Investigating the Role of Personality Traits and Influence Strategies on the Persuasive Effect of Personalized Recommendations Gkika Sofia Skiada Marianna Lekakos George PhD Student PhD Student Assosiate Professor ELTRUN, E-Business Center ELTRUN, E-Business Center ELTRUN, E-Business Center Athens University of Economics and Athens University of Economics and Athens University of Economics and Business Busines Business Evelpidon 47-A & Lefkados 33, Room Evelpidon 47-A & Lefkados 33, Room Evelpidon 47-A & Lefkados 33, Room 801, GR-11362, Athens Greece 801, GR-11362, Athens Greece 801, GR-11362, Athens Greece gkikas@aueb.gr mskiada@aueb.gr glekakos@aueb.gr Kourouthanasis Panos Assistant Professor ELTRUN, E-Business Center Athens University of Economics and Business Evelpidon 47-A & Lefkados 33, Room 801, GR-11362, Athens Greece pkour@ionio.gr ABSTRACT Recommender systems provide suggestions for products, services, or information that match users’ interests and/or needs. However, 1. PERSUASIVE MESSAGE PROCESSING not all recommendations persuade users to select or use the Persuasive Technologies utilize several techniques in order to recommended item. The Elaboration Likelihood Model (ELM) shape, reinforce or/and change humans’ attitudes and behaviours suggests that individuals with low motivation or ability to process without coercion or deception (Fogg, 2002). On the other hand, the information provided with a recommended item could Recommender Systems represent a class of personalization eventually get persuaded to select/use the item if appropriate technologies that aim to tailor products/information/services peripheral cues enrich the recommendation. The purpose of this according to their users’ interests, preferences and needs. Thus, research is to investigate the persuasive effect of certain influence personalized recommendations can significantly strengthen the strategies and the role of personality in the acceptance of effect of persuasive interventions due to the inherent influence of recommendations. In the present study, a movie Recommender personalized communication. Berkovsky et. al. (2012) suggest System was developed in order to empirically investigate the that most of the extant research examine personalization and aforementioned questions applying certain persuasive strategies in persuasive technologies in isolation although “both personalized the form of textual messages alongside the recommended item. and persuasive technologies aim to influence user interactions or The statistical method of Fuzzy-Set Qualitative Comparative the users themselves”, acknowledging “…the huge untapped Analysis (fsQCA) was used for data analysis and the results potential of personalization to maximize the impact of persuasive revealed that motivating messages do change users’ acceptance of applications” (Berkovsky et. al., 2012). the recommender item but not unconditionally since user’s In information-theoretical terms, persuasion is modeled by the personality differentiates the effect of the persuasive strategies. Elaboration Likelihood Model (Petty and Cacioppo, 1986), which suggests that individuals with low motivation or ability may not Keywords elaborate the information provided (e.g. through a Persuasion, Persuasive Technologies, Personalization, recommendation) and therefore users’ neutral or negative Recommender Systems, Personality, Elaboration Likelihood behavioural response in recommendations (expressed in the form Model. of low rating or non-selection of the recommended item) may not depict their actual intention towards the recommended item. In such cases, the utilization of additional peripheral cues (motivating elements) may increase the persuasive effect of recommendations by engaging users to further elaborate the provided information (Fogg, 2009) in order to investigate the potential to adopt the recommendation. In Recommender systems, explanations are typically used to provide users additional EMPIRE 2016, September 16, 2016, Boston, MA, USA. information that will support them in their decision making Copyright held by the author(s). process and can be eventually utilized as the means to pass users in Recommender Systems since an explanation is a mean through persuasive messages (Tintarev and Masthoff, 2011). which a consumer perceives the value of the recommended item so as to decide whether is close to his/her interests or not. Along the above lines, the first objective of this research is to Explanations can operate like motivators and are being used by investigate the persuasive effect of the influence strategies several systems such as MovieLens (Herlocker et al., 2000) and proposed by Cialdini (1993), namely Reciprocation, Consistency, Social software items (Guy et al., 2009). However, there is no Social Proof, Liking, Authority, Scarcity, which are implemented clear indication in extant literature about what would be the as persuasive messages in the form of recommendations content of explanations (i.e. the message passed to users) that can explanations in a movie recommender system developed for the actually lead to persuasion. For example, a description of how the purposes of this study. recommendation has emerged (i.e. transparency of Moreover, previous studies (e.g. Kaptein and Eckles, 2012) recommendations) has been shown to be associated with an suggest that persuasive messages do not always achieve their goal increase of trust in recommendations (Herlocker et al., 2000) to persuade users. Indeed, if users receive “wrong” messages (i.e. while still there is no enough empirical evidence that irrelevant or annoying) then negative behavioural responses may demonstrates what type of messages could lead to persuasion be generated. In this context, previous studies (e.g. Halko and (Halko and Kientz, 2010). Kientz, 2010) have demonstrated the significance of the A number of persuasive (or influence) strategies have been individual’s personality in the (negative or positive) behavioural proposed in the literature and can be eventually be utilized in the responses to persuasive messages. Following the above design of persuasive messages. For example, Fogg (2002) argumentation, the second objective of this study is to examine describes 42 persuasion strategies and Cialdini (2001) 6 influence the role of personality in the acceptance of the recommendations strategies (also known as Six Weapons of Influence) In this study, and identify possible differentiations in the users’ response on the we rely upon Cialdini’s influence strategies since they have been persuasive strategies that may attributed on their personality type. broadly used and verified there are evidences that if influence In this study, we focus on peripheral cues such as short persuasive strategies are implemented in a system then they increase its messages, developed upon Cialdini’s (2001) six influence persuasive effect (e.g. Fogg, 2002). According to Cialdini (2001). strategies, presented to user as recommendation explanations. We Cialdini’s (2001) influence strategies are the following: consider such messages as peripheral cues because they neither  Reciprocity: humans have the tendency to return favors, affect the quality of argumentation (i.e. how close to the users interests the recommended items are) nor change the  Commitment or consistency: people’s tendency to be recommended item but when users lack of motivation or ability, consistent with their first opinion, these peripheral variables influence users by triggering internal heuristic processing rules (Tam and Ho, 2005), which eventually  Social proof: people tend to do what others do, would lead to persuasion  Scarcity: people are inclined to consider more valuable The rest of the paper is organized in five sections. In Section 2 the whatever is scarce, hypothesis development. Our experiment is presented in Section  Liking: people are influenced more by persons they like and 3, while in Section 4 the experimental results are discussed. Discussion of the study’s findings and a discussion of areas for  Authority: people have a sense of duty or obligation to further research conclude the paper. people who are in positions of authority. 2. HYPOTHESIS DEVELOPMENT Cialdini (1993) suggested that when a compliance professional (e.g. salesperson) uses the above six influence strategies 2.1 Influence strategies as messages in (Reciprocity, Commitment, Social proof, Scarcity, Liking and recommendation explanations Authority) in his/her strategy then (s)he managed to influence The mainstream of research in Recommender Systems has more successfully the customer to consume a traditionally focused on designing and developing accurate product/service/information. In the same vein, Kaptein et al. recommendation algorithms (e.g. Xiao and Benbasat, 2007). More (2012) suggests that applying the influence strategies on text specifically, extant research indicates that the factor that mostly messages people get persuaded to reduce snacking consumption. determines the success of a Recommender System is the provision We adopted Cialdini’s influence strategies because they have of recommendations that are more close to consumer’s already been tested and validated in other domains such as in e- preferences. According to the ELM perspective, the accuracy of commerce (Kaptein, 2011), use of credit cards (Shu and Cheng, recommendation algorithms determines the quality of 2012). They also provide a solid framework in order to investigate argumentation. In other words, if the recommended item is close the persuasive power of messages as peripheral cues in to the user preferences, this will eventually lead to persuasion recommender systems. The above leads to following hypothesis through the central route, i.e. through in-depth processing of the of our study: recommendation. ELM suggests that the alternative (peripheral) H1: Influence strategies (applied as peripheral cues through path may also lead to persuasion if appropriate cues are provided. messages in recommendations explanations) will have a positive Such peripheral cues may be implemented as motivating messages persuasive effect on individuals’ disposition towards the in the form of recommendation explanation (Herlocker, 2000). recommended item. A recommendation explanation can be considered as any type of The examination of the above hypothesis will allow us to additional information accompanying a system’s output, having as demonstrate (if validated) that when the preference matching level ultimate goal to persuade users to try or purchase the item that is of the recommended item is low (i.e. when the recommended item recommended (Tintarev and Masthoff, 2011). Tintarev and is not close to the user’s preferences and interests), then Masthoff (2012) indicate that explanations have an important role enhancing the recommendation by applying influence strategies in the form of short explanatory messages, the user will be Additionally, Halko and Kientz (2010) combined persuasive persuaded to use the recommended item, thus changing his/her strategies with user’s personality using Big Five Dimensions of original negative behavior towards the recommended item to Personality and the results of their study revealed relationships positive intention to use item. between individuals’ personalities and persuasive technologies which means that not all people are affected from the same Influence strategies rely upon different psychological principles persuasive means. Finally, Smith et al. (2016) examined the that may lead to persuasion and therefore it is expected that they impact of patients personality on Cialdini’s influence strategies in will present different degrees of persuasive effect on the recipients the form of reminders. The research indicated that patient’s with of the respective persuasive messages. Thus, the second high emotional stability seem to be more responsive to all hypothesis of our research is: strategies of persuasion, while patients with low agreeableness H2: Influence strategies lead to different degrees of persuasive rated all Cialdini’s strategies higher than those with high. Finally, effect on individuals’ disposition towards the recommended item. the research demonstrated that the reminders of “Authority” and “Liking” are the most popular. 2.2 Personality Kaptein and Eckles (2012) in their study demonstrated that 3. EXPERIMENTAL DESIGN AND influence strategies do not always lead to persuasion. They PROCEDURE indicate that in case a consumer receives a message with ‘wrong’ principle then this can bring undesired effects. The above suggests 3.1 Design of Persuasive Explanation that there are also other factors that should be taken into For the execution of the experiment we had first to design the consideration when a persuasive message is used, one of which is persuasive explanations that would accompany each individual’s personality. A human’s personality is defined as ‘a recommended movie. For this task we followed the methodology dynamic organisation, inside the person, of psychophysical proposed by Kaptein et al. (2012). More specifically, a group of systems that create the persons’ characteristic patterns of three researchers familiar with Persuasive Technology, created behaviour, thoughts and feelings’ (Allport, 1961, p. 11). thirty (30) textual explanations, i.e. five (5) for each Cialdini’s influence strategies. The content of each explanation was Given that, one of the major aims of a Recommender System is to developed in order to comply with the main purpose of each help consumers in decision making processes, the fact that principle in the movie domain. For instance, for the influence personality influences how people make their decisions (Nunes et strategy of Social Proof, the five possible persuasive explanations al., 2012), consumer’s personality should be taken into that were constructed are: (1) The 85% of this research’s users consideration when a persuasive message is provided with a rated the recommended movie with four (4) or five (5) stars. (2) recommendation. Indeed, previous studies suggest a relationship The recommended movie is on ‘to watch’ list of 85% of this between human’s preferences and tastes with their personality in research’s users. (3) Most of the users with the same age and sex different domains such as movies (e.g. Chausson, 2010), music as yours, rated the recommended movie with 4 stars! (4) The and paintings (Rawlings et al., 2000). recommended movie’s video trailer on youtube has more than There is a variety of personality taxonomies one of which is Big 5 550,000 views. (5) The recommended movie’s video trailer on Dimensions of Personality (John et al., 2008). The personality youtube has more than 1600 likes and only 200 dislikes. traits suggested by the Big Five taxonomy are: Extraversion, Seventeen (17) experts in the field of Information Systems and Agreeableness, Conscientiousness, Neuroticism and Openness. Marketing were invited in order to evaluate each explanation in According to psychological research (Jang et al., 2012) the facets terms of its compliance with the respective influence strategy. for each personality trait are: First, a brief presentation of the strategies was given to the  Extraversion: Gregariousness, Assertiveness, Activity, evaluators so as to be more familiar with the influence strategies Excitement-Seeking, Positive Emotions, Warmth. and then they were asked to evaluate the set of persuasive explanations. Each evaluator declared the compliance of each  Agreeableness: Trust, Straightforwardness, Altruism, explanation to the respected influence strategy through a 1 to 5 Compliance, Modesty, Tender-Mindedness. rating scale (from “Completely Disagree” to “Completely  Conscientiousness: Competence, Order, Dutifulness, Agree”). The persuasive explanation with the highest average was Achievement Striving, Self-Discipline, Deliberation. considered as the best-matching explanation for this particular influence strategy.  Neuroticism: Anxiety, Angry Hostility, Depression, Self- The six (6) best-matching persuasive explanations (one for each Consciousness, Impulsiveness, Vulnerability. strategy), were chosen for the experiment are the following:  Openness: Ideas, Fantasy, Aesthetics, Actions, Feelings, Reciprocity: A Facebook friend, who saw the movie that you Values. suggested him/her in past, recommends you this movie. The first study that examined message-person congruence effects Scarcity: The recommended movie will be available to view from with a comprehensive model of personality traits is that of Hirsh 15/1/2014 to 31/1/2014 on cinemas. et al. (2012). Since then message-person congruence effects have been examined in relation to a variety of psychological Authority: The recommended movie won 3 Oscars! characteristics (Dijkstra, 2008). Hirsh et al. (2012) demonstrated Social Proof: The 87% of users in this survey rated the that persuasive messages are more effective when they are recommended movie with 4 or 5 stars! custom-tailored to their interests and concerns. Moreover, Tintarev et al. (2013) demonstrated that people who are Liking: Your Facebook friends like this movie. characterized from Open to Experience (one of the Big 5 personality traits) tend to prefer diverse recommendations. Commitment: This movie belongs in the kind of movies you enjoy movie is considered in our study as a measure of persuasion (i.e. to watch. acceptance of the recommendation), which is operationalized by computing the difference between the original and the final 3.2 Experiment design and execution ratings. However, the exact meaning of the “acceptance the A within subjects experimental design was followed in this recommendation” depends on the business objectives of a site. For research. One of our main concerns in the execution of the example, in some cases (as in e-commerce) the desired behaviour experiment was to manage participants’ burden by avoiding may be to request more information, or to purchase the product extensive exposure to treatments and questionnaires (only the and so on. psychographic questionnaire consisted of 44 items) while preserving the validity of the experiment. One option to deal with problem was to expose different groups to different cues (i.e. follow a between subjects design). However, this would significantly reduce the sample size within each group and also taking into account the anticipated low number different personality types represented in each of the groups it would have limited our ability to produce valid statistical results. Thus, we selected the within subjects design. At the first step of the experiment, a set of 20 movies where presented to participants (with no explanations besides the typical information provided by iMDB, such short description of the story, lead actors etc.), where they were asked to state (by Figure 1. Second Step of the experiment. checking the appropriate option) whether they have watched each movie and then provide their ratings (in 1-5 scale). Users were As mentioned above, the recommended movie was enriched with explicitly instructed to provide their intention to watch a movie persuasive explanations, based on Cialdini’s Principles (i.e. the (for all unwatched movies) in the form of a rating. For the movies explanations designed in the first part of the experiment) and the they had already watched they provided their actual evaluation. participant was asked to assess the recommended movie in order Recommendations were drawn from the set of unwatched movies. to examine whether (and which) strategies influenced users in order to change their intention to watch the recommended movie. The set of 20 movies was randomly selected from a pool of 60 More specifically, the recommended movie was presented with movies from different genres and presented to the participants the same set of information as the first step (title, actors, etc.) along with the typical information for each movie (movie’s genre, while participants were asked to declare their intention to watch its plot, and the starring actors). The first criterion for the the recommended movie, taking into consideration one of the 6 inclusion of a movie in the pool of 60 movies was its genre persuasive explanations each time, which were presented as a list (action, drama, romance, etc.). In the pool of 60 movies there below the recommended movie. The order of the persuasive were at least three movies from each genre, although most of the explanations was appeared in a random way to each user but there movies belong to more than one genre. The second criterion was were the same texts for all of them. For that reason the the popularity of the movie, With the term popular movie is meant expressions that were used in the persuasive explanations were in a movie with high average rating (above 8.0) from a large amount a generic form, e.g. the wording ‘the recommended movie’ was of users (above 1000 users).. Since popular movies are more used instead of the actual title of the recommended movie and so likely to collect higher ratings while unpopular ones may not be on. known to the experimental participants (and therefore attract lower ratings), we included in the sample both popular and The absolute difference between the original and the final rating unpopular movies according to their iMDB ratings. Although that was used to measure the persuasive effect. As the “final” rating the number of 20 movies was large enough to ensure that at least with respect to the first hypothesis (examining if there are some of them wouldn’t have been watched by the participant, the differences before and after the application of the persuasion system was designed to select from the pool of 60 movies and strategies) we used the highest rating that users provided present to participants alternative movies in the extreme case that (independently of the strategy that corresponds to that rating). For all 20 movies have been actually watched by the user. the evaluation of the second hypotheses (examining if there are differences among strategies with respect to their persuasive At the second (recommendation) step of the experiment (see effect), the rating given by the users’ as evaluation of each Figure 1), the (unwatched) movie for which the participant has strategy was considered as the “final” rating. expressed the lowest intention to watch (note that if more than one movie was rated with the lowest score, then the recommended At the third and last step of the experiment participants were movie was selected randomly from the above set of low-rated asked to complete the psychographic questionnaire that was used movies) was presented to the user exactly as the original to classify users into the Big 5 personality traits. The Big Five presentation but enhanced with persuasive explanations. Selecting Inventory- 44 (BFI) was used, constitutes from 44 questions (John to present users with the lowest rated movie, is in alignment with et al., 2008), and is already used in other studies (Bouchard and our theoretical ELM foundations, which suggest that when the McGue, 2003; Shiota et al., 2006). preference matching of the user with respect to the recommended item is low then the peripheral route will be followed. Moreover, 3.3 Sample this choice enable us to track more easily any changes in the The experiment participants were invited through posts in user’s intention to watch the movie since in computational terms it University’s Facebook groups (e.g. undergraduate, postgraduate is much easier to identify changes in intentions from the lower to and PhD students) and authors’ personal mailing lists to the higher levels of the 1-5 scale. It must be noted that the rating participate in this research. The invitation message was asking expresses the users’ intention to watch (or not) the recommended recipients to participate in a research in which they would be asked to rate recommendations provided by an online application a condition implies that an outcome may not derive without the as well as to fill in a psychographic questionnaire. The link to presence of the condition; nevertheless, the condition alone is not access the system was provided and a clear suggestion concerning able to produce the outcome. Sufficiency of a condition implies the anonymity of their participation was included in the message. that the condition alone is capable of producing the outcome. In The invitation did not specify that the research involved movies practice, if a solution includes the presence of only one condition evaluation. The participants’ average scores for the items (i.e., a solution requires the presence or absence of only one measuring the personality types in the 44-item psychographic personality trait),, then this condition is sufficient to produce the questionnaire are (the standard deviation is included in the outcome. To estimate the sufficiency and/ or necessity of parentheses) Extraversion: 3.34 (0.49), Agreeableness: 3.47 hypothesized conditions, fsQCA follows a Boolean minimization (0.42), Conscientiousness: 3.34 (0.42), Neuroticism: 3.30 (0.48), process based on truth table analysis. The outcome of this process Openness: 3.24 (0.46). The above descriptives showcase that the includes the generic combinations of conditions that are sufficient sample does not exhibit certain personality types more (or less) for the outcome whilst remaining logically true. These are than others. encapsulated in three solutions that differ based on their complexity, named as complex, intermediate, and parsimonious. In total 117 users participated in our research. 61 (52%) Of interest is the parsimonious solution, which reduces the causal participants of our sample were males while the rest 56 (48%) recipes to the smallest number of conditions possible. were females. Additionally, the 46% of the sample was aged between 18 and 24 years old, the 52% was between 25 and 34 This research explores how individuals’ personality traits, in the years old and the 2% at the age of 35-44 years old. form of five alternative dimensions, fit with different persuasive strategies. Nevertheless, an individual may not be exclusively 3.4 Analysis Methodology categorized under a unique personality trait. Instead, individuals This research employs the prescriptions of the fuzzy-set may exhibit elements of multiple traits, which collectively form qualitative comparative analysis methodology (fsQCA) to explore their personality. Moreover, these personality traits are not fixed which personality traits explain the effectiveness of each within all individuals; a particular persuasive strategy may be persuasion strategy. Opposed to variance-based statistical perceived as equally appropriate to individuals that exhibit methods (e.g. structural equation modelling or partial-least completely dissimilar values on their fundamental personality squares based regression models) in which the independent qualities. As a result, we cannot assume that there is a single, variables ‘compete’ with each other to explain one or more universal, personality profile that explains the impact of a given dependent variables, fsQCA treats the hypothesized causal factors persuasion strategy, which would call for the application of as conditions that may be related to the phenomenon under traditional statistical analysis methods based on regression investigation either by themselves or in combination with one models, but we need to examine how the different combination of another (Rihoux and Ragin,, 2009; Rihoux et al., 2011). Hence, the personality traits interweave in order to explain the suitability fsQCA does not compute a single, optimal, solution that attributes of a given persuasion strategy. The modus operandi of fsQCA weights to the independent variables; instead, the methodology covers this requirement, thus warrants us to adopt it as our guiding proposes multiple alternative solutions, which require the analysis methodology. presence or absence of each hypothesized causal factor. This is a fundamental difference from variance-based statistical methods 4. RESULTS and calls for operationalization of the variables in the dataset. The first step of our analysis is involved investigating effect of In effect, fsQCA employs fuzzy set theory and Boolean algebra to each influence strategy on individuals’ attitude towards watching evaluate whether the cases in the dataset belong or not in a certain a movie that they, initially, were unmotivated to watch. We conceptual state. For example, in this research cases may be performed two different comparisons to examine the persuasive evaluated in order to assess whether an individual is extravert, effect of the influence strategies. In the first test, we measured the open, agreeable, conscious, or neurotic. Likewise, the impact of difference between the maximum of the ratings that each user each persuasion strategy on individuals’ attitude change may also provided for the six influence strategies and the original rating. be operationalized to capture the degree to which the strategy The t-test results suggested that on average there are significant actually manifested a behavior change. Such operationalizations differences (p< .001) between the original rating and most are captured through fuzzy set membership scores ranging from 0 persuasive (for each user) strategy (original and final ratings (non-membership to the set) to 1 (full membership to the set). In- average scores: 1.49 and 3.05 respectively with standard deviation between scores indicate the distance of each case from the 0.50 and 1.23). In the second statistical test, we performed a t-test outbound scores. The researcher may transform the cases’ original analysis that compares their initial beliefs and the ones formulated values to fuzzy-set membership scores by using specialized after the application of the strategy. The results suggest that all fsQCA software. This process is coined with the term influence strategies were successful in increasing the likelihood of ‘calibration’. In this research we used fsQCA 2.0 developed by individuals to watch the movie (Table 1) nevertheless, this the University of Arizona. The software was also employed increase is marginal in absolute figures.. throughout the remaining methodology stages. Fuzzy-set QCA identifies conditions or combinations of Table 1: T-test results. All comparisons are significant at conditions that are necessary or sufficient to explain an outcome. p<.001 In this research, a combination of conditions reflects the Influence Mean (SD) T-statistic (Original rating personality profile of an individual. Such profile would include Strategy – intention after influence specific membership values to each personality trait following the strategy is applied) calibration procedure. As such, a value close to 1 in a particular personality trait implies that the individual exhibits this trait. In Original Rating 1.49(0.50) n.a contrast, membership values close to zero imply that the Reciprocity 1.84(0.89) -4.707 (p<.001) individual does not exhibit the said personality trait. Necessity of Scarcity 1.73(0.97) -2.953 (p<.001) Table 3 illustrates the results of fsQCA for the Reciprocity influence strategy. The methodology, identified four solutions Authority 2.57(1.16) -10.941 (p<.001) leading to high influence of an individual by the application of the Social Proof 2.67(1.17) -12.349 (p<.001) respective strategy. The results indicate that the absence of even one personality trait is sufficient to individuals in order to be Liking 2.07(1.04) -6.698 (p<.001) influenced by the Reciprocity strategy Moreover, a one-way ANOVA test between the attitude changes Table 3: fsQCA results for the paths leading to high of individuals for each influence strategy (see Table 2). The acceptance of Reciprocity. results of this analysis indicate that there are statistical differences among the six strategies at the p<.05 level (F= 14.941, p= .000). Solutions leading to high acceptance of To probe for differences between the strategies we performed a Reciprocity influence strategy Games-Howell Post Hoc Test. Based on these results we accept Personality Traits 1 2 3 4 H1. Extraversion  Table 2: ANOVA results (Sign. < 0.05) Agreeableness  Persua Recipro Autho Scarc Soci Liki Consist Conscientiousness  sive city rity ity al ng ency Openness  Strateg Pro y of Neuroticism Sign. .001 .001 .006 .001 .003 .007 Consistency 0.672 0.636 0.644 0.70 Coverage 0.578 0.624 0.572 0.639 H2 was evaluated through the application of fsQCA methodology. We used the five personality traits as possible conditions that Overall solution 0.611 influence the acceptance of each influence strategy. As a first step, consistency the prescriptions of fsQCA require for calibration of the cases into Overall solution 0.970 membership sets. Calibration was performed using the coverage corresponding function provided by fsQCA 2.0 software. The function demands as input three threshold points; a full- membership value, a non-membership value and a cutoff point. The methodology identified 6 alternative paths leading to high Because the dataset consists of subjective cases, we used cluster acceptance of the Scarcity influence strategy. The majority of analysis following the k-means algorithm (k=3) to calculate the paths require two personality traits to be present in an individual’s three membership sets. More specifically, high values are personality in order to be influenced by Scarcity strategy (Table correlated with the full-membership set, medium values are 4). For example, individuals that are both agreeable and correlated with the crossover point set and finally low values are conscious, but do not exhibit traits of neuroticism are likely to be correlates with the non-membership set. influenced by the scarcity influence strategy (solution 6). For the independent variables (personality traits) no cluster Table 4: fsQCA results for the paths leading to high analysis was conducted due to the fact that the differences among acceptance of Scarcity. the personality traits’ scores were imperceptibly small. Thus, for Solutions leading to high acceptance of this case we calculated the independent variables (personality Scarcity influence strategy traits) through frequencies with cut points for 4 equal groups, in SPSS. The percentiles that emerged correspond to the full- Personality 1 2 3 4 5 6 membership set for the high values, the crossover point set for Traits medium values and finally the non-membership set for low Extraversion  values. Agreeableness     The results of fsQCA indicate 3-7 alternative solutions per Conscientiousness    influence strategy comprising of alternative combinations of the personality traits that lead to high acceptance of each influence Openness     strategy. Black circles indicate the required presence of a Neuroticism   personality trait in a solution. White circles indicate the required absence of a personality trait from the solution. Blank cells indicate that in that particular solution, the presence or absence of Consistency 0.797 0.7 0.7 0.87 0.7 0.873 that personality trait is indifferent. Each solution is accompanied Coverage 0.295 0.416 0.358 0.193 0.376 0.206 by two additional measurements of fitness, which express the Overall solution 0.685 ‘predictive power’ of each solution, namely the consistency and consistency coverage indexes. Consistency presents how consistent is the Overall solution empirical evidence with the outcome which is investigated while 0.747 coverage coverage estimates the proportion of cases that address the outcome which is under investigation. The remaining Tables present the different paths, consisting of Personality Traits 1 2 3 combinations of personality traits, which lead to high acceptance Extraversion  of the remaining four influence strategies. These tables may be interpreted as an atypical personality profile of individuals (one Agreeableness   per produced fsQCA solution) in order to be influenced by each Conscientiousness   strategy (Table 5 – Table 8). Similar to the previous solutions, Openness  each table should be interpreted as a combination of mandatory personality traits (indicated with black circles) coupled with the Neuroticism  mandatory absence of one or more personality traits (indicated with white circles). Hence, each solution represents a unique combination of the personality traits that should exist in order to Consistency 0.47 0.48 0.64 explain the acceptance of a persuasive strategy. Coverage 0.41 0.31 0.192 Table 5: fsQCA results for the paths leading to high Overall solution 0. 456 acceptance of Authority. consistency Overall solution 0.643 Solutions leading to high acceptance of coverage Authority influence strategy Personality 1 2 3 4 5 6 Table 8: fsQCA results for the paths leading to high Traits acceptance of Consistency. Extraversion  Solutions leading to high acceptance of Agreeableness     Liking influence strategy Conscientiousness    Personality Traits 1 2 3 Openness     Extraversion  Neuroticism    Agreeableness   Conscientiousness   Consistency 0.598 0.604 0.62 0.674 0.677 0.636 Openness  Coverage 0.294 0.303 0.357 0.182 0.252 0.25 Overall solution Neuroticism  0.566 consistency Overall solution 0.752 coverage Consistency 0.47 0.48 0.64 Coverage 0.41 0.31 0.192 Table 6: fsQCA results for the paths leading to high Overall solution 0. 456 acceptance of Social Proof. consistency Solutions leading to high acceptance of Overall solution 0.643 Social Proof influence strategy coverage Personality 1 2 3 4 5 6 7 5. DISCUSSION Traits This research emphasizes on two elements of persuasive/ Extraversion      recommender systems. First, we empirically validate that the Agreeableness   application of an influence strategy may indeed positively shift Conscientiousness    the attitude of an individual towards a specific recommended item. Nevertheless, not all influence strategies have the same Openness     persuasive effect. We attribute this deviation to the personality Neuroticism        traits of the recommender system users. Hence, the second contribution of this study reflects on the development of personality profiles per influence strategy. Each profile, measured Consistency 0.698 0.645 0.619 0.604 0.581 0.698 0.637 as a combination of personality traits that need to be present or Coverage 0.31 0.25 0.303 0.317 0.250 0.31 0.190 absent from the personality mix, reflects the set of traits that fit most with each influence strategy (i.e., individuals sharing the Overall solution 0. 713 consistency same profile would indeed be persuaded following the application of the respective strategy). It must be noted that an important issue Overall solution 0. 577 in utilizing recommendation explanations is that persuasive coverage messages may be perceived as promotional ones and therefore impact users’ trust in the recommender systems. For this reason Table 7: fsQCA results for the paths leading to high we used a control variable measuring (in an 1-5 scale) users’ trust acceptance of Liking. in the system, which has shown that no such effect occurred (i.e. Solutions leading to high acceptance of no significant differences were found between the trust levels Liking influence strategy before and after the presentation of the persuasive messages, which was on average 2.96 for the users with low intention to watch the movie and 3.27 for the users with high intention to “Reciprocity” is presented to them. Despite our expectations, watch a movie). humans with low conscientiousness changed their intention to watch the movie influenced by the “Consistency” strategy rather In effect, most studies in the field of recommender systems have than humans with high conscientiousness. This may be attributed primarily focused on the algorithmic perspective through the to the fact that individuals with high conscientiousness avoid to proposition of algorithms that provide recommendations tailored take risks because that might make them feel uncertain or cause to users’ interests and preferences. In contrast, this study provides unexpected delays to their work (James and Mazerolle, 2002; Raja insights indicating that the provision of properly selected (i.e. and Johns, 2004). taking into account users’ personality) motivating messages have a persuasive effect on users intention to “use” the recommended On the other hand, people with high openness tend to be item, e.g. to watch a movie. characterized by creativity, sophistication, and curiosity (Barrick and Mount, 1991). This might explain why in most cases, the trait According to the Elaboration Likelihood Model (ELM), when an of openness is absent from the solutions indicated by fsQCA. individual has low motivation (or ability) to process a Finally, individuals with low neuroticism lack confidence. This recommendation then she will not proceed through the central may explain why the application of the “Social Proof” strategy on route of persuasion, i.e. he will not thoroughly assess the quality neurotics in most of cases depicts low neuroticism and Liking, of argumentation in order to get persuaded. Instead, if appropriate because they tend to be influenced by people who they like or peripheral cues are implemented (such as persuasive strategies what the majority says. Additionally, neurotics are characterized applied in the form of messages, as suggested in our study) then by anxiety and typically they do not trust others (Raja and Johns, she will eventually be influenced (i.e. motivated) to elaborate the 2004), so they tend to be consistent with their original thoughts in recommendation following the peripheral route to persuasion. order to deal with their insecurity and therefore it is expected to Such peripheral cues act as extra motivating triggers that get persuaded by the “Consistency” strategy. influence a user by “diverting attention, reallocating cognitive resources, and evoking affective responses and behaviours” (Tam The findings of the study must be interpreted taking into account and Ho, 2005). its limitations. The sampling frame (students) and the relatively low sample size restrict the possibility of having an actual Current recommendation applications typically disregard items representation of the population in the sample in terms of with low degrees of fitness with the users’ current interests. The personality types. By extending the experiment, in future research, confirmation of the first hypothesis of this study indicate that even to a larger sample of users we would also have the opportunity to for such items, there is strong possibility that they may be avoid possible self-selection bias as well as to follow a between favoured by the users if they are presented with the appropriate subjects design, showing not only more movies to each user but motivating peripheral cue. Moreover, not all people are influenced most importantly avoiding the learning effect associated with the from the same persuasive messages. This study provides empirical presentation of all six strategies to all experiment participants. It evidence that there is a relationship between personality and must be noted that we tried to control the learning effect bias by Persuasive Strategies. People with specific combination of showing to users recommendations with persuasive explanations personality traits are affected more from particular persuasive in a random way, i.e. the mix of recommendations representing messages. different persuasive strategies was presented in varying order to The results of the experiment that was conducted surfaced that each of the participants. It is clear that this study provides insights motivating messages are not uniformly applied to all recipients of concerning the movie recommendation domain in which it was recommendations. Users’ personality traits are an important applied. The generalization of our findings would be enabled only factor that differentiates the effect of influence strategies applied if this research is extended to other application domains. In our as persuasive explanations. More specifically, a person who is future research plans, besides the extension of our research to characterized by high extraversion seems to be influenced by all other domains (e.g. e-commerce) we aim to investigate additional Six Persuasive Strategies. This is reasonable if we take into factors that may influence persuasive communication, as for consideration that they enjoy interacting with the environment example the need for cognition, which is a personality variable whilst such people have the tendency to seek for stimulation and reflects people’s intrinsic motivation to engage in and enjoy (Zhao and Siebert, 2006). Moreover, people with high thinking (Cacioppo and Petty, 1982, p. 116). extraversion have the tendency to be curious, novel, sociable, active, energetic (Costa and McCrae, 1992; Goldberg, 1992), and 6. ACKNOWLEDGMENTS positive (Watson and Clark, 1997). Along this line, the fact that The first author acknowledges the financial support of the this type of people favour networking with others (Watson and Department of Management Science and Technology and the third Clark, 1997) make them more prudent to be influences by author the financial support of the Research Center of the Athens “Liking” strategies. University of Economics and Business for the presentation of this work. Individuals with high agreeableness are eager to help other people (Costa and McCrae, 1992) while they have the tendency to be 7. REFERENCES kind, generous, fair and unconditional (Goldberg, 1992), so [1] Allport, G. W.: Pattern and growth in personality. New York: people with high agreeableness tend to be motivated from the Holt, Rinehart & Winston (1961). “Reciprocity” influence strategy. The fact that people with low [2] Barrick, M. R., Mount, M. K.: The Big Five personality agreeableness tend to be suspicious (Digman, 1990). dimensions and job performance: A meta-analysis. Personnel People with high conscientiousness are dutiful (Major et al., Psychology, 44, 1-26 (1991). 2006). Ιn other words, they are careful to fulfil obligations, and [3] Berkovsky, S., Freyne, J., Oinas-Kukkonen, H. Influencing thus when someone helps them they feel obligated so they become Individually: Fusing Personalization and Persuasion. ACM more motivated when a persuasive explanation implementing Transactions on Interactive Intelligent Systems, 2(2) (2012). [4] Cialdini, R. B.: Influence: Science and practice (3rd ed.). development activity. Journal of Applied Psychology, 91, New York: HarperCollins (1993). 927-935. doi:10.1037/0021-9010.91.4.927 (2006). [5] Cialdini RB. Influence, Science and Practice, Allyn & Bacon, [21] Nunes, M. A. S. N., Hu, R.: Personality-based Recommender Boston (2001). Systems: An Overview. In: Proc. of the 6th ACM [6] Chausson, O.: Who Watches What? Assessing the Impact of Conference on Recommender Systems (RecSys’12), pp. 5-6 Gender and Personality on Film Preferences. myPersonality (2012). project, Univeristy of Cambridge (2010). [22] Petty, R.,Cacioppo, J.: The elaboration likelihood model of [7] Costa, P. T., Jr., McCrae, R. R.: Revised NEO Personality persuasion. Advances in experimental social psychology, Inventory (NEO PI-R) and NEO Five-Factor Inventory (1986). (NEO-FFI) professional manual. Odessa, FL: Psychological [23] Raja, U., Johns, G.: The impact of personality on Assessment Resources (1992). psychological contracts. Academy of Management Journal, [8] Digman, J. M.: Personality structure: Emergence of the five- 47, 350-367 (2004). factor model. Annual Review of Psychology, 41, 417-440 [24] Rawlings, D., Barrantes, N., Furnham, A.,: Personality and (1990). aesthetic preference in Spain and England: two studies [9] Fogg, B. J.: Persuasive Technology: Using Computers to relating sensation seeking and openness to experience to Change What We Think and Do. Morgan Kaufmann (2002). liking for paintings and musi. European Journal of Personality, Volume 14, Number 6, November/December [10] Fogg, B. J.: A behavior model for persuasive design. 2000, pp. 553-576(24) (2010). Proceeding of Persuasive '09 Proceedings of the 4th International Conference on Persuasive Technology (2009). [25] Rihoux, B., and Ragin, C.C. 2008. Configurational Comparative Methods: Qualitative Comparative Analysis [11] Goldberg, L. R.: The development of markers for the Big- (Qca) and Related Techniques. SAGE Publications, Five factor structure. Personality Assessment, 4, 26-42 Incorporated. (1992). [26] Rihoux, B., I. Rezsohazy & D. Bol (2011): Qualitative [12] Halko, S., Kientz, J.A.: Personality and persuasive Comparative Analysis (QCA) in Public Policy Analysis: an technology: An exploratory study on health-promoting Extensive Review. German Policy Studies, 7(3), 9–82. mobile applications. Persuasive technology, pp. 150–161 (2010). [27] Smith, K.A., Dennis, M., Masthoff, J. Personalizing Reminders to Personality for Melanoma Self-checking. [13] Herlocker, J.L., Konstan, J.A., Riedl, J.: Explaining Conference on User Modeling Adaptation and collaborative filtering recommendations. CSCW 2000: 241- Personalization, p. 85-93. (2016) 250 (2000). [28] Tam, K.Y., Ho, S.Y.: Web personalization as a persuasion [14] Hirsh, J. B., Kang, S. K., Bodenhausen, G. V.: Personalized strategy: An elaboration likelihood model perspective. Persuasion Tailoring Persuasive Appeals to Recipients’ Information Systems Research, pp. 271 - 291 (2005). Personality Traits.Psychological science, 23(2), 578–581 (2012). [29] Tintarev, N., Masthoff J.: Designing and evaluating explanations for recommender systems. Recommender [15] Jang, K.L.,Livesley,W.J., Vemon, P.A.:Heritability of the Systems Handbook, Springer, p. 479–510 (2011). Big Five Personality Dimensions and Their Facets: A Twin Study. Journal of personality, 64(3), pp. 577–592 (1996). [30] Tintarev, N., Masthoff J.: Evaluating the effectiveness of explanations. User Model User-Adap Inter. pp.399–439 [16] James, L. R., Mazerolle, M. D.: Personality in work (2012). organizations. Thousand Oaks, CA: Sage (2002). [31] Tintarev, N., Dennis, M., Masthoff J Adapting [17] John, O. P., Naumann, L. P., Soto, C. J.: Paradigm shift to Recommendation Diversity to Openness to Experience: A the integrative big-five trait taxonomy: History, Study of Human Behaviour. User Modelng, Adaptation and measurement, and conceptual issues. In O. P. John, R. W. Personalization, vol. 7899, pp.190–202 (2013). Robins, & L. A. Pervin (Eds.), Handbook of personality: Theory and research (3rd ed., pp. 287–310). NewYork: [32] Watson, D., Clark, L. A.: Extraversion and its positive Guilford Press (2008). emotional core. In R. Hogan, J. A. Johnson, & S. R. Briggs (Eds.), Handbook of personality psychology (pp. 767–793). [18] Kaptein, M., Eckles, D.: Heterogeneity in the effects of San Diego, CA: Academic Press (1997). online persuasion.Journal of Interactive Marketing, 26(3), pp.176–188 (2012). [33] Xiao, B., Benbasat, I.: E-commerce product recommendation agents: use, characteristics, and impact. Mis Quarterly, 31(1), [19] Kaptein, M., De Ruyter, B., Markopoulos, P., Aarts, E.: pp. 137–209 (2007). Adaptive persuasive systems: a study of tailored persuasive text messages to reduce snacking. ACM Transactions on [34] Zhao, H., Seibert, S. E.: The Big Five personality dimensions Interactive Intelligent Systems (TiiS), 2(2) (2012). and entrepreneurial status: A meta-analytical review. Journal of Applied Psychology, 91, pp. 259-271(2006). [20] Major, D. A., Turner, J. E., Fletcher, T. D.: Linking proactive personality and the big five to motivation to learn and