A NEW WAY OF LOOKING AT THE BARNUM EFFECT AND ITS LINKS TO PERSONALITY TRAITS IN GROUPS RECEIVING DIFFERENT TYPES OF PERSONALITY FEEDBACK

The present study proposes a new way of looking at the Barnum effect. A theoretical model describing the relationship between personality traits, type of personality feedback, and acceptance of personality feedback is proposed and tested. The results of the study provide proof for the model, supporting the notion that personality feedback is accepted both rationally and irrationally, demonstrating that the Barnum effect is linked to personality traits and these links are moderated by the type of personality feedback that is provided.


Introduction
The Barnum effect, also known as the Forer effect, is the tendency of an individual to perceive vague and general personality descriptions as accurate and descriptive of himself or herself (Furnham & Schofield, 1987;Mason & Budge, 2011).It was first demonstrated by B. Forer who sought to prove that personality tests should not be validated using personal testimony from test-takers, and he called it "the fallacy of personal validation" (Forer, 1949).This classical experiment was replicated many times, and the Barnum effect had been in the center of attention of scientists for a few decades.Now, despite its relevance to psychological testing in general, the Barnum effect is all but forgotten, and there is little new research on the topic.To this day, the most comprehensive review of the topic is presented by Furnham and Schofield (1987).This is worrisome, because psychological testing is a substantial part of psychologists' practice; however, the acceptance of personality feedback is not clearly understood to this day.
Although the Barnum effect is traditionally defined as the acceptance of vague and general statements that could apply to most people, some studies venture beyond this definition and present stimuli that are comprised of inverted personality profiles (Greene, Harris, & Macon, 1979), or that describe positive or negative traits (Michels & Layne, 1980;Weinberger & Bradley, 1980), or are written in a positive or negative tone (Layne, 1978;O'Dell, 1972).Keeping in mind the varied stimuli used in the Barnum effect studies, it is adequate to define the Barnum effect as a tendency to accept personality feedback as true despite its validity (Poškus, Kairys, Liniauskaitė, & Žukauskienė, 2014).The present study is constructed with this definition in mind.
To our knowledge, there are no published studies that investigate the Barnum effect's links to personality traits, at least not in the Big Five paradigm.The scarce evidence of the Barnum effect's links to personality factors is quite outdated (Furnham & Schofield, 1987) and offers little practical use in today's science of psychology.It is known that the external locus of control is correlated with a greater acceptance of fake personality descriptions (Cuperman, Robinson, & Ickes, 2014;Snyder & Larson, 1972).Most recent research on the topic also suggests that the Barnum effect is related to schizotypy -the tendency to experience unusual cognitive and emotional states (Claridge, Clark, Powney, & Hassan, 2008;Mason & Budge, 2011).There is also some evidence that mixed-handed persons also are more susceptible to the Barnum effect (Christman, Henning, Geers, Propper, & Niebauer, 2008), so we can reasonably assume that innate personality factors do have some role in the acceptance of fake personality descriptions.
Another issue that arises from the variety of types of stimuli in the Barnum effect studies is that we cannot definitely say that those stimuli are equivalent.In fact, most studies find that different types of stimuli are accepted differently (Furnham & Schofield, 1987;Weinberger & Bradley, 1980).Therefore, it is possible that the acceptance of different types of personality feedback is linked to different types of variables.
There is some criticism directed to the concept of the Barnum effect itself.Research shows that the accuracy ratings of personality descriptions are related to actual traits of the subject (Layne & Ally, 1980;Layne, 1978;Michels & Layne, 1980), so the acceptance of personality descriptions may not be naïve, people may be accepting fake descriptions as accurate simply because of the high base validity of the presented stimuli.However, statements presented as feedback, rather than questionnaire items, are accepted more readily (Layne, 1978), so not all acceptance of personality feedback can be explained by actual traits of the subject.We propose that the acceptance of personality feedback is comprised of two components: the rational or objective part which is directly related to actual traits of a person, and a naïve or subjective part which could be confidently called the Barnum effect (Fig. 1).Asking a person to rate a description of his or her personality would encompass both these aspects, and it is up to the researcher to separate the rational part of the acceptance from the Barnum effect.
The proposed model integrates the classical assumption that overall personality description ratings produce the Barnum effect (Furnham & Schofield, 1987) and its critique -the assumption that ratings of separate statements of a personality description are highly correlated with actual traits (Layne & Ally, 1980;Layne, 1978;Michels & Layne, 1980).The aim of this study is to test the proposed model empirically and by doing so: • to provide a more concrete way of measuring the Barnum effect; • to explore the moderating role of the type of personality feedback to the links between personality traits and the Barnum effect.
Modified "Your NEO Summary".In order to accomplish our goals, we used a modified "Your NEO Summary" sheet as a stimulus for the study.The sheet was modified so that the participant's name was presented in the top right corner of the sheet.In the manner of the real "Your NEO Summary" sheet, presented in the top left was the title of the sheet: "Your PERSONALITY Summary".Below the title there was a short introduction based on the real "Your NEO Summary" introduction, stating that the summary describes five large personality traits that describe the reader in relation to his or her peers.After the description, five numbered statements were presented, each describing one of the Big Five traits.
Below the statements, there was a field asking participants to rate the accuracy of the personality feedback on a scale from 1 (entirely inaccurate) to 7 (entirely accurate).Below, the participants were asked to read their personality description once more and rate the accuracy of each statement individually.The rating scale was the same as before.
There were four types of possible experimental conditions (groups): • the control group received feedback that was accurate and based on their real personality profiles derived after completing the NEO PI-R questionnaire.Each personality trait was described using one of three possible statements from the "Your NEO Summary" sheet, describing either highly, averagely or slightly expressed trait.Statements were chosen based on the participants' T scores and the instructions in the test manual: a statement identifying a slight expression of a trait was presented for T scores of 44 and less, a statement of high trait expression was presented for a T score of 56 or more, for a T score ranging from 45 to 55, a statement of moderate trait expression was presented; • the inverted feedback group received personality descriptions that were as far from their real personality profiles as possible.For a T score of 50 points or more, a statement describing a slight trait expression was presented, and for a T score of less than 50 points, a statement of high trait expression was presented;  & Žukauskienė, 2014).All presented statements described averagely expressed traits applicable to T scores ranging from 45 to 55.
Participants.A convenience sample of 359 (79 males and 280 females) first and second year university students was chosen for the study.The mean participant age was 19.6 years (SD = 1.84), and ranged from 18 to 36 years.The design of the study required to meet with the participants two separate times a few weeks apart, and some of the participants did not show up the second time, so the final sample size was 243 participants (67.7% of the original sample, 49 males and 194 females); the mean participant age was 19.54 years (SD = 1.65) and ranged from 18 to 36 years.Before the second meeting, all participants had been randomly assigned to one of the four experimental groups: control group (final N = 61), inverted feedback group (final N = 64), positive feedback group (final N = 60), and universally valid feedback group (final N = 58).
Procedure.The researcher met with the participants at a prearranged time during their lectures.Participants were verbally informed that the aim of the study was to test the validity of a personality feedback sheet and those who were willing to volunteer would receive their personality descriptions after a few weeks.Those who volunteered to participate in the study were presented with NEO PI-R personality questionnaires and were asked to fill them out.Participants were also asked to provide their names or pseudonyms and their e-mail addresses in order to identify and, if need be, contact them.Participants were informed that all collected personal information would be destroyed upon completion of the study and would be used in no other way than to identify them in order to provide personality feedback.After participants had filled in the questionnaires, a second meeting was arranged.This procedure was repeated 9 times with groups of about 40 participants, and the second meeting time ranged from two to three weeks from the first meeting.
During the second meeting, the researcher provided the participants with personality feedback.Participants had been randomly assigned to one of four experimental conditions beforehand and received one of four possible types of personality feedback: actual (control), inverted, positive, or universally valid.Participants were asked to sit apart from one another by at least one seat in order to ensure their privacy.The researcher called aloud participant names and personally presented them with their feedback sheets.Participants were asked to privately read their descriptions and follow the additional instructions presented in the feedback sheet.When all participants had completed the procedure, the researcher asked them to return the feedback sheets.
Upon completion of the study, all participants were debriefed about the real aim of the study via e-mail and were provided with their real personality profiles and descriptions.Some participants responded to the debriefing via e-mail and expressed interest in the study, asking additional questions and providing positive comments.No negative feedback was received from the participants.
Results.First, the accuracy ratings of different types of personality feedback were tested to see if they depended on experimental conditions.The very nature of the Barnum effect suggests that the accuracy ratings of any type of personality feedback would be skewed, because people tend to rate most feedback as accurate; therefore, it was decided to use the nonparametric Kruskal-Wallis test for the analysis.It was found that the accuracy ratings differed significantly between experimental conditions (H(3) = 29.333;p < 0.01).For post hoc analysis, the Mann-Whitney test was used, and the significance was corrected for multiple comparisons.It was found that only the inverted feedback was rated as less accurate than other types of feedback (the median rating was 5 points out of 7, where 7 is "entirely accurate"), while there were no differences between other experimental conditions (median accuracy ratings in all other conditions were 6 points out of 7).The difference between inverted and actual personality feedback ratings showed the greatest effect (U = 995.5,d = -0.988),while comparing the ratings of the inverted feedback to universally valid (U = 1112, d = -0.776)and positive (U = 1217.5,d = -0.694)feedback ratings showed smaller effects.
It was tested if personality traits were adequately correlated with the ratings of separate feedback statements.Only the correlations in the positive feedback experimental group were analyzed, because it was the only group where a linear relationship between personality traits and acceptance of feedback statements could be observed (Table 1).
It was found that separate "Your NEO Summary" statements, rated as feedback, in fact correlated with actual personality traits of the rater, the only exception being the openness to experience trait.This might be due to some validity issues of the Lithuanian translation of the "Your NEO Summary" sheet; these issues are covered in the previous research by Poškus et al. (2014).On the whole, the observed correlations confirm the notion that separate feedback statements are rated quite rationally (Layne & Ally, 1980;Layne, 1978;Michels & Layne, 1980).It was also important to establish whether separate feedback statement ac-curacy ratings correlated with the ratings of the whole personality description.The results presented in Table 2 show moderate and strong relations between the variables in all experimental conditions, meaning that the rating of the whole description does indeed encompass some degree of objectivity and is not due only to the Barnum effect.
Before testing the model it was necessary to determine if there were relations between the acceptance of the whole feedback and personality traits, without controlling for objec-tive ratings.This type or similar correlational methods is quite common to Barnum effect studies (e.g., Claridge et al., 2008;Cuperman et al., 2014;Macdonald & Standing, 2002;Mason & Budge, 2011), albeit we would argue that it has significant drawbacks.As we can see in Table 3, there are some relations between personality traits and feedback ratings, most of which are different between experimental conditions.However, we cannot say that the observed correlations show the Barnum effect's relations to personality traits because we must bear in mind that these ratings might not be naïve.
To test the proposed model, a partial correlation analysis was conducted, correlating the ratings of the whole feedback to personality traits and their facets, controlling for objective ratings (ratings of separate statements).Results of this analysis are presented in Table 4.When controlling for separate feedback statement ratings, the correlations between personality traits and their facets changed, the old ones (Table 3) disappeared and the new ones (Table 4) appeared, demonstrating that the objective part of feedback acceptance was eliminated from the results, thus extracting that part of feedback rating which is irrational and could confidently be called the Barnum effect.It was decided to flag marginally significant correlations as well because of their potential relevance to the further research.The observed results support the proposed model and demonstrate its utility in the Barnum effect studies.

Discussion
The accuracy ratings of personality feedback between experimental conditions differed only to the extent that inverted feedback received accuracy ratings lower than other experimental conditions, while actual, positive or universally valid feedback received equally good accuracy ratings.These findings in general are consistent with previous research (Furnham & Schofield, 1987) and support the previous research done with the Lithuanian translation of the "Your NEO Summary" (Poškus et al., 2014).
A model of the relationship between personality traits, the received feedback type and feedback acceptance was proposed and tested.The data supported the proposed model, showing not only the twofold structure of feedback acceptance and that the Barnum effect is linked to personality traits, but that these links are moderated by the type of feedback used to produce the Barnum effect.
The Barnum effect was found to be correlated with various personality facets among experimental groups, but pro bably the most easily interpretable is the Barnum effect's correlation with the trait of agreeableness and some of its facets in the inverted feedback experimental group.These results are somewhat reminiscent of the studies demonstrating that people with an external locus of control are more susceptible to the Barnum effect (Cuperman et al., 2014;Furnham & Schofield, 1987;Snyder & Larson, 1972), however, links between agreeableness and the Barnum effect are observed only in that one experimental condition, although some facets of agreeableness are correlated with the Barnum effect in all experimental groups.
It is important to note that in the actual feedback (control) group there were no significant correlations between the Barnum effect and personality traits.One marginally significant inverse correlation between the Barnum effect and the facet of altruism was observed, however, for the most part the acceptance of actual feedback was not affected by the Barnum effect.These findings provide additional proof of the validity of the Lithuanian translation of the "Your NEO Summary".
One of the limiting factors of the study was its sample: the male population was underrepresented, excluding the possibility to investigate gender differences.Further research should be done with a larger, more representative sample.Another potentially limiting factor could be the stimuli that were used, since the stimuli were identifying the very traits to which their acceptance was being correlated.Further studies should investigate the acceptance of feedback derived from tests other than those encompassing the variables the feedback ratings are being correlated to.
The findings of this study suggest that acceptance of personality feedback is in-deed twofold, consisting of a rational evaluation of one's own personality traits as well as of a subjective component -the Barnum effect.Therefore, we suggest that researchers engaged in studying the Barnum effect should avoid using the traditional overall personality feedback ratings as a measure of the Barnum effect.Rather, researchers should not only gather the acceptance ratings of the whole personality description but its component statements as well.Any correlational analysis between the acceptance rating of the whole personality description and other variables should be done controlling for the ratings of separate statements of the description.In this way researchers could obtain a more objective measure of the actual Barnum effect and not the rational acceptance of personality descriptions.

Figure 1 .
Figure 1.Theoretical links between personality traits and acceptance of feedback.

Table 2 . Rank (r s ) correlations between the ratings of accuracy of the whole personality descrip- tion and the ratings of separate feedback statements
Note: RSS = rating of separate statements; N = neuroticism; E = extraversion; O = openness to experience; A = agreeableness; C = conscientiousness.*p < 0.05, **p < 0.01.