PART II

EMPIRICAL PAPERS

PAPER 3

An experiment onthe recall of reputation gossip

Abstract

In this paper I present the results of an experimental study about the recall of Reputation Gossip. I focus on Reputation Gossip (RG) that concerns trait/behavior information attached to a specific other person. I presented 85 students from the University of Antwerp 16 gossip stories about 25 imaginary soon to be colleagues. After a thirty minute distraction, they were subjected to a surprise recall test. Scope of the study was to investigate selection in recall according to different conditions.

In general the results show that respondents remember a lot from the presented RG, which implies that differences between-respondents and within-respondents for different stories are rather low. My results indicate the chances of future encounters might not be a primary decision criterion to decide whether to store or ignore RG. Next, when eliciting ‘as if’ encounters with the gossipees, by adding pictures to the presented RG, the recall rates significantly increase. Third, focusing on Mating RG I could not support the prediction that single respondents would recall more about single opposite sex gossipees than their colleague respondents who reported to be in a long-term relationship. For Mating Structure RG I could not find support that gossip about troublesome relationships would be recalled more than gossip about good relationship. My results do support the hypothesis that cues of attractiveness are recalled more for female gossipee than for male gossipees, while cues of wealth status are recalled more for male than female gossipees. Last, I added some information about cheaters and altruists among the gossiped about (imaginary) soon to be colleagues. However recall rates for this kind of Co-operation RG were very low.

1 Introduction

In chapter 4 I have proposed a classification system for gossip as a general noun. I distinguished Strategy Learning Gossip from Reputation Gossip. In this paper I focus on Reputation Gossip. Reputation Gossip (RG) is information about the traits or behaviors of specific other individuals. The gossiped about traits and behaviors cannot be detached from the gossipee, since RG then looses value.

RG functions to teach receivers about the reputations of other people and for senders to manipulate these reputations. While senders of RG benefit most from manipulating the reputation of RG gossipees, for receivers RG is most beneficial to learn from the reputations of other individuals. RG teaches receivers about who is skilled, talented, who is a cheater and who is an altruist, who would be a good potential mate, who can be a rival, a friend or a foe, and who is related to whom.

Gossipers must know the gossipees of RG for the RG to be functional. If gossipees are unknown to the gossipers, the gossipers do not benefit to learn about these unknown gossipees, nor do they benefit from manipulating the reputations of unknown gossipees.

1.1 Out of sight, out of reach, and out of gossip

1.1.1 Who we can meet in the future is who we gossip about

Encounters with the gossipees of Reputation Gossip are crucial for the value of RG. Since RG is not functional for gossipees with whom we do not interact, individuals who we have never met and who we are unlikely to meet in the future are worthless to gossip about.

In chapter 5 I have put forward a graphical decision tree to outline how human decision making for acquiring gossip occurs. The criteria, receivers of RG take into account in their decision whether they will ignore the received RG or store the received RG and calibrate their attitude towards the gossipee, are ‘believability’ and ‘encounter opportunities’. When the information is unlikely to be true, the best strategy for receivers is to ignore this. When they believe the content is true, the best strategy is to estimate future encounter possibilities. If it is unlikely that the receiver will ever meet the gossipee in the future, the gossip can be best ignored. If future encounters are likely, the receiver optimally stores the RG information and calibrates his or her attitude towards the gossipee (see figure 3.1).

A cue we can easily rely on to estimate the chances of future encounters are past encounters with an individual. If we have met an individual in the past, this signals that this person is somehow part of our social network. Either we know this person directly, or we know this person indirectly, as a person related to someone we know directly. Anyhow, if we have encountered an individual in the past, it is likely that we can encounter this person in the future again. I have therefore put in this criterion in the decision tree (see figure 3.1), as a second criterion to decide whether to store or ignore received RG. If the outcome of estimating this criterion is negative, this is, if encounters in the past are not a fact, the receiver must try to estimate future encounters using other cues.

Figure 3.1. Fast and frugal tree for acquiring Reputation Gossip

Because gossipees with whom future interactions are unlikely, are not relevant for RG receivers, I predict that:

Hypothesis 1:

Reputation Gossip about individuals with whom chances of future encounters are low will have lower recall rates than Reputation Gossip about individuals with whom chances of future encounters are high.

1.1.2 Who we see is who we meet

It has been shown (Furnham et al, 2002) that picture-loaded information elicits higher recall rates than non-picture loaded information. Reason for this might be because we process pictured information ‘as if’ we experience the portrayed situation as real. Barkow (1989, 1992) suggests that our interest in celebrities stems from our daily encounters with images of these people in media products. We process this pictured information ‘as if’ we encounter these people in real life. Who we see is who we meet.

Funded on Barkow’s (1989,1992) idea I assume that presenting people pictures about imaginary characters elicits feelings ‘as if’ the audience encounters these people. Having seen an image, being able to recall the face of imaginary characters, receivers experience feelings ‘as if’ they have encountered these characters. Because past encounters are triggered in respondents who see pictures of characters, these respondents should be more eager to remember RG about the characters they have seen (encountered). Receivers who are given RG information with no pictures of the imaginary characters present do not experience ‘as if’ they have encountered these individuals. Therefore I predict that:

Hypothesis 2:

Receivers who are presented Reputation Gossip with pictures of the gossipees added will have higher recall rates than receivers who obtain Reputation Gossip with no pictures added.

1.2 Mating Reputation Gossip

I have narrowed down Reputation Gossip into more detailed sub categories that each have different specific functions. On a first classification level within RG I have distinguished Mating RG from Social RG. Social RG functions to teach receivers about the reputations of others that are important for social (non-sexual) interactions, and to manipulate these social reputations, when Social RG is shared. Examples of crucial social reputations are for instance having a good or bad reputation as being a cheater or an altruist in the context of co-operation, having skills which make you a valuable friend, and so on (see chapter 4, section 4 for more details).

Mating Reputation Gossip functions to teach receivers about the reputations of others as valuable mates and to manipulate these reputations when sharing Mating RG. In this paper I focus on Mating RG that is most valuable for receivers. Receivers benefit from hearing information about the reputation of a potential mate (Mates Detection RG), from learning who is a rival in their search for a potential mate (Sexual Rival Detection RG), and they benefit from hearing who is having a relationship with whom (Mating Structure RG).

1.2.1 Mates Detection RG

Mates Detection RG informs receivers about the reputation of others as potential mates. Although both men and women who are engaged in a relationship still potentially seek extra or other potential mates (Buss & Schmitt, 2001), Mates Detection RG is most valuable for receivers who are not (yet) engaged in a relationship. Likewise, although individuals who are in a committed relationship can be good potential mates to poach, individuals who are single still are more frequent target of Mates Detection RG. Therefore I predict that:

Hypothesis 3a:

Recall rates for Mates Detection Reputation Gossip about single male gossipees will be higher of single female receivers than for female receivers who are engaged in a relationship.

Hypothesis 3b:

Recall rates for Mates Detection Reputation Gossip about single female gossipees will be higher for single male receivers than for male receivers who are engaged in a relationship.

Because men and women desire different traits in potential mates (Buss, 1999), I expect different traits to be recalled by receivers for male and female gossipees of Mates Detection RG. In general, men value traits that signal fertility and reproductive value of potential female partners. Such cues are for instance youth and physical attractiveness. Women, on average, desire investment of male potential partners, and value traits that signal availability of wealth and willingness to invest, such as wealth status and willingness to commit (Buss, 1994). I therefore expect that:

Hypothesis 4a:

For Mates Detection Reputation Gossip with cues signaling physical attractiveness, recall rates will be higher for female than for male gossipees

Hypothesis 4b:

For Mates Detection Reputation Gossip with cues signaling wealth status recall rates will be higher for male than for female gossipees

These hypotheses are not only true for Mates Detection RG, but will exactly the same for Sexual Rival Detection RG. In our search for potential mates we have to deal with same-sex rivals (Schmitt & Buss, 1996). Sexual Rival Detection RG functions to solve the problem of detecting who our sexual rivals are. Therefore I expect that female respondents will recall information about valued traits of same-sex competitors, which are cues of youth and physical attractiveness. Men will recall gossip about valued traits of same-sex competitors as well, and these traits are for instance wealth status.

In general both men and women will recall gossip about valued traits of both male and female gossipees, but for different reasons. Men will store gossip about the physical attractiveness of female gossipees to learn who is a good potential mate. Women will recall the same information, not to learn who is a good potential mate (at least not in the context of heterosexual relationships) but to learn who is a threatening rival. Women will recall cues about the wealth status of male gossipees to learn who is a good potential male mate. Again, men will recall the same information, not to learn who is a good potential male mate (I here only focus on heterosexual relationships) but to learn about threatening same-sex rivals.

1.2.2 When the relation is troubled, the gossip gets doubled

As I have already said, in their search for potential mates, humans do not restrict their options to individuals who are not committed in a relationship, poaching occurs both among men and women (Buss & Schmitt, 2001). In our search for potential mates, getting information about ‘who sleeps with whom’ is valuable. Mating Structure RG is information about the sexual relations between members of our social network, and functions to inform others about who is (sexually) related to whom.

Mating Structure RG is extra valuable when information about the quality of ‘who sleeps with whom’ is present. If you hear two people are happily committed, you learn that chances of poaching one of these two individuals are unlikely. However, hearing a relationship is troubled teaches receivers that poaching might be an option. This latter form of troublesome Mating Structure RG is highly valuable for individuals searching a potential mate.

Taken this into account, combined with the fact that in general bad news has a greater impact and is retained better by receivers (e.g. Ito, Larsen, Smith & Cacioppo, 1998) I expect that Mating Structure RG about relationship troubles will be retained better than Mating Structure RG about good relationships:

Hypothesis 5

Recall rates for Mating Structure RG about bad relationships will be higher than for Mating Structure RG about good relationships.

1.3 Social Reputation Gossip

Although in this paper I focus mainly on Mating RG, I did add some Social RG information in the study I here discuss. Social RG, as I already mentioned, functions to teach receivers about the reputations of social network members. Especially Co-operation RG is of great value for receivers. This form of RG functions to label co-operators with a good (altruist) or bad (cheater) reputation as co-operator.

It has been shown that humans possess a cheater detection mechanism (Cosmides, 1989, Cosmides & Tooby, 1987, 1992) and might possess an altruist detection mechanism as well (Brown & Moore, 2000). However, as Dunbar (1998) has commented, these mechanisms function well in small social settings where every individual interacts with every other individual on a regular basis, but in larger social settings cheaters can operate without being noticed. Dunbar (1998) therefore suggests that communication, and more specific gossip, functions to exchange information about the reputation of co-operators. Through gossip cheaters can be detected and signaled to others. The idea that reputations are crucial in the context of co-operation has been stresses by many other researchers as well (see e.g. Alvard & Nolin, 2002; Gintis et al, 2001; Gurven, in press; Hawkes 1991; Smith & Bliege Bird 2000).

Since knowing who acts altruistic and who cheats is of great relevance, I expect that any receiver will recall this kind of information about gossipees with whom future interactions are likely:

Hypothesis 6:

All receivers will recall Co-operation RG about cheater and altruist gossipees with whom future encounters are likely.

2 Method

To test the predictions I outlined above, I presented 85 respondents Reputation Gossip stories to read. After a distraction of thirty minutes they were presented a surprise recall test.

2.1 Apparatus

The gossip stories were presented in written style to the respondents. Every respondent received a 4-pages booklet with all stories, either with or without pictures added. The pictures I used were all non-colored.

2.2 Materials

Initial oral instructions informed the respondents that the experiment consisted of two related sessions. They were told that they would first have to read some stories, and then fill out a questionnaire about their personality and social life. I informed them that I wanted to investigate the influence of reading information on filling out questionnaires, and that another group would get other information to read. They were not at all aware that the experiment involved a recall test about gossip stories. Only at the end of the experiment the respondents were informed about the actual experimental scope.

In total each respondent was presented 16 stories about 25 characters (for a list of these stories see attachment of this paper). All imaginary gossip stories were about ‘soon to be colleagues’. I asked the respondents to imagine they would soon be employed at an office, where they would have to co-operate with several new colleagues. I told them to imagine that their office co-worker took them out to an informal party were most of these new colleagues were present. This co-worker then told them some information about each new colleague. The sex of the gossip sender was not mentioned.

To rule out biases in recall because of the order wherein the gossip stories were presented, I distributed four different versions of the gossip conversation. In each condition all 16 stories and 25 characters remained the same, but the order wherein they were presented was shuffled. This shuffling was done randomly. Each gossip story was given a number, and these numbers were shuffled four times to define the order of each condition.

Half of the gossip stories-booklets I distributed were plain text with no picture added. The other half of the booklets consisted of the exact same gossip stories, but pictures of the fictive characters were added. The pictures I used were taken from internet websites and were all unknown people, with whom no familiarity from the respondents’ side could be expected.

2.3 Procedure

The gossip-stories booklets were presented to the respondents during 10 to maximum 15 minutes, and were taken away as soon as they had finished reading all 16 gossip stories. Next, they were distracted with a survey about their personal and social life that took about 25 minutes to fill out. At the end of the survey each respondent received a personal number to remember. I had informed them that they had to keep this number, because I would organize a lottery to win a prize, as a reward for participating in this study.

Thirty minutes after the gossip-stories booklets were taken away, the respondents were informed about the recall test, and the fact that their personal number was needed for this. No lottery was organized to reward the participants, but each respondent received a small gift (CD-single). I distributed blank sheets on which they had to write down their personal number, and all they recalled from reading the gossip stories.

2.4 Design

The experiment consisted of a 2 (picture condition: pictures added vs. no pictures added) X 4 (sequence of gossip stories) mixed factorial design with picture condition and sequence of the gossip stories as between subjects factors. Respondents were randomly assigned to each of the eight groups resulting from the combination of the two between-subject factors.

2.5 Participants

The experiment was conducted with 85 student participants at the University of Antwerp. In total 22 males and 63 females participated. Their mean age was 19.45 years (S.D.= 1.67). Of all 85 participants 45 were single and 40 were in a committed long-term relationship at the time the experiment was conducted.

Although this experiment was run at a Belgian university (University of Antwerp) not all respondents had the Belgian nationality. Of all 85 participants 56 were Belgians, 27 were from the Netherlands and 2 were German. However, most of them were living in Belgium; only 4 reported to still live in the Netherlands. Of these 81, 69 participants lived in the Province of Antwerp.

3 Results

For each of the 16 gossip stories, I encoded if the respondents remembered something from this “1” or not “0”. Only correct recalls were encoded. If respondents remembered wrong information, this was not encoded. Further I encoded every trait and behavioral action of all 25 characters as present “1” or non-present “0” in the recall reports of all 85 participants.

In general respondents remember much from the presented information. The average recall rate is 10.70 (S.D.= 2.71) of a total of 16 RG stories. The minimum amount of RG stories recalled by 1 respondent is 3, while the maximum is 16. Two respondents succeeded to recall something from all presented RG stories. Using a T-test for Equality of Means for Independent Samples I looked at sex differences in this overall recall rate. I notice that female respondents significantly (p<.05) recall more gossip stories (M_female= 11.10) than the male respondents (M_male= 9.59).

To investigate the effect of potential future encounter rates (hypothesis 1), I looked at the recall rates for those soon to be colleague characters about whom I had explicitly mentioned information that can influence the future potential encounters. Of the 25 characters 18 were soon to be colleagues and 7 were partners of these soon to be colleagues. Future encounters with these 18 colleagues is most likely for the respondents. However, I manipulated this by ensuring low potential encounter rates for three characters. About male character “John” it was mentioned he lives in New York and is only around for advertisement campaigns. Female character “Stephanie” lives and has an office in London, so she is not around a lot either. Third, about “Emily” respondents were informed that she would soon move to Geneva, and that she would not be around a lot anymore. About three soon to be colleagues I mentioned explicitly they lived in Antwerp. Since most respondents (69) live in Antwerp as well, encounters with “Mary”, “Ricardo”, and “Rob” are most likely in the future, both at the office an maybe even outside the office as well. An overview of residence information and all general recall rates for each character can be found in table 3.1.

Table 3.1. Recall rates of N=85 respondents for each of all 25 characters

If potential future encounter rates are the first decision criterion receivers use to decide whether they store or not store Reputation Gossip about specific other individuals, it can be expected that the presented RG about Emily, John, and Stephanie will be recalled by fewer respondents than RG about other characters, with whom future encounters are more likely. From the results (see table 3.1) follows that recall rates for RG about Emily indeed are low (34.5% of all respondents recalls something about her). Still, recall rates for both John (78.8%) and Stephanie (89.3%) are very high compared to the others (see table 3.1). The presented RG information is different for these three characters, but if the decision tree model for acquiring RG I presented in chapter 5 were correct, respondents would have not paid as much attention to all three characters with whom future encounters are fairly unlikely.

For the characters of whom I explicitly mentioned they lived in Antwerp (Mary, Ricardo and Rob) I had expected highest recall rates, if chances of encountering the gossipee in the future are a primary decision criterion to decide whether to store or ignore RG. Recall rates for Mary (70.6%) and Rob (73.8%) are high, but recall rates of Ricardo (50.0%) are moderate. Again this does not confirm my prediction strongly.

Next, using a T-test for Equality of Means for Independent Samples I computed the difference in recall rates between the respondents who were presented the gossip stories with pictures (N1= 49) and the respondents who were presented the gossip stories without pictures N2= 35, 1 missing value). In line with what I predicted in hypothesis 2, I found a significant (p<.05) difference between both groups. As predicted those respondents who received gossip stories with pictures added have higher recall rates (M_pictures= 11.33) than those respondents who did not get pictures added to the gossip stories (M_no-pictures= 9.83).

For the third hypothesis on the difference between single/non-single respondents in recalling Mates Detection RG about the single gossipees, I had to compare recall percentages of single vs. non-single respondents. I compared the recall percentages of single female respondents to the recall percentages of female respondents in a relationship for Mates Detection RG about the single male imaginary gossipees (John, Ricardo, Dirk, Rob). Likewise I compared the percentages of single male respondents and male respondents in a relationship recalling Mates Detection RG about the single female imaginary gossipees (Mary, Aida, Stephanie, An, Emily).

I first selected only the female respondents from my research population (N= 63), and split my file to compare groups for variable ‘single’, where “1”= single and “0”= not single, but in a committed relationship. Looking at the results (see table 3.2) there is some indication that single female respondents score slightly higher on recalling RG about single imaginary male gossipees than female respondents in a relationship, but none of these differences are significant (using Statistica Other Significance Tests for Difference Between Two Proportions). I comment however that this might be due to the low number of respondents in this study.

Table 3.2. Recall rates for Mates Detection RG about single male imaginary characters for single/non-single female respondents (N= 63)

* Statistica Other Significance Tests/ Difference between two proportions

After that I selected all male respondents from my research population (N= 21, 1 missing value), and split my file again to compare groups for variable ‘single’, where “1”= single and “0”= not single, but in a committed relationship. Looking at the results (see table 3.3) there is no indication that single male respondents score higher on recalling RG about single imaginary female gossipees. Again, this might be due to the low number of respondents of this study. Male respondents who are single or in a relationship either score equal in the valid percentage of respondents recalling gossip about single female imaginary gossipees, or the male respondents who are in a relationship score even higher than the single male respondents. Again none of these differences are significant though (using Statistica Other Significance Tests for Difference Between Two Proportions).

Table 3.3. Recall rates for Mates Detection RG about single female imaginary characters for single/non-single male respondents (N= 21)

* Statistica Other Significance Tests/ Difference between two proportions

Overall, hypotheses 3a and 3b cannot be confirmed.

Turning to hypothesis 4a, I tested whether cues of physical attractiveness are more recalled for female gossipees than for male gossipees and whether wealth status cues are more recalled for male vs. female gossipees. I here compared recall percentages of all N=85 respondents. I compared detailed information of single female gossipee “Stephanie” and single male gossipee “Ricardo”. The presented RG information about these gossipees was similar. Both were described as young, attractive, creative, social, and working as graphic designers for the company. Both are graphical designers, which are quite prestigious positions. I used their job as a cue for wealth status. About both was mentioned they were physical attractive, which was a direct cue to measure recall of physical attractiveness.

To compare the recall rates of both gossipees, I first selected all respondents who recalled something from the RG information about Ricardo (N1= 42), and looked into detail which percent of this sub population remembered the fact that Ricardo was a graphic designer (as cue for wealth status) and which percent of this sub population remembered the fact that Ricardo was attractive (as cue for physical attractiveness). I then selected the respondents who recalled at least something about the presented RG about Stephanie (N2= 76) and again looked which percentage of this sub population remembered that Stephanie was a graphic designer (as cue for wealth status) and which percentage of this sub population remembered Stephanie was physical attractive (as cue for physical attractiveness).

As predicted the wealth status cue was recalled by more respondents for the male gossipee (Ricardo: 45.2% of N1= 42) than for the female gossipee (Stephanie: 28.9% of N2=76). This difference is marginally not significant (p= 0.08 using Statistica Other Significance Tests for Difference Between Two Proportions). Looking at the results for the physical attractiveness cue, I did find highly significant (p<.001 using Statistica Other Significance Tests for Difference Between Two Proportions) differences for both gossipees. From all 76 respondents who recalled something about Stephanie, 71.1% remembered she was physical attractive. For Ricardo only 33.3% of all 42 respondents who remembered him recalled he was physical attractive. An overview of this information is presented graphical in figure 3.2.

Figure 3.2. Percentage recall of cues to physical attractiveness and wealth status for a male and female gossipee

I have explained in the introduction that both male and female respondents would recall this information. Only their motives are different. When looking at the differences between male and female respondents for recalling cues of physical attractiveness and wealth status for a male and female gossipee, I indeed did not find significant differences (using Statistica Other Significance Tests for Difference Between Two Proportions, see table 3.4).

Table 3.4. Male Female differences in recall of gossip about physical attractiveness and wealth status for a male and female gossipee

* Statistica Other Significance Tests/ Difference between two proportions

In hypothesis 5 I predicted that recall rates would be higher for gossip about troubled relationships than for gossip about good working relationships. In the gossip stories I presented 6 stories reported about the relationships of 7 soon to be colleagues (one story had two soon to be colleagues as gossipees). Three stories reported about the good relationship of soon to be colleagues, and can be labeled as positive Mating Structure RG. In the first story the respondents learned that one of their soon to be male colleagues was happily in love with his new girlfriend, whom he recently met. The second good Mating Structure RG was about a young couple who are expecting a baby soon. The third and last positive Mating Structure RG reported about an established long term relationship of one of the older soon to be managers of the company. He is happily married and has three kids.

The three bad Mating Structure RG stories concerned troublesome relationships. In the first bad Mating Structure RG a couple is mentioned where the boy parties a lot and the girl (soon to be colleague) is jealous and complains about her boyfriend a lot. The second troublesome Mating Structure RG reports about two soon to be colleagues who are present at the party with their partners, but who are having an affair. The boyfriend of the cheating girl suspects something and many people of the company already gossip about this affair. The last bad Mating Structure RG talked about the boss who was about to divorce his wife because he no longer loved her and actually fancied one of the female soon to be colleagues.

Looking at the recall rates of all N=85 respondents for these six Mating Structure RG stories, two troublesome Mating Structure RG stories get very high recall rates. The story about the cheating colleagues is remembered by 89.3% of all respondents, and the gossip about the boss divorcing his wife was recalled by 92.9% of all respondents. This last percentage is significantly (p< .05, using Statistica Other Significance Tests for Difference Between Two Proportions) higher than the highest valid percent recall from the three good Mating Structure RG’s (79.9% for pregnant couple). Still, the bad Mating Structure RG about the party boy and his jealous and complaining girlfriend does not get high recall rates (31.8%). This low score is even very significantly (p< .01) lower than the lowest score of recall for good Mating Structure RG (57.1% for the couple with kids). In general, the three positive Mating Structure RG’s get rather high recall rates as well. For all three more than half of the respondents recall something (see table 3.5).

Table 3.5. Recall rates for Mating Structure RG about troublesome and good relationships

Overall, these results cannot confirm my prediction that Mating Structure RG about troublesome relationships would be remembered by more respondents because of the higher value and negative news value.

Last, turning to the results of the Social RG information, I first present the recall rates for Cheater Detection RG. I have pictured one of the soon to be colleagues as a potential cheater, by informing the respondents that he most probably has stolen files form his colleague-secretary. I expected that this form of Cheating Detection RG would be recalled by all respondents. However, only 63.1% of all 84 (1 missing value) respondents remembered secretary Mark (labeled as cheater), and of those only 61.9% explicitly recalled him as a cheater. This is, of all 84 respondents, only 34 (40.5%) mentioned in their recall test that they remembered Mark had (probably) stolen files from his colleague. Thus, not even half of the respondents remember this (highly important) information, which does not support my prediction.

I had mentioned about two soon to be colleagues that they were very helpful, labeling them as altruists. I informed the respondents that Isabel always listens to anyone’s troubles and is very supportive (emotional support), and that Rob is a very helpful person, especially when it concerns technical problems (technical support). Of all 84 (1 missing value) respondents only 5 (6.0%) recalled that Isabel was the person to turn to when you wanted to talk about your problems and get support. The valid recall percent for Rob being a helpful technician is 11.9% of all 84 (1 missing value) respondents. Also these results do not at all confirm my prediction that respondents would recall altruists among their soon to be colleagues.

To conclude my results, I want to mention that I had informed my respondents that secretary Jenny was the source of social information (gossip) of the company. Little more than half (56.0% of all 84 respondents) recalled this, and no sex difference was found for this. Of the 22 male respondents 54.5% recalled Jenny as the source of gossip, and 56.5% of the 63 female respondents did the same.

4 Conclusion

Presenting 16 gossip stories about 25 characters to 85 student respondents from the University of Antwerp I wanted to elicit selection in recall of the presented gossip. However, in general the 85 participants of this study recalled a lot from the 16 presented gossip stories about 25 characters, so that analyzing differences in recall rates for different conditions was hard to establish.

I had expected that RG about gossipees with whom future encounters could not be secured would be remembered less than RG about gossipees with whom future encounters were more likely to occur. My results however do not really support this hypothesis. Some RG about gossipees with whom future encounters were unlikely got high recall rates, while some RG about gossipees with whom future encounters could be assured got low recall rates. I suggest future research to further investigate whether chances of future encounters with the gossipee are a primary criterion to decide whether to store or ignore RG. Maybe my graphical decision trees need to be adapted.

My prediction that RG stories with pictures of the characters added would elicit stronger recall rates can be confirmed. Barkow (1989, 1992) has suggested that our minds process visual stimuli of individuals as some kind of encounters with these individuals. Presenting pictures I created situations ‘as if’ respondents encountered the (imaginary) characters. I had predicted that this fake past encounter would elicit more interest and better recall, since RG makes most sense for gossipees with whom encounters can be secured.

For Mating RG I had predicted that single female respondents would remember more about the single male gossipees and that single male respondents would recall more about the single female gossipees than vice versa, but my results cannot confirm this at all. Also the prediction that more would be remembered about troublesome relationships, because this signals poaching opportunities, and also because bad news impacts more in general, cannot be confirmed with this study. What I did confirm was my prediction that cues about physical attractiveness would be recalled more for a female gossipee than for a male gossipee. Although not significant and not as strong as for the gossip about attractiveness, I also found some support for the fact that cues about wealth status are recalled more for a male than for a female gossipee.

Surprisingly gossip about potential cheaters and altruists among the imaginary soon to be colleague gossipees was recalled by very few respondents. The fact that one of the gossiped about soon to be colleagues had most probably stolen files was remembered by about 40% of the respondents. For the information about their helpful (altruistic) imaginary soon to be colleagues, the recall rates were very low (6.0% and 11.9%).

5 Discussion

Because the general recall rates in this study were so high, it was difficult to look at differences for the different conditions. I therefore suggest future researchers to present more gossip stories about more characters, so that selection in recall can be assured.

The method I used to investigate the impact of potential future encounters might not be the most optimal method. The RG information I presented for the three characters, with which future encounters were unlikely, was very different. About Emily little RG was given, while Stephanie was gossiped about in more details, and was also part of the ‘juicy’ gossip that the big boss fancied her a lot. I therefore suggest future research to further investigate this, by using similar gossip stories and only manipulating chances of future encounters with the gossipee.

This critique also counts for my results on Mating RG. The gossip stories about the single gossipees and the couples gossiped about, were sometimes very different from each other. A variety of other factors might influence my data, and more controlled experiments are necessary to investigate if single respondents recall gossip about single gossipees better than respondents in a relationship. Also for the Mating Structure RG’s I suggest future researchers to reply this study with stories that are not as different on other aspects. In this study the future encounter opportunities, age and status of the gossipees might be too different to control for good/bad relationship information.

The two gossip stories about the single gossipees being wealthy graphic designers and being physical attractive as well were fairly similar. My results are in proof of my prediction, but I still suggest future research to investigate more cues, such as for instance the age of female gossipees and willingness to commit of male gossipees.

Last, another interesting experimental design to test the impact of pictures added to RG information would be to first present respondents with a series of pictures, and then to present them with RG stories about gossipees that either were present or not present in the series of pictures. It can be expected that RG stories about gossipees of whom pictures have been shown before the RG was transmitted will be better recalled than RG stories about gossipees of whom no pictures had been shown at all.

Attachment: used Reputation Gossip stories

I here present the Reputation Gossip stories I used in this study in a random order (version A of the distributed booklets) and with pictures added:

“Imagine you are graduated and you are just employed to start working at a new company. This means that soon your social network will be expanded with soon to be colleagues, working in this company. To introduce you to your new social environment, your office companion takes you to an informal party, organized by one of the other employees of the company. Your office companion wants to introduce you to some of the employees, who will soon play an important role in your daily life at the office and maybe even in your private life as well. You are talking, just the two of you and observing the other people. Your office companion tells you the following information:

And this conversation continues like that for quite a while...