PART II

EMPIRICAL PAPERS

PAPER 6

Celebrity gossip:

learning through gossip about the experiences of parasocial members of our social networks

Abstract

Part of our daily gossip conversations are about celebrities. One way to explain all of us our interest in the private lives of these people is the fact that we can vicariously learn from their behavior. This is what I call the Learning Hypothesis. In addition to this hypothesis, a second explanation for our interest in talk about celebrities is embedded in the Parasocial Hypothesis. Some of us regard celebrities as being members of our social networks, with whom they have parasocial, or one-way, interactions. The Parasocial hypothesis says that some are interested in Celebrity Gossip for the same reasons as they are interested in gossip about real social network members. Questioning 838 Belgian people about their interest in Celebrity Gossip and their real life social network, I found support for the Learning Hypothesis rather than for the Parasocial Hypothesis. Age seems to be the most important factor in explaining interest in gossip about celebrities. Younger, less experienced people, are interested the most and as age increases, the interest in presented celebrity gossip stories a declines to almost no interest at all for older, more experienced respondents.

Since, in evolutionary terms, women have been more subject of relocation to new social environments, I further argue that women are the more active social sex. From this I expect that women will have a higher interest in gossip about celebrities. My data confirm this hypothesis, showing that women are more interested in celebrity gossip and that the effect of age on the interest in celebrity gossip is significantly stronger for female respondents than for male respondents.

1 Introduction

When was the last time you saw Jennifer Aniston? Maybe yesterday, when watching ‘Friends’. Chances that you ever met, or will ever meet her in real life are as good as non-existing. Since we never encounter celebrities in real life, these people will always be some kind of strangers to us. Still on the other hand, the media overload us with information about celebrities. We read gossip stories about them, become interested in those people’s private lives and start gossiping interpersonally as well about those famous persons with real friends. In this paper I explore which mechanisms drive this behavior.

1.1 Gossip as a learning strategy

Using an evolutionary approach to explain our interest in celebrity gossip I suggest that one possible explanation is embedded in the learning function of gossip. Baumeister et al (2004) have suggested that gossip can be regarded as a learning device. They see gossip as an extension of observational learning. In my classification system of gossip (see chapter 4) I have distinguished Strategy Learning Gossip from Reputation Gossip on a first level of classification. The first focuses on the exchange of information about fitness-relevant strategies. Receivers of SLG vicariously learn about successful and unsuccessful strategies from the trial and errors of others. It is not the gossipee that makes the content of SLG relevant to a recipient, but the behavior strategies of the gossipee. Gossipees of SLG are mere carriers of fitness-relevant strategy information. Reputation Gossip, on the other hand, functions to teach receivers about the reputations of members of their social network, and enables senders to manipulate these reputations in their own benefit.

The, what I call Learning Hypothesis says that celebrities, being unknown subjects to an average media audience person, are mere carriers of fitness-relevant behavior strategies. Interest in Celebrity Gossip rests on the fact that celebrity gossipees carry over fitness-relevant strategy information. Celebrities teach receivers of Celebrity Gossip about successful and unsuccessful strategies. Moreover, celebrities are prestigious people, and it has been shown that mimicking the behavior of higher status people is an adaptive strategy, leading to an increase of your own status (Henrich & Gil-White, 2001; Henrich et al, 2001). In this perspective, it is far from being ridiculous to show interest in celebrities’ private life and acquire gossip about them, since you (accidentally) learn from this information.

1.2 Celebrities as parasocial members of our social networks

In 1984 John Caughey wrote a book about our ‘Imaginary Social Worlds’, in which he explains how people often overlook the fact that most of us have ‘imaginary’ friends such as, fantasy figures, spirits and media stars: “people characterize unmet media figures as if they were intimately involved with them, and in a sense they are…” (Caughey, 1984: 33). Caughey (1984) and Kanazawa (2002) as well, use the term ‘imaginary friends’ since even though the celebrities do exist and are not imaginary people, the friendships with celebrities are not real. Actually our interactions with celebrities are one-way interactions. Celebrities reveal their private lives to us, we show emotions towards them, but we don’t share our private issues with them, and they don’t get emotional over us. These kinds of relations, where reciprocity is lacking, are called parasocial interactions (Horton and Wohl, 1956; Rubin, Perse, and Powell, 1985). Labeling celebrities as our parasocial friends or one-way friends still only covers part of the situation. Some may regard Jennifer Aniston as a friend, and others will consider her as an enemy, or some men might even think of her as being a potential partner. Since celebrities are not only our friends, but can take different social positions, just like real members of our social network, I will refer to them as parasocial or one-way members of our social network. This term covers both the kind of relation we have with them and the different kinds of positions these people can take.

Framing this parasocial membership of our social networks in an evolutionary perspective, Barkow (1989, 1992) explained why those one-way interactions exist. Barkow (1992) says that Celebrity Gossip is a recent by-product of Interpersonal Gossip, and that our stone aged minds, as products of evolution, are not adapted to present media environments. Our minds are adapted to an environment where every image of a person meant a real life encounter with this person. Photographical images, only existing since a little more than a hundred years, are a too recent phenomenon in terms of evolution, to be adapted to. Our brains process photographical images of people as encounters, as if we meet these people in real life. The results is that we treat celebrities as members of our social networks and show interest in their lives for the same reasons we are interested in the lives of our real friends:

“We see them in our bedrooms, we hear their voices when we dine: If this hypothesis is correct, how are we not to perceive them as our kin, our friends, perhaps even our rivals? As a result, we automatically seek information about their physical health, about changes in their relative standing, and above all about their sexual relationships.” (Barkow, 1992: 629-630)

Thus, individuals exposed to media products potentially establish parasocial relationships with celebrities. According to Kanazawa (2002) these ‘imaginary friends’ give us the same friendship satisfaction as real friends do. When parasocial relations are a fact, individuals search for gossip about their parasocial network members for the same reasons as they search for gossip about real (two-way) network members.

1.3 Gossip, unity and social networking

"Telling what's happening in your life and the lives of those you talk to is a grown-up version of telling secrets, the essence of girls' and women's friendships." (Tannen, 1990: 97). According to Tannen (1990) gossip is essential in the construct and maintenance of friendships. She focuses on female friendships, but gossip is bonding in the context of male friendships as well (e.g. Benwell, 2001). This is, because gossip establishes feelings of unity, as Gluckman (1963, 1968) has outlined in his functionalist approach to gossip. Gossip is embedded in our relationships and is used by individuals to better understand other network members and to constitute relationships (Goldsmith & Baxter, 1996). And, since gossip is about people who are known to the gossipers, it creates a sphere of intimacy, people who gossip are connected in a way, and they trust each other (e.g. Ayim, 1994, Bergmann, 1993; Nevo & Nevo, 1993). According to Morreall (1994) part of the bonding effect is due to the fact of shared laughter. People who gossip, giggle, have fun and their laughter has a social function.

Still, in our current societies part of our daily gossip conversations are fading. Locke (1998a, 1998b) explained how feelings of unity with our neighbors are disappearing, because gossip across the fence, to maintain the social bonds with the people surrounding us, does not occur in big cities where one hardly ever meets his neighbors. Szwed (1966) already noticed such changes in the 1950’s, when he was observing a parish in Newfoundland and noticed that men were excluded from important information, spread through gossip, since they started working outside the parish and lost touch with the local gossip circuits. Szwed (1966) attributed an important role to ‘social drinking’ to bridge this knowledge gap. He argued that the gathering of men in pubs for a drink was important to unite these male members of this parish. However, what he did oversee was that these men needed to have mutual contacts to gossip about. In the parish Szwed (1966) observed this might not have been problematic, but in many current western societies, people move to new places regularly. New schools, new jobs, leaving everything behind and moving to another city or country, people are obliged to constantly renew their social network.

Caughey (1984) thinks that our imaginary (or better parasocial) relations with media figures facilitate real life social contacts; media figures can become shared interests to talk about with real social contacts. When people who don’t know each other start a conversation, they try to find mutual interests. Following the argumentation of Caughey (1984), I claim that in our modern western societies, celebrities, being regarded as one-way-members of our social networks, are the mutual acquaintances to establish relationships with new people, we need to integrate with. When individuals shift their life in a new direction, celebrities console by giving friendship satisfaction (Kanazawa, 2002). Celebrities are easy access cards to start gossip conversations with new real encounters. With her research about soap-opera gossip, Riegel (1996) referred to this effect, since she noticed that people who interpersonally gossip about soaps feel bonded. The gossip about the soap gives them a shared interest.

1.4 Sex differences in sociality due to differences in relocation patterns throughout human evolutionary history

“Women who move away from a familiar community often report suffering from lack of gossip in their new environment. They don’t know what there is to talk about, they lack the context of knowledge and acquaintances which situates new information and points its meaning. But they also lack people to talk with.” (Spacks, 1982: 29)

Patrilocality, which means the husband lives with his wife and his relatives, occurs in 70% of our human societies (Burton et al, 1996). Studies tend to confirm that in our evolutionary past women migrated more often than men to his residence after marriage (Oota et al, 2001, Rodseth et al, 1991). Dominance of female migration patterns is not only common among humans, but also occurs in other non-human primates (Gerloff et al, 1999) and would have even impacted the population structure of large mammals (Tiedemann, et al 2000). These facts support the hypothesis that throughout the evolutionary history, our female ancestors were more confronted with adaptive problems due to relocation, such as learning about their new social environment and establishing new social contacts. If this is true, we could suppose that those behavioral adaptations still exist today, and we could expect females to be the socially more active sex. Support for this can also be found in the fact that social contacts in general are more important for woman than for men. By forming allies with others, women secure themselves, and can even achieve power (Yanca & Low, 2004). Hess and Hagen (2002) have already argued that women use gossip as a manipulative strategy to attack the reputation of others in conflicting situations. Their theory on ‘Informational Warfare’ also supports the idea of women being the socially more active sex.

2 Hypotheses

Funded on this theoretical background, I now first outline two different predictions about our interest in Celebrity Gossip. For the first prediction, I fall back on the Learning Hypothesis that explains our interest in Celebrity Gossip. Celebrity Gossip is a source of experiences about the behavioral strategies of prestigious people. Receivers learn from gossipees’ mistakes how to avoid nasty situations, and learn from gossipees’ triumphs how to achieve success. If Celebrity Gossip is so interesting because of the experiences receivers can acquire almost for free, it can be expected that inexperienced receivers show higher interest than more experienced receivers in Celebrity Gossip. More specific, From the Learning Hypothesis I expect that younger people, being less experienced than older people, will have higher interest in celebrity gossip:

Hypothesis 1 (in support of Learning Hypothesis):

Celebrity Gossip appeals to a broad audience because of the fitness-relevant experiences transmitted. Therefore younger receivers, having less experience, will show higher interest than older, more experienced receivers.

Opposed to this prediction in support of the Learning Hypothesis I propose a two more predictions that rather support the Parasocial Hypothesis. The Parasocial Hypothesis predicts two things. First of all, this view on our interest in Celebrity Gossip says that interest in Celebrity Gossip will be higher for those receivers who are more exposed to media products that elicit (pseudo-)encounters with celebrities. Second, this Parasocial Hypothesis forecasts that people with very few real life social network members might seek compensation in celebrities. Celebrities become parasocial network members than can replace real life social network members. From this it can be assumed that individuals, who recently moved to a new social environment, lacking social contacts, will be more interested in Celebrity Gossip than people who have been living in the same social environment for years. Celebrities, as parasocial network members can give friendship satisfaction. Moreover, celebrities can also serve as mutual acquaintances to gossip about with real life social contacts, when we lack real mutual acquaintances. From this I predict the following:

Hypothesis 2a (in support of the Parasocial Hypothesis):

Celebrity Gossip appeals to and audience because of the parasocial interactions some have with celebrities. The more individuals are exposed to media products, the more they encounter celebrities and can establish parasocial bonds with them and, eventually the stronger their interest in Celebrity Gossip will be.

Hypothesis 2b (in support of the Parasocial Hypothesis):

Celebrity Gossip appeals to an audience because of the parasocial interactions some have with celebrities. Therefore the shorter the time a respondent has been living at his/ her current address, the less he/she will have strong real life social bonds and the more chance this individual might seek compensation. Celebrities as parasocial network members fill the social network niches and therefore the stronger these individuals’ interest in Celebrity Gossip will be.

Next, assuming that women are the socially more active sex, because of a more prevalent relocation selection pressure, I predict that women will show higher interest in Celebrity Gossip. This can be supported both by the Learning Hypothesis and the Parasocial Hypothesis. Gossip about strangers might have been adaptive, first to learn about successful and unsuccessful strategies, and second to learn about new social network members and to establish new contacts with these individuals. Therefore:

Hypothesis 3 (Sex Hypothesis):

Female respondents will have a stronger interest in Celebrity Gossip than male respondents.

3 Methodology

An anonymous questionnaire was distributed to 1000 Flemish (Belgian) respondents with the help of 100 graduate students from Ghent University. I used a purposive sampling method to recruit participants. Each student was responsible to distribute 10 surveys during the 2003 two-week Christmas holiday, among people from their social network. They were asked to use quotas to come to an equal male/ female and age distribution. All of our Belgian respondents volunteered to take part in this research, and were not rewarded.

The questionnaire contained 31 celebrity gossip stories, 9 about Belgian celebrities and 22 about non-Belgian celebrities, of which the majority were Americans. All stories were collected from very recent gossip magazines; the internet and some were invented or changed a little to secure to have ‘new’ information. Some examples of stories I presented are:

Out of control! Courtney Love, 39, just can’t shake her rep as Hollywood’s bad seed. She appears to be spiraling out of control (again) after making a mess at a party last week. She was rushed to the emergency room for an apparent drug overdose. Get a grip, Court!

Olivier Martinez, 37, cut a forlorn figure at the airport as he headed back to France alone after a showdown with his girlfriend Kylie Minogue in London.It seemed to be further proof that their two-day reunion had not gone well. Kylie, 35, has been linked to one of the dancers of her latest video clip. "Everyone had heard the rumors about Kylie and we were waiting to see what happened.” Someone from the set told us. "Olivier arrived early and went straight to Kylie's trailer. There were no raised voices, but neither of them looked too happy."

The past couple of months have been hard for Pamela Anderson, 36. She discovered she has hepatitis C, a possible fatal virus that can cause liver failure and liver cancer. “First, my doctor just told me I had a glitch in my blood work, the he told me about the disease. He said that it was going to be what I would eventually die from.”

For all stories, I asked the respondents “How interested are you in this story?” which they had to indicate on a 7-point Likert scale, ranging from “1” “not at all” to “7” “very much”. With this question I measured the respondents’ ‘interest in each presented Celebrity Gossip story’. For each respondent I then averaged all of their 31 separate scores on interest in the presented stories. I recoded this in a new variable: “interest in presented Celebrity Gossip”. This variable ranged from 1 to 7, with “1” “having no interest” and “7” “having very high interest” in presented celebrity gossip.

In a second part of the survey, I asked the respondents questions about their media use. To measure ‘media exposure’ I used several variables with which I measured the respondents’ use of a variety of media products, where encounters with celebrities can be expected. The variables I used are ‘reading gossip magazines, ‘reading newspapers’, ‘watching movies at home’, ‘watching movies in movie theatres’ and ‘watching the news’. These are all media products were celebrities are present. Gossip magazines are the most common product associated with celebrity encounters. News might not commonly be seen as a media product where gossip is present, but as I have argued in chapter 7 a lot of the content of newspapers and television news programs can be classified as gossip.

In a third and last series of questions I asked the respondents some personal questions, such as their sex and age. I asked some questions about their real life social contacts and some socio-demographical information. Concerning this last, I asked them where they grew up, where they were living at the time of taking the survey, and most important for this study, I asked them “How long have you been living at your current address?”. This last question gives information about any recent relocation pattern of my respondents. The time my respondents had been living at their current address ranged from 0.5 to 82 years.

From the returned questionnaires 838 were complete and included in the survey (83.3%): 397 men and 441 women. Ages ranged from 10 to 82, and did not differ between the sexes (using Independent Samples T-test for Equality of Means): M_age=34.9 +/- S.D.=13.6 years for women, M_age=34.2 +/- S.D.=15.9 for men (df= 834; t= -.64; p= .52).

4 Results

To test my first two, contrasting, hypotheses, I used a linear regression to measure the impact of “age”, “media exposure” and “the time respondents lived at their current address” on the respondents’ “interest in the presented Celebrity Gossip”.

I forecasted that respondents with higher media exposure would show greater interest in Celebrity Gossip. The Parasocial Hypothesis predicts this because the more we encounter celebrities the more likely we will regard them as parasocial, or one-way, members of our social network. Media exposure enables frequent encounters with celebrities, so that more media exposure can increase the number of celebrities respondents regard as one-way members. This will in the end increase the interest in Celebrity Gossip.

Although it could be argued that the relation between media exposure and interest in Celebrity Gossip is rather a bidirectional correlation than a causal relation, I did use Regression Analysis to look at the effect of different kinds of media exposure on interest in Celebrity Gossip. Reason for this is that my theoretical framework, and more specific the Parasocial Hypothesis, does predict a causal relation, with media exposure influencing Celebrity Gossip interest. I do admit there is a feedback loop, causing a correlation relation between both variables, as higher interest in Celebrity Gossip will cause respondents to read more tabloids, watch more soaps, watch more movies and so on, because they will seek information about their parasocial network members. Both phenomena influence each other, but still the Parasocial Hypothesis predicts that encounters (media exposure) precede interest in Celebrity Gossip. I do add that once interest in Celebrity Gossip exists in an individual he or she will increase his or her media exposure, and then it is hard to say which influences which most.

Looking at the results of the effect of a variety of variables on the respondents’ interest in the presented Celebrity Gossip (see table 6.1) I notice that age seems to be the one and only most important and significant factor (β_age= -.53; p<.001). Age has a negative impact on the interest in Celebrity Gossip, meaning that the younger my respondents, the higher their interest in Celebrity Gossip. This results supports the Learning Hypothesis.

Table 6.1. The effect of “Age” and “Time living at current address” on interest in “presented celebrity gossip”, using linear regression.

Dependent Variable: Interest in Presented Celebrity Gossip (scoring from 1.00 to 7.00).

From all media exposure variables, only two indicate to have some effect. Reading gossip magazines slightly affects the interest in the presented Celebrity Gossip (β_{gossip-magazines}= .18; p<.001). However, I comment that the effect is low and especially this media exposure variable has a bidirectional relation rather than a single causal relation with interest in Celebrity Gossip. Next to this, watching movies at home somehow influences the interest in the presented Celebrity Gossip, though the effect is very small (_{βmovies-home}= .08; p<.05). All other media exposure variables have no significant influence on the interest in the presented Celebrity Gossip. Overall these results cannot confirm my Parasocial Hypothesis.

For the second part of testing the Parasocial Hypothesis, the time respondents are living at their current address has a very low, though significant (p<.05) effect on the interest in the presented- (β_relocation= .10) Celebrity Gossip. However, I had expected a negative correlation, meaning that the shorter the respondents were living at their current address, the higher their interest in Celebrity Gossip. At least, that is what I predicted with my Parasocial Hypothesis. Nevertheless, looking at the results (see table 6.1), the relation is positive, indicating the opposite effect.

I added some extra control variables, such as the number of friends the respondents have, how satisfied they are about their real life friendships and with how many people they are living with, to seek further support for the Parasocial Hypothesis. These variables all correlate negative, indicating that, as expected by the Parasocial Hypothesis, the fewer friends and roommates respondents have, and the less satisfied they are about their real life friendships, the higher their interest in the presented Celebrity Gossip is. Still, all effects are extremely low and non-significant. Overall hardly any results support my Parasocial Hypothesis.

When presenting the effect of age on the interest in the presented Celebrity Gossip graphically (see graph 6.1) this results in regression lines with a negative slope. Younger respondents show some interest in the presented Celebrity Gossip. As the age of my respondents increases, which is when you move to the right side of the graph, it shows that the interest in presented celebrity gossip declines.

Graph 6.1. The effect of age on interest in celebrity gossip

For the third, Sex Hypothesis, which focuses on the sex differences in interest in celebrity gossip, I can conclude that women show a little more interest than men do. Although both mean scores are rather low, female respondents significantly (p<.01) scored higher on ‘interest in presented Celebrity Gossip’ than male respondents (M_female=2.29 vs. M_male=2.05).

I have argued that this sex difference is due to different selection pressures for men and women, with women being more exposed to problems of relocation. Since my data does not support the Parasocial Hypothesis, but does support the Learning Hypothesis, I only looked at the influence of sex differences for the latter one. My Learning Hypothesis claims that Celebrity Gossip appeals to an audience because of fitness-relevant experience-information being transmitted. In our evolutionary past, female ancestors might have benefited more from this adaptive value of gossip than male ancestors, because they migrated more often to new social environments. If gossip as a learning device was, and still is, more valuable to women than to men to get fitness-relevant experience-information, then the effect of age on the interest in celebrity gossip should be greater for female respondents than for male respondents. I already controlled for the effect of the sex of our respondents in my first analyses, noticed that this variable as a low (β_sex= .12), though significant effect (p<.001) on this interest in presented Celebrity Gossip (see table 6.1).

Graphical results (see graph 6.2) show that the slope is greater for female respondents than for male respondents. At younger age, girls show more interest in the presented celebrity gossip. As age increases, this is when you move to the right side of the graph, you notice that the difference in interest between men and women disappears. Both men and women of older age, when having accumulated a lot of experiences themselves and through gossip, show (almost) no interest in celebrity gossip.

Graph 2 Sex differences in the impact of age on interest celebrity gossip

Looking more into detail how strong and how significant the sex difference is in the effect of age on the interest in celebrity gossip, I again used a Linear Regression, splitting the data file to compare between male and female respondents. As follows from table 6.2, a stronger β is found for female than for male respondents for interest in the presented Celebrity Gossip (β_female= -.48; β_male= -.38). Using a significance test to control for differences between correlation coefficients (using Statistica Other Significant Tests for Difference Between Two Betas) this sex differences can be considered to be significant (p<.05).

Table 6.2. Sex differences in the effect of age on interest in celebrity gossip”

5 Conclusion

At some point in life almost everyone engages once or more in acquiring Celebrity Gossip through media channels or exchanges Celebrity Gossip in interpersonal interactions. In this paper I presented two alternative explanations for why we like Celebrity Gossip. To test these I questioned 838 Flemish (Belgian) respondents about their interest in Celebrity Gossip, their media use, their real life social network and their recent relocation pattern.

My data show first of all that the use of media does hardly influence my respondents’ interest in the presented Celebrity Gossip. The only effects I could find come from reading gossip magazines and watching movies at home. The effects of both variables are low, however. Secondly, relocation patterns, indicating that the respondent moved to a new social environment, do not affect the interest in Celebrity Gossip either. Both these effects of media use and relocation patterns were expected from the Parasocial Hypothesis.

An alternative hypothesis that explains our interest in Celebrity Gossip is the Learning Hypothesis, which says that gossip about celebrities is liked because it transmits fitness-relevant experience-information. From celebrities’ behavior, receivers learn about successful and unsuccessful strategies to solve problems. In this perspective Celebrity Gossip is most useful to less experienced individuals. My data support this Learning Hypothesis, in the sense that younger people, who are less experienced, are significantly more interested in Celebrity Gossip than older, more experienced people.

Because of sex differences in relocation patterns in our evolutionary history, I argue that women more than men should be better skilled to use Celebrity Gossip and have a greater tendency to acquire Celebrity Gossip. My data indeed presents stronger correlations between age and interest in Celebrity Gossip for female than for male respondents.

6 Discussion

In our current western societies people relocate often to new social environments. When meeting new people, a way to establish relationships is by talking about mutual acquaintances. However, when entering an entirely new social environment, mutual acquaintances to gossip about are hard to find. Celebrities, as parasocial members of our social network can bridge this gap. Gossip about celebrities is an easily accessible method to establish relationships with new people we encounter.

In the Western world we nowadays live in societies with growing individualism and loss of small communities. It is no longer a fact that everyone knows everyone else. People no longer establish and maintain friends for life. They switch places to live and therefore switch social networks and friends on a regular basis. Celebrities are somehow a stable factor in this all, securing mutual acquaintances to talk about.

I had expected that individuals who recently moved to a new social environment would have a greater interest in Celebrity Gossip. My data did not support this idea, but I suggest it might be interesting not only to look at the time respondents lived at their current address, but to keep track of their relocation pattern in general. People who often switch social environments might, according to the Parasocial Hypothesis, be more interested in Celebrity Gossip, than people with a more stable social environment. I must also admit that relocations of my respondents might often not be very geographically differential. Belgium is a small country, and this study involved only Flemish respondents, who most often relocated only to a neighboring town. I therefore suggest to further test this Parasocial Hypothesis with respondents who relocate on a larger geographical basis, such as people moving to a new country.