specialists and clients. Therefore, we believe, that we should create another measurement that will use non-obvious questions that will test the predispositions and propensities of a person to manipulate others. As follows from previous descriptions, we believe, that IAT will serve these purposes.



Chapter 2. Development and validation of Implicit association test for propencsity to be engaged in manipulative communication

The current study is aimed at creating and validating the Implicit Association Test for propensity to be engaged in manipulative communication. In order to do it we set the following tasks:

1. Collect association-rows for the words selected for the IAT

2. Develop 360° questionnaire to test the predictive power of the methods

3. Develop software for IAT

4. Collect final data, using MAC, 360° and IAT

5. Analyze the data with relevant statistic procedures

Study 1: Collection of association rows

Sample

Participants were 156 people aged 18 to 26 years (M =24. 38, SD =1. 37). There were 98 (62,2%) females and 59 (37,8%) males. All participants had Russian as their first language and lived in Russia. Subjects were recruited by convenient sampling.

Procedure

Participants were recruited both online on the LimeSurvey platform and directly in person. They were given the same form to complete (See Appendix A). The admission criterion was to be a native Russian speaker. In online recruitment participants were sent the standard contact message. It had a brief description of the study and the information about ethical principles, as well as a link that directed participants to the online form. When participants were recruited in person, they received the same information on the first page of a questionnaire. The time was not limited.

Results

We have collected around 450 associations per item. After frequency analysis, made via IBM SPSS, the most frequent associations were: Обман [Lie] (n = 15), Лицемерие [Hypocracy] (n = 8), Эгоизм [Selfishness] (n = 7), Выгода [Benefit] (n = 6), Хитрость [Slyness] (n = 5) for “manipulation”; Честность [Honesty] (n = 15), Верность [Faithfullness] (n = 9), Искренность [Sincerity] (n = 8), Честь [Honor)] (n = 5), Уважение [Respect] (n = 5) for “decency”; Спокойно [Calmly] (n = 18), Тепло [Warmth] (n = 11), Прекрасно [Wonderful] (n = 9), Радость [Joy] (n = 7), Приятно [Pleasant] (n = 6)» for “good”; Слезы [Tears] (n = 11), Болезнь [Illness] (n = 8), Предательство [Betrayal] (n = 5), Бедность [Poverty] (n = 6), Глупость [Stupidity] (n = 4)» for “bad”.

Discussion

Stage 1 of the research allowed us to construct the Implicit Association Test, using the most frequent associations, given to the relevant constructs. The associations, gathered for wards “good” and “bad” differ from ones that are used traditionally in implicit tests. As the sample in Stage 1 and the process of validation possessed the same characteristics, it is important to use associations that have predictive outcome. Moreover, there is no collected data on associations to the words “manipulation” and “decency”. Thus, this stage was vital for next stages of this research.

Study 2: Development of 360° questionnaire

Sample

Participants were 32 people who specialize in Psychology, Pedagogics (18 subjects) and HR management (14 subjects). The education level of subjects specialized in Psychology and/or Pedagogics varied from postgraduate students to PhD. The positions of HR specialists included consultants, senior consultants, managers of department and head of the branch. All specialists were more than 2 years of experience in the sphere of personnel selection.

Procedure

First, we have analyzed the theoretical data about common practices of manipulators, creating lists of behavioral patterns that are used by manipulators (1 markers), those that they do not use (-1 markers), that further will be used as reversed items and random characteristics, that have no confirmation in literature (0 markers) to attract accidental points, resulting in 38 behavioral patterns (see Appendix B and C).

Second, experts were given the questionnaire in paper form. The first page of the questionnaire contained our definition of manipulation. Participants were asked to evaluate how often a manipulator uses each pattern on a 7-point Likert scale, according to which «1» corresponds to «never», «2» - «very rarely», «3» - «rarely», «4» - «sometimes», «5» - «often,» «6» - “very often” and «7» is “always”. The time was not limited.

Results

During the expert-review, some items were reconsidered. For example, «Эксплуатирует вашу порядочность» (“Exploits your decency”) was changed to «Пользуется вашим стремлением помогать» (“Uses your desire to help”) in order to avoid strongly negative wording, that could bias the results.

Direct and indirect patterns were analyzed. The analysis of obtained results was made via R program. It showed that Kendall concordance level varied from .64 to .95, p < .05 (see Appendix D). We selected markers with the highest weight, resulting in total 10 markers, including 11 direct statements (2, 4, 5, 7, 9, 10, 11, 13), 2 reversed items (1, 3). The zero markers were not analyzed due to their purpose - to attract biases. They were selected randomly, resulting in 3 zero-statements (6, 8, 12) (see Appendix E).

Discussion

We have created the 360° Assessment from theoretical assumptions that were evaluated by experts. Due to the fact that our sample included both psychologists and specialists in personnel selection, the obtained results represent two perspectives. Gathered results demonstrated high level of concordance level on the majority of items. On the one hand, it might indicate that we have deduces the right behavioral patterns from literature. On the other hand, it may be explained by other factors, such as the presence of professional deformation in experts, the absence of independent information or other biases.

Study 3: Validation of IAT on propensity to be engaged in manipulative communication

Sample

Our sample for the final stage of the study consisted of 98 respondents, 39 (40. 2%) males and 58 (58. 8%) females (see Figure 1), age 18 - 48 (M=24. 7, SD=. 406). All participants were native Russian speakers, due to the requirements of IAT. All of them were from Russia, locating in Moscow at the time of the testing.

The sample consisted of 21 groups of two (N = 42), 13 groups of three (N = 39), 3 groups of four (N = 12) and 1 group of five (N = 5). The admission criterion for groups was the presence of at least one year of constant communication. Participants estimated the closeness of relationships on a scale from 0 to 5, where 1 meant “do not communicate” and 5 was “very close relationships”. Due to the initial instructions, all participants indicated their relationships as at least on the fourth level.

Procedure

Participants were recruited online via standard contact message that included general information about the study, contacts of the researcher, ethical principles of participation. Subjects were recruited together with their friends or partners due to the requirements of 360° Assessment. All participants were instructed to create the common name for their group and agree on a name for each member. After this they were sent the invitation with and two links - first led to the Implicit Association Test, second led to MAC scale and 360° Assessment on LimeSurvey platform. The first link was the same for all groups; the second differs depending on the quantity of respondents in the group. The difference was in the number of 360° questionnaires. The order of instruments-exposure was fixed: IAT, MAC-scale and 360° Assessment in order to avoid disturbance in the results of the IAT. The time of participation was not limited. After a month of initial testing, 17 respondents were tested on IAT one more time. They were recruited online.



Figure 1. The pie-chart of gender distribution in the sample (N = 98)

Measures

1. Our own Implicit Association Test for propensity to be engaged in manipulative communication. It was situated on a personal domain. The elements of coding could be seen in Appendix H.

2. MAC-scale, adapted to Russian by Znakov (see Appendix F). In this scale participants were asked to evaluate 20 statements in 5-point Likert-type scale, where 1 is determined as “strongly disagree”, and 5 - “strongly agree”. This method includes two types of items. Ten items are evaluated on a direct scale (1, 2, 5, 8, 12, 13, 15, 18, 19 and 20), and ten on the reverse scale (3, 4, 6, 7, 9, 10, 11, 14, 16 and 17), respectively. The questionnaire was adapted by Znakov in 2000 and is commonly used to measure the Machiavellianism.

3. Developed 360° Assessment (see Stage 2 of this study)

Validation strategy

There are two general concepts of testing that we need to address in order to validate a new method: validity and reliability.

First of all, we need to control the construct validity:

We will address the results of IAT and MAC-scale. Our test would be valid if its results will not correlate with the results of MAC-scale (see Figure 2).

Figure 2. The hypothyzed correlations of the results of IAT and MAC-scale

Also we would like to see the impact of Social Desirability:

We hypothesize that the majority of respondents will try to show themselves from a positive side and will be in the first fourth of the graph (social desirability bias).

We control criterial validity:

For this purpose we would look for correlations between IAT scores and 360° Assessment scores. Our test will be valid if its results will correlate with the results of the 360° Assessment (see Figure 3).

Figure 3. The hypothyzed correlations of the results of IAT and 360° Assessment

We also control discriminative validity:

In order to evaluate discriminative validity of IAT and MAC-scale we would use regression analysis. The test with higher rІ on the results of 360° Assessment will have better predictive power and will be more informative.

As for retest reliability of our test, we will conduct the second trial of IAT on a portion of our sample after a month of the first data collection. Our test will have high retest reliability if the results of both trials will have high level of correlation.

...