“Is my picture good enough?”
Testing an inoculation message as intervention for the
effect of social comparison with different Instagram
sources on self-esteem
______________________
Master’s thesis
Graduate School of Communication
Master’s programme Communication Science
(Persuasive Communication)
University of Amsterdam
Brenda Montijn (10776184)
Supervisor: Dr. C. Scholz
June 29, 2018
Word count: 7496
2
Abstract
On a daily basis, female Instagram users are confronted with many perfect-looking photos. Instagram has a wide range of users, such as influencers and unknown peers. It is expected that such photos stimulate females to compare themselves with the persons in the photos. Past research examined the effects of social media use on social comparison, and the effect of social comparison on mostly appearance related variables, while self-esteem remained relatively unresearched. Unknown peers and influencers are two interesting groups to look at because social media users do not personally know these people. Additionally, unknown peers are more similar to normal social media users and influencers seem to be better off. However, unknown peers and influencers have barely been researched. This experiment examined (1) whether unknown peers’ photos had a stronger positive effect on social comparison than influencers’ photos, (2) whether social comparison negatively influenced self-esteem, (3) and whether an inoculation message reduced the effect of social comparison on self-esteem. Results showed that influencers induced more upward social comparison and that social comparison induced a more negative self-esteem. Inoculation messages did not proof to be effective. The study suggests that more future research is needed on self-esteem and effective intervention strategies.
Introduction
When you go through the content of the social networking site Instagram, you often come across photos of people in which they seem to have a great life and a perfect appearance. These photos do not form a realistic or complete reflection of real life. Instagram users may feel bad about themselves because they compare themselves to the persons in these photos. As Instagram emerged, a new group of users that seems to have an even greater life arose: the
influencers. They are followed by many people and show off their great and often luxury lives
in pretty photos.
Social comparison is related to severe consequences. Studies on social media have found that it is related to depressive symptoms, loneliness and low self-esteem (e.g. Nesi and Prinstein, 2015; Yang, 2016; Cramer, Song & Drent, 2016). Females might even experience stronger negative effects. A recent study found that females were more active than males in taking and posting selfies and using photographic filters on photos (Dhir, Pallesen, Torsheim & Andreassen, 2016). Young adults were also more likely to do these things in comparison to
3 adults. Female young adults thus seem to be busy with their appearance and are faced with many selfies of other women to compare with. Several studies have shown that women are indeed likely to compare themselves on social media and attain a lower self-esteem, more body dissatisfaction, more drive for thinness and a negative mood (e.g. Fardouly & Vartanian, 2015; Hendrickse, Arpan, Clayton & Ridgway, 2017; Ho, Lee & Liao, 2016; Brown &
Tiggemann, 2016). A study on user profiles found that females were more likely than men to look at other people's online profiles on social network StudiVZ to compare with others (Haferkamp, Eimler, Papadakis & Kruck, 2012).
This study will look at how the effect of social comparison on self-esteem can be reduced. For this a change in the cognitions about oneself and about photos is needed. In earlier research, for instance on smoking behavior (Hersey et al., 2005), researchers exposed participants to an inoculation message. Such an inoculation message should protect media users from persuasive and attitude-threatening messages in the media with the help of counterarguments (McGuire, 1961). People are made resistant to potential messages in the media that threaten their current beliefs (or in this case, Instagram photos that threaten self-esteem). When a young woman thinks she looks good today but then sees a photo of a photoshopped person, she might start to compare herself to this person and think her own appearance is not good enough anymore. An inoculation message might prevent negative effects caused by social comparison by sending a “forewarning”. This could be a message that emphasizes that many Instagram photos are photoshopped. When a female is later exposed to a photo, she remembers that photos can be photoshopped and do not reflect reality.
To conclude, studies did find psychosocial effects of social media content. Although there is a large amount of studies conducted in the field, there are five gaps in the literature. First, a remaining challenge is to examine how effects of social comparison on self-esteem can be reduced. To make the translation from theory to practice, this experiment will test an inoculation message as a mode to reduce negative effects on self-esteem. Second, self-esteem has barely been researched within the context of social media and social comparison.
However, self-esteem has been shown to be related to social comparison on social media (e.g. Wang, Wang, Gaskin & Hawk, 2017) and is relevant to explore in a social media context because it is a broader concept than body dissatisfaction. It also measures satisfaction with your social skills, complete appearance and performance (Heatherton & Polivy, 1991). Third, most studies have focused on celebrities, friends, family and (distant) peers, but not on
4 the viewer may know but does not regularly socialize with in person” (Fardouly & Vartanian, 2015, p. 83). Both influencers and unknown peers are people that social media users do not personally know. However, it is expected that influencers are more familiar to social media users than unknown peers, because they are well-known people. Additionally, influencers have something extra, such as success or a good appearance, while unknown peers are expected to be more similar to the normal social media user. The reason that some females follow unknown people, is because they are found via “hashtags” used with photos or because people have profiles with open access. An online survey among young adults showed that the amount of strangers followed moderated the relationship between Instagram use and social comparison (Lup, Trub & Rosenthal, 2015). This suggests that comparison with unknown peers does happen. Fourth, regarding the measurements, research has applied trait
measurements more than state measurements. It is unlikely that a onetime exposure to a photo would change your overall self-esteem or tendency to compare with others. Fifth, most of the studies that have been done in this field were surveys. This makes it difficult to examine whether social comparison really influences self-esteem, or that it is that females with a lower self-esteem are more likely to compare with others on social media.
To overcome some of these shortcomings in the existing literature, an online
experiment was conducted among female young adults. This study looked at the effects that unknown peers’ and influencers’ photos on Instagram have on the state of upward social comparison and in turn on the state of self-esteem. Additionally, inoculation theory was applied to examine whether it is a helpful strategy in reducing the possible negative effect of upward comparison on self-esteem.
Influencers on Instagram
Instagram is a social network where a diversity of people (e.g. normal users, influencers and celebrities) share photos and videos. “Normal” Instagram users have a personal profile to share their photos and videos with friends and other people they allow to see their content. Many celebrities are also on Instagram to share their lives and music. However, a new and even more flawless group of celebrities emerged together with the popularity of social media: influencers. There is thus far no steady definition on what an influencer is. This might be due to the different amounts of followers they have. Chae (2018) defines influencers as “online celebrities who exhibit their personal lives to many followers via social media” (2018, p. 246). Influencers or micro-celebrities have in common that they show their personal lives and lifestyles to a large amount of followers on their social media profile, while they follow a
5 small amount of people themselves (Abidin, 2016). Some influencers have knowledge about specific domains such as beauty, fashion or travel influencers (Chae, 2018). Other influencers “tend to be conventionally good-looking, work in “cool” industries such as modeling or tattoo artistry, and emulate the tropes and symbols of traditional celebrity culture, such as glamorous self-portraits, designer goods, or luxury cars” (Marwick, 2015, p. 139). It has been said that the posts of influencers are “catalogs of what many young people dream of having and the lifestyle they dream of living” (p. 155).
Exposure to Instagram photos and social comparison
Because influencers are living the dreams that some other social media users have, their photos can stimulate social comparison, which is “the process of developing subjective assessments of one’s opinion and ability by making comparisons to other persons” (Festinger, 1954, in Cramer, Song & Drent, 2016, p. 739). The social comparison theory by Festinger (1954) argues that there exists a drive in humans to evaluate the own abilities and opinions. These evaluations are often based on comparisons with other persons. As a result of the extra things that influencers have (e.g. clothes, fit bodies, exotic travels), “normal” social media users might think that they are not good enough. They could question their own appearance or performances in life and gain a lower self-esteem at that moment. By making comparisons to influencers, these social media users might also seek improvement of themselves. This is called upward social comparison (Cramer, Song & Drent, 2016; Yang, 2016).
Even though influencers might cause social comparison, the effect on social comparison is expected to be larger for unknown peers´ photos, in line with the social
comparison theory (Festinger, 1954). Festinger’s theory describes that “a person does not tend to evaluate his opinions or his abilities by comparison with others who are too divergent from himself” (p. 120). Since peers are more likeable to have abilities or appearances close to other “normal” social media users, it is easier to evaluate the own ability and appearance by
comparing with this group. According to the theory one would say that it is not possible for a normal social media user to accurately evaluate the own abilities and appearance in
comparison to the influencer, because the influencer is too different.
Many recent studies have been conducted on social media use and social comparison and they show that the two are indeed related (e.g. Wang, Wang, Gaskin & Hawk, 2017; Chae, 2018; Fardouly & Vartanian, 2015; Brown & Tiggemann, 2016). However, most of these studies are surveys and cannot make causal assumptions. A survey among young adults
6 showed that passive social media use was related to more upward social comparison with friends (Wang, Wang, Gaskin & Hawk, 2017). A longitudinal panel survey found that female young adults’ exposure to a diversity of influencers’ social media (including Instagram) relates to comparison of one’s life with that of influencers (Chae, 2018). Two studies found that females compared themselves more with peers (Fardouly & Vartanian, 2015) and friends (Fardouly & Vartanian, 2015; Ho, Lee and Liao, 2016) than with celebrities. Contrary to these findings, an experiment on Instagram use and appearance-related comparisons among female young adults found that the extent of comparisons did not differ between peers’ and
celebrities’ photos (Brown & Tiggemann, 2016). In short, the following is proposed (Figure 1):
H1: Female young adults’ exposure to unknown peers’ Instagram photos is associated with more upward social comparison, in comparison to when someone is exposed to an
influencers’ photo.
Exposure to Instagram photos and self-esteem
If a female upwardly compares herself with another, one can argue that her self-esteem will be lower for that moment because she seeks improvement of herself and evaluates her own downsides. Self-esteem “reflects an overall subjective evaluation of personal worth” (Alessandri, Vecchione, Eisenberg, Laguna & Reynolds, 2015, p. 621). It is plausible that social comparison influences self-esteem because concepts such as appearance and success are not only related to social comparison (Allan & Gilbert, 1995) but also to self-esteem (Heatherton & Polivy, 1991).
However, there is a lack of studies that examine the influence of social comparison on self-esteem. An online survey concerning social networking sites WeChat and Qzone studied these concepts and found a mediation effect for social comparison on the relationship between passive social media usage and self-esteem (Wang, Wang, Gaskin & Hawk, 2017). More studies have been conducted on body dissatisfaction and a drive to be thin. These concepts are partly measured in self-esteem scales with questions as “I am dissatisfied with my weight” (Heatherton & Polivy, 1991). A recent survey among female young adults found that appearance-related comparisons on Instagram were associated with greater body dissatisfaction. This study also found a mediation effect for Instagram photo activity, appearance-related comparisons and drive for thinness (Hendrickse, Arpan, Clayton & Ridgway, 2017). Specific studies on celebrities found that young adult (Fardouly &
7 Vartanian, 2015) and adolescent (Ho, Lee & Liao, 2016) females’ social comparison on social media is associated with more body image dissatisfaction, body image concerns and a drive to be thin. For male adolescents, a study showed that there was no significant relationship between social comparison with celebrities and body dissatisfaction or a drive to be muscular (Ho, Lee & Liao, 2016).
It is thus plausible that good-looking or interesting photos have the potential to impair females’ self-image. Social comparison has been shown to mediate the relationship between social media use and self-esteem (related variables) for both influencers and unknown peers. Therefore, the current study proposes that (Figure 1):
H2: upward social comparison mediates the relationship between female young adults’ exposure to unknown peers’ and influencers’ Instagram photos and self-esteem, such that more upward social comparison results in a lower self-esteem.
Inoculation messages
To reduce negative consequences of social comparison, female young adults should become resistant towards photos that seem perfect. Inoculation messages can do this and “involve pre-exposing the person to the mention of counterarguments against his beliefs together with a detailed refutation”(McGuire, 1961, p. 184). First of all, inoculation messages consist of the mentioning of counterarguments against the existing beliefs. These counterarguments threaten these beliefs to make people alert that they need to substantiate their beliefs for when they are faced with a real attitude-threatening message (McGuire, 1964, in Szybillo & Heslin, 1973). This stimulates people to find new support for their position to make their beliefs stronger. The second element is a detailed refutation of the earlier mentioned counterarguments. Because of the mentioning of conflicting arguments, people are expected to better pick up the information in the refutation part. Only emphasizing the beliefs that people already hold should not induce enough resistance to attacks because it would state the obvious.
Inoculation messages have been studied in the field of health communication (e.g. Parker, Ivanov & Compton, 2012; Niederdeppe, Heley & Barry, 2015). They have proven to be successful in making people resistant to attitude threatening messages for a diversity of health topics. In an experiment on sexual risk behavior, young adults who received an inoculation message were better able to develop counterarguments concerning unprotected sexual activity, and had more negative attitudes towards the attack message on condom use (Parker, Ivanov & Compton, 2012). Another experiment on several anti-policy messages
8 (concerning sugary drinks and obesity, cigarette smoking, and prescription painkiller abuse) showed that inoculation pro-policy messages were successful in stimulating counterarguing against the anti-policy messages (Niederdeppe, Heley & Barry, 2015).
Inoculation messages as an intervention for Instagram photos
Social media are known for positivity bias. Lin and Utz (2015) found that 72.6% of the content in the Facebook news feed was perceived as positive by the respondents. The same bias can be observed on Instagram. Users have the tendency to share mostly positive photos. For instance, many photos are photoshopped. Especially influencers are good at
photoshopping their pictures and they often do so (Abidin, 2016). They make adjustments such as editing away blemishes or smoothing wrinkles. Instagram also has a build-in function to add “filters” (i.e. color and brightness layers) on your photos, but there are also many external smartphone apps providing editing functions. These make it easier and more accessible for normal social media users to edit their photos.
Inoculation theory has not yet been studied in the context of social media and social comparison. Even though young adults might be aware that photos can be photoshopped or that they are just a snapshot of someone’s life, they might forget this once they get exposed to a perfect looking photo. An inoculation message can remind them of these facts. In an
inoculation message one uses a “forewarning” (Niederdeppe, Heley & Barry, 2015), which in this case means warning young adults that they will be exposed to photos that can change their beliefs about themselves.
To give an example: when a young woman believes she looks good and then sees a photo of a photoshopped influencer, she might start to compare. Her self-esteem can be threatened at that moment, because she thinks that her own appearance is not good enough anymore. An inoculation message has the potential to weaken the effect of this negative comparison on self-esteem. Such a message would start with a threatening part: “influencers always seem to look extra good”. Next, a refutation part is included, for instance: “but their photos are often photoshopped”. This helps the woman to counterargue her own negative comparisons for when she is exposed to an influencers’ photo.
9
H3: The relationship between upward social comparison and self-esteem is weaker when
someone is also exposed to an inoculation message, in comparison to when someone is not exposed to an inoculation message.
Figure 1. Conceptual model
Conceptual model with the effect of exposure to different sources on self-esteem
Methods Participants and procedure
To reach the largest amount of participants within a short period, this study made use of a convenience and snowball sample which consisted of 104 female young adults between 18 and 25 years old (M = 22.22, SD = 1.82) who use Instagram. See Appendix 3 for the descriptive statistics of the separate conditions. The experiment was distributed via social media (Facebook, Instagram and Whatsapp) and shared via social media by contacts of the researcher.
The questionnaire first asked questions to select whether the participants were appropriate for the experiment. They needed to be female, between 18 and 25 years old and use Instagram on a weekly basis. Participants were randomly assigned to one of the conditions in which they read or did not read the inoculation message and next they saw an influencers’ or unknown peers’ photo. Hereafter they answered questions for state upward comparison and self-esteem. Next, they answered questions to check whether the inoculation message and photo were perceived as intended and a question to check whether they paid attention to the photo. Finally, questions about trait upward comparison and self-esteem were asked.
10
Design
An online experiment was conducted with a 2 (exposure to inoculation message: message vs. no message) x 2 (exposure to different sources: influencers´ photo vs. unknown peers´ photo) full-factorial between-subjects design.
Measures
Independent variables
Inoculation message conditions. Participants read an inoculation message consisting of a
threatening and a refutation part (Appendix 1). To stimulate participants to pay attention to the text, they were able to click through after 20 seconds. The threatening part consisted of arguments as “sometimes it can appear like Instagram users are continuously traveling to pretty countries.”. The refutation part consisted of arguments as “photos only give a limited view into the daily life of the Instagram user”.
Exposure to different sources. Participants were exposed to either one of the three
photos of a female influencer or one of the three unknown peers’ photos (Appendix 1). Three different persons were used for both conditions, to exclude the chance that effects would be due to one specific photo. The influencers are selected because they are well-known in the Netherlands, they have many followers and gain much publicity. In this way they are made distinct from the unknown peers. The Instagram users that are chosen as a peer because they are all female, they are about the same age as the participants and are all Dutch. These unknown peers have a smaller number of followers, which makes the chance lower that the Dutch participants will actually know the peer. However, the two sources were made as similar as possible to make them comparable to each other, except for the fact that one is an influencer and one a peer. Both groups were selected based on the tropical countries with a sea, their pretty clothes and photo quality.
The influencers have pretty branded clothes or accessories where a luxury brand logo is visible. The participants in the unknown peer condition were exposed to one of the three photos of a regular Dutch person who is not an influencer. Again, the person is in a tropical area nearby the sea, but this time the country is within Europe and thus less expensive to travel to. The clothes are pretty, but not branded, in comparison to the influencers who often get branded clothes for free or have more money to buy them. Participants were asked to look at the photo and they were able to click through after ten seconds. A time span longer than ten
11 seconds would not be realistic, because Instagram is a network in which people rapidly
“swipe” through the content.
Dependent variables
State self-esteem. The 20-item scale by Heatherton & Polivy (1991) was translated to Dutch to
measure one’s feelings about oneself. Examples of items are “I feel satisfied with the way my body looks right now”, “I feel inferior to others at this moment” and “I feel confident about my abilities” (Appendix 2). The items are rated on a 5-point Likert-type scale (1 = not at all, 2 = a little bit, 3 = somewhat, 4 = very much, 5 = extremely). Four items of the performance factor were left out, because they are irrelevant to the experiment or too general for a state measurement. These items might be too difficult or vague for the participants to give an answer to.
A principal component analysis revealed four factors with an eigenvalue greater than 1.00 which accounted for 67.3%. Ten items were reversed as was done by Heatherton and Polivy (1991). Items with a factor loading above 0.4 were included (Field, 2005). The item “I feel self-conscious” was removed due to an incorrect translation in Dutch. The item “I feel good about myself” was replaced to the performance factor where the item had a loading of 0.46, as done by Heatherton and Polivy (1991). The item “I feel that others respect and admire me” had low loadings on all three remaining factors and was therefore removed. Three
reliable averaged scales for state self-esteem were created in line with the original scales, namely social (α = 0.87, N = 6 items), appearance (α = 0.77, N = 4 items), and performance (α = 0.72, N = 4 items) state self-esteem. Another averaged scale for overall state self-esteem was computed (α = 0.88, N = 15 items).
State upward social comparison. The measurement of Vogel, Rose, Roberts & Eckles,
2014) was translated to Dutch and used to measure the extent of state upward social
comparison (Appendix 2). The question for Downward social comparison has been left out. The measurement was adjusted to the Instagram environment and to a state measurement. This means that the question was adjusted to the comparison during the specific exposure to the Instagram photo in the experiment. The question was answered on a 5-point Likert-type scale ranging from “not at all” to “a great deal”. The assumption of normality was met for state upward social comparison.
12
Inoculation message. To measure whether the inoculation message was perceived as intended,
participants were asked to rate two statements on a 7 point scale (1- not at all to 7- very much). The statements were “the pictures on Instagram do not represent reality” and “by seeing Instagram photos of others, my self-confidence can be reduced”.
Exposure to different sources. To check whether the peer was actually unknown and
whether the influencer was more familiar than the peer, participants were asked: “did you already know the person in the photo before this study?” Next to this, to measure whether participants thought that the unknown peer was more similar to them than the influencer, they rated the item “the person in the photo is similar to me” on a 7 point scale (1- not at all to 7- very much).
Control variables
Trait self-esteem. To measure trait self-esteem, the Rosenberg General Self-esteem Scale was
translated to Dutch (Rosenberg, 1965, in Fetzer Institute, n.d.). This scale has recently been applied in a study on exposure to influencers’ social media and envy (Chae, 2018). It consists of 10 items about general feelings about oneself. Items included for example “On the whole, I am satisfied with myself” and “I take a positive attitude toward myself” (Appendix 2). The items are rated on a scale that ranges from 1- strongly disagree to 5- strongly agree
(Alessandri, Vecchione, Eisenberg, Laguna & Reynolds, 2015). Five items were reversed (Fetzer Institute, n.d.). A reliable averaged scale with all the 10 items was computed for trait self-esteem (α = 0,90). The assumption of normality was met for trait self-esteem.
Trait upward social comparison. The trait measurement for upward social comparison
created by Vogel, Rose, Roberts and Eckles (2014) will again be used here to measure trait upward social comparison (Appendix 2). Again, solely the question for upward social comparison is used. The measurement here is only slightly adjusted to Instagram use. It is a more general measurement than the state measurement because it asks about Instagram comparisons in general and not just the comparisons within the experiment. The assumption of normality was met for the variable trait upward social comparison.
Attention to Instagram photos. To measure whether participants paid attention and
really looked at the Instagram photo, the following question was asked: “what was the color of the top or dress of the person in the picture?” (Appendix 2). Participants could answer the question with either blue, white, orange, black or “I do not know”.
13
Analyses
Before the hypotheses were tested, surveys taken in preview mode and incomplete surveys were removed from the dataset. Dummy variables were created for the source (0 = influencer, 1 = peer) and inoculation conditions (0 = no message, 1 = message). Crosstab analyses were conducted for knowing the source and recall of the color of the top. To control for perceived similarity with the sources, an Independent-Samples T test was conducted with source condition as grouping variable. Then, two other T tests were performed, one with inoculation condition as grouping variable and perceived reality as dependent variable, and the other with self-confidence as dependent variable. Two more T tests were performed to do a
randomization check for age, with source condition and inoculation condition as grouping variables.
To test Hypothesis 1, a multiple regression analysis was conducted with state upward comparison as dependent variable. Photo condition was the independent variable in step 1 and trait upward comparison was added in step 2. To test Hypothesis 2, a mediation (regression) analysis was conducted using the Process Macro for SPSS. Source was added as independent variable, state self-esteem as dependent variable and state upward comparison as mediator. To test Hypothesis 3, a moderation analysis was performed using Process. State upward
comparison was added as independent variable, overall self-esteem as dependent variable and inoculation message as moderator variable. To control for the effects of trait upward
comparison and trait self-esteem, the regression analyses were repeated with these two variables as covariates. Extra exploratory analyses were performed for the state self-esteem subscales if only partial support was found for the mediation or moderation effects.
Results Manipulation checks and control variables
Familiarity and similarity. First, I conducted a manipulation check, assessing whether the
influencers and unknown peers were perceived differently by the participants. Specifically, the influencers were expected to be more familiar to the participants than the unknown peers and unknown peers were expected to be perceived as more similar than influencers.
Additionally, it was expected that none of the participants knew the peer. The biggest part of the participants did know the influencers (74.5%). Influencer Laura Ponticorvo was known by 64.7%, Anna Nooshin by 77.8%, and Monica Geuze by 80.0%. As expected, none of the
14 participants knew any of the peers. Furthermore, participants indeed felt more similar to unknown peers (M = 3.84, SD = 1.82) than to influencers (M = 2.44, SD = 1.39), t (102) = -4.45, p < 0,01, CI = [-2.03, -0.78]. In short, the manipulation for source worked as expected.
Inoculation message. Next, I examined whether participants in the inoculation
message were more likely to think that Instagram photos did not present reality and that exposure to Instagram photos can threaten their self-confidence, in comparison to participants who did not read the inoculation message. There were no significant differences in the
perceptions of reality between the non-inoculation (M = 4.83, SD = 1.55) and inoculation message (M = 4.74, SD = 1.67) groups, t (102) = 0.27, p = 0.792, CI = [-0.55, 0.72], or in the perception that Instagram photos can threaten one’s self-confidence (non-inoculation
message: M = 4.43, SD = 1.79, inoculation message: M = 4.43, SD = 1.96), t (102) = 0.01, p = 0.992, CI = [-0.73, 0.74]). Both groups scored high on a 7 point scale, thinking that
Instagram photos do not reflect reality and can threathen one’s self-confidence. To conclude, the inoculation manipulation did not work as expected.
Attention. Participants were asked to recall the color of the top or dress of the person
in the photo to allow for an attention check. Most of the participants in the influencer (87,3%) and unknown peer condition (93.9%) passed this test. The analyses will be conducted both including and excluding those individuals who failed the attention test to assess the robustness of the results.
Randomization check
It is possible that some participants were older or younger than the person in the photo, leading to different levels of social comparison, or that participants of different ages react differently to the inoculation manipulation. There was no difference in the means between the influencer (M = 22.04, SD = 1.85) and unknown peer (M = 22.43, SD = 1.78) groups, t (102) = -1.10, p = 0.274 , CI = [-1.10, 0.32], or between the non-inoculation (M = 22.22, SD = 1.89) and inoculation (M = 22.22, SD = 1.78) message groups, t (102) = -0.02, p = 0.985, CI = [-0.72, 0.71]. Age was thus equal over all these conditions and does not form a confounding factor.
Exposure to different sources and state upward social comparison
Hypothesis 1 predicted that participants in the unknown peer condition would engage in more state upward comparison than participants in the influencer condition. The regression model
15 with source as independent variable and without the covariates significantly predicted state upward comparison, F (1, 102) = 6,45, R2 = 0.06, p < 0.05. Source, b = 0.53, β = 0.24 t = -2.54, p < 0.05, 95% CI [-94, -0.12] had a strong negative and significant association with state upward comparison. Thus, participants who saw a photo of an influencer had a stronger tendency to compare themselves with this person than participants who saw the unknown peer picture.
When adding trait upward comparison as a covariate, this effect became of moderate strength but remained significant and the regression model was improved, Fchange = 36.32, R2change = 0.25, p < 0.01, F (2, 101) = 22.50, p < 0.001. The two variables explained 30.8% of the variance in state upward comparison. The effect of trait upward comparison on state upward comparison was moderately strong and positive, b = 0.47, β = 0.50, t = 6.03, p < 0,01, 95% CI [0.31, 0.62]. This indicates that young adult females who have a general tendency to compare themselves with others also compared themselves more with the persons in the pictures. Because participants in the influencer condition tended to compare themselves more with these persons, Hypothesis 1 had to be rejected.
State upward social comparison
Hypothesis 2 argued that state upward comparison would mediate the relationship between source and state self-esteem. The explained variance of the mediation model for overall state self-esteem (i.e. scale with all the items) was significant but low, R2 = 0.06, F (1, 102) = 6.45, p < 0.05. State upward comparison had a significant but weak effect on overall state self-esteem, b= -0.24, t = -4.40, p < 0.001. The more one compared herself to the person in the photo, the lower the state self-esteem was. This effect held even when controlling for trait upward comparison and trait self-esteem, but became weaker, b = -0.11, t = -3.03, p < 0.01. Neither source or trait upward comparison had a significant direct effect on overall state self-esteem. Trait self-esteem, however, had a strong and significant effect, b = 0.88, t = 13.93, p < 0.001 on overall state self-esteem. Thus, participants who already had a lower self-esteem previous to the experiment also seemed to have a lower state self-esteem during the experiment. Source had an indirect effect on overall self-esteem which was significantly greater than zero, b = 0.13, 95% CI [0.03, 0.26]. The effect remained significant after controlling for the control variables, but it diminished, b = 0.05, 95% CI [0.01, 0.11]. Hypothesis 2 was thus supported.
16 Hypothesis 3 predicted that an inoculation message would moderate the relationship between state upward comparison and state self-esteem. The explained variance of the moderation model for overall state self-esteem without the covariates was significant, R2 = 0.19, F (3, 100) = 7.75, p < 0.001 (Appendix 3). As outlined above, state upward comparison had a negative and weak, marginally significant effect on overall self-esteem, b= -0.14, t = -1.80, p < 0.1. As participants compared themselves more to the person in the picture, their state esteem diminished. This effect held when controlling for trait social comparison and trait self-esteem. Inoculation condition did not have a significant main or interaction effect with state upward comparison on overall state self-esteem (Appendix 3). Trait self-esteem, however, had a significant and strongly positive effect on overall state self-esteem, b= 0.90, t = 13.56, p < 0.001, but trait upward comparison did not. Because there was no significant interaction effect, nor a direct effect of inoculation condition on overall state self-esteem, Hypothesis 3 would not be supported.
Because Hypothesis 3 was not supported, additional exploratory analyses were performed for the subscales of state self-esteem. Three multiple regression analyses were performed using the Process Macro for SPSS. State upward comparison was added as independent variable, social, appearance and performance state self-esteem as dependent variables and inoculation message as moderator variable. Three other multiple regression analyses were performed including the covariates trait upward comparison and trait self-esteem.
The explained variances for all three models were significant (social self-esteem: R2 = 0.43, F (3, 100) = 7.38, p < 0.001, appearance self-esteem: R2 = 0.08, F (3, 100) = 3.06, p < 0.05, performance self-esteem R2= 0.12, F (3, 100) = 4.41, p < 0.01). State upward
comparison was a significant predictor of social, b= -0.20, t = -2.03, p < 0.05, appearance, b = -0.22, t = -2.92, p < 0.01, and performance self-esteem, b= -0.16, t = 2.75, p < 0.01. The inoculation condition was no significant predictor of any of the three self-esteem scales (Appendix 3). The interaction between state upward comparison and inoculation condition was not significant for social or appearance self-esteem and did not add anything to the model. However, the interaction was significant for performance self-esteem, b= 0.24, t = -2.15, p < 0.5, and significantly added to the explained variance in performance self-esteem, R2change = 0.04, F (1, 100) = 4.60, p < 0.05.
When adding the covariates to the moderation model, trait upward comparison only had an effect on social self-esteem, b = -0.12, t = -2.01, p < 0.05. Trait self-esteem, however,
17 had a strong effect on social, b= 0.88, t = 7.74, p < 0.001, appearance, b = 0.98, t = 7.49, p < 0.001, and performance self-esteem, b= 0.83, t = 9.33, p < 0.001. The direct effect of state upward comparison on appearance self-esteem held when controlling for the covariates, but became weaker. The effects on social and performance self-esteem became insignificant. Notable, the interaction effect between inoculation condition and state upward comparison on performance self-esteem also became insignificant. The interaction effect was thus not robust when adding the covariates. Therefore, the support for the moderation analysis is weak and needs more research. Because all interaction effects were insignificant, Hypothesis 3 had to be rejected. However, the results seem to indicate that young adult females who already had a lower self-esteem previous to the experiment, also had a lower self-esteem state.
As mentioned earlier, the analyses were repeated without the respondents that could not correctly recall the color of the top and did not know the influencer (N = 82). Direct effects of the inoculation condition and interactions did not reach significance. One exception was the interaction effect between state upward comparison and inoculation condition on social state self-esteem, which became significant, b = -0.30, t = -2.00, p < 0.05. However, when adding the covariates, this effect diminished and became insignificant. The rejection of the
moderation effect was thus not due to respondents who did not pay attention or did not know the influencer.
Conclusion and discussion
This experiment aimed to examine the psychosocial impact of two relatively unresearched groups on social media, namely unknown peers and influencers on Instagram. The results showed that different Instagram sources indeed stimulate different effects among female young adults. Participants who saw a photo of an influencer tended to compare themselves more to this person than participants who saw an unknown peer. In turn, more upward social comparison led to a decrease in state self-esteem. To summarize, exposure to influencer posts may have a more negative effect on female users’ self-esteem by increasing upward social comparison compared to posts by unknown peers.
To the best of my knowledge, this study was the first to implement an inoculation message as an intervention for negative psychosocial effects caused by exposure to social media content. However, the inoculation message did not change the relationship between upward comparison and self-esteem. This may be a function of the specific inoculation message used here.
18
Theoretical implications
The findings reported here stand in contrast to the Social Comparison Theory by Festinger (1954), which states that people do not tend to evaluate their opinions or abilities with other people who are too different from themselves. Although participants thought they were more similar to the unknown peers than to the influencers, they still compared themselves more to the influencers. However, this finding is logical because influencers seem to have a better life than unknown peers. As a matter of fact, upward comparison is about comparing yourself to others who you perceive to be better than yourself. There may have been more downward comparison with unknown peers. I tried to prevent this difference between upward and downward comparison by selecting photos of unknown peers and influencers that were similar, based on characteristics such as pretty clothes and countries.
Influencers, as a relatively new concept, have only been examined in one study in the context of social comparison and self-esteem (Chae, 2018). While familiar peers (Brown & Tiggemann, 2016; Fardouly & Vartanian, 2015) and friends (Fardouly & Vartanian, 2015; Ho, Lee and Liao, 2016; Wang, Wang, Gaskin & Hawk, 2017) have been looked at in a larger amount of studies, unknown peers are still a new group. Comparing influencers to friends would manipulate both popularity and personally knowing the source. However, by using unknown peers, this study was a more controlled experiment because both sources were not known personally. Participants compared themselves less to these unknown peers than to influencers. This is in contrast with studies that found that females compare more with peers (Fardouly & Vartanian, 2015) and friends (Fardouly & Vartanian, 2015; Ho, Lee and Liao, 2016) than with celebrities. It is plausible that the different results are due to the difference between unknown peers and known peers. As said, people do not tend to compare themselves to people who are too different from themselves (Festinger, 1954). Whilst unknown peers were perceived as more similar than influencers, this difference was not very large. Known peers and friends might have been more similar to female young adults, because they have even more in common, such as the same job, study or interests. It could also be that
participants liked to compare themselves to people who were better off than themselves. This could go via the process of self-enhancement, which means that females have a stronger tendency to compare themselves with desirable others who score the highest on positively valued characteristics (Arrowood & Friend, 1969). So it might be that females in this study compared themselves more to influencers because they thought that influencers had higher scores on achievements in life or appearance.
19 There was a lack of experimental studies in the research field of social media use, social comparison and self-esteem. Relationships were found between social media use and social comparison, but often no causal inferences were possible (Wang, Wang, Gaskin & Hawk, 2017; Chae, 2018; Fardouly & Vartanian, 2015; Brown & Tiggemann, 2016). The current study therefore explored this relationship in an experimental design and found such effect for the two different Instagram photo sources. Former studies also made assumptions about how more social comparison would lead to less self-esteem and related psychosocial factors, but there was again not enough causal evidence for this (Wang, Wang, Gaskin & Hawk, 2017; Hendrickse, Arpan, Clayton & Ridgway, 2017; Ho, Lee & Liao, 2016; Fardouly & Vartanian, 2015). The current study extended the evidence as it found that upward
comparison indeed mediates the relationship between exposure to specific Instagram sources and self-esteem.
Inoculation messages have mainly been studied in other more physical health domains such as smoking, binge drinking or condom use (e.g. Parker, Ivanov & Compton, 2012; Niederdeppe, Heley & Barry, 2015). These studies have proven the effectiveness of inoculation messages on a diversity of psychological outcomes, such as developing
counterarguments towards risk behavior and resistance towards attitude threatening messages. The current study was the first to look at the effect of an inoculation message on psychosocial factors in a social media context. However, no moderation effect was found.
The latter can be explained because there was no difference for the non-inoculation and inoculation conditions in the perceived reality of Instagram photos, nor in the perceived threat of Instagram photos to the self-confidence. This means that either the threatening part of the message was not threatening enough, or the refutation part did not contain information that was persuasive enough (Szybillo & Heslin, 1973). However, the means for perceived reality and perceived threat to the self-confidence were high in both inoculation conditions. Thus, it could be that participants already thought that Instagram photos do not represent reality or form a threat to the self-confidence before this study and therefore did not learn new information. For instance, even though to the best of my knowledge there are no studies available on the awareness of photoshopped images, it is plausible that most female young adults are aware of photo editing. A study found that this group uses photographic filters on their photos themselves (Dhir, Pallesen, Torsheim & Andreassen, 2016).
20 Another possible explanation for why no significant moderation effect was found is the small sample size (N = 104). Although small interaction effects were found, these were not
significant. A larger research group might produce higher or significant interaction effects. Another limitation of the research design is that mainly university students participated, derived from the network of the researcher. The effects of the inoculation message could be weaker among university students because they might already have more knowledge about the reality of Instagram photos. Then the inoculation message has less space to make a change in their cognitions. Results show that participants indeed knew that Instagram photos do not represent reality. However, the study did not control for education level. Additionally, this study did not include a no exposure condition. It is thus important to acknowledge that the effect on social comparison was caused by different sources, but not by Instagram photos in general.
Future directions
Thus, more experimental research is still needed with larger and different research groups, such as older adults, males and non-university students. Since the inoculation message did not work, it is important to create more threatening or stronger messages. Because unknown peers and influencers are barely examined in prior studies, more research on these two groups is still needed. Additionally, the effect of upward comparison on self-esteem also still has to be explored in future studies. Future studies might want to look at other intervention methods such as narratives, in which you try to convince people in a more narrative style by using a storyline with characters. In a recent study on multiple health behaviors, narratives were more successful than inoculation messages in stimulating support for health policy (Niederdeppe, Heley & Barry, 2015).
Practical implications
The results of this experiment confirmed that there is indeed a difference in the extent of upward comparison for different Instagram sources. While upward comparison directly led to a decrease in overall, social and appearance self-esteem, the inoculation message did not serve as an effective intervention for this. This highlights the need for practitioners to develop more effective messages. Based on the results, it is important to acknowledge that there is a new group of users on Instagram called influencers, who especially can have a negative effect on females’ upward comparisons. Furthermore, practitioners might want to target those females that already have a lower self-esteem. The current experiment thus shows that female young
21 adults do experience a negative effect from influencers’ photos on self-esteem via upward comparison, which should be prevented by an appropriate message that targets the
comparisons and negative cognitions about oneself.
22
References
Abidin, C. (2016). “Aren’t these just young, rich women doing vain things online?”: Influencer selfies as subversive frivolity. Social Media+ Society, 2(2). doi: 2056305116641342
Alessandri, G., Vecchione, M., Eisenberg, N., Łaguna, M., & Reynolds, C. R. (2015). On the Factor Structure of the Rosenberg (1965) General self-esteem Scale. Psychological
Assessment, 27(2), 621-635. doi: 10.1037/pas0000073
Allan, S., & Gilbert, P. (1995). A social comparison scale: Psychometric properties to psychopathology. Personality and Individual Differences, 19(3), 293-299. doi: 10.1016/0191-8869(95)00086-L
Arrowood, A. J., & Friend, R. (1969). Other factors determining the choice of a comparison other. Journal of Experimental Social Psychology, 5(2), 233-239. doi: 10.1016/0022-1031(69)90049-3
Brown, Z., & Tiggemann, M. (2016). Attractive celebrity and peer images on Instagram: Effect on women's mood and body image. Body image, 19, 37-43. doi:
10.1016/j.bodyim.2016.08.007
Chae, J. (2018). Explaining females’ envy toward social media influencers. Media
Psychology, 21(2), 246 – 262. doi: 10.1080/15213269.2017.1328312
Cramer, E. M., Song, H., & Drent, A. M. (2016). social comparison on Facebook: Motivation, affective consequences, self-esteem and Facebook fatigue. Computers in Human
Behavior, 64, 739-746. doi: 10.1016/j.chb.2016.07.049
Dhir, A., Pallesen, S., Torsheim, T., & Andreassen, C. S. (2016). Do age and gender
differences exist in selfie-related behaviours?. Computers in Human Behavior, 63, 549-555. doi: 10.1016/j.chb.2016.05.053
Fardouly, J., & Vartanian, L. R. (2015). Negative comparisons about one's appearance mediate the relationship between Facebook usage and body image concerns. Body
23 Festinger, L. (1954). A theory of social comparison processes. Human relations, 7(2),
117-140. doi: 10.1177/001872675400700202
Fetzer Institute. (n.d.). Rosenberg self-esteem Scale. Retrieved from
http://fetzer.org/sites/default/files/images/stories/pdf/selfmeasures/Self_Measures_for_s elf-esteem_ROSENBERG_SELF-ESTEEM.pdf
Field, A. (2005). Factor analysis using SPSS. Retrieved from http://users.sussex.ac.uk/~andyf/factor.pdf
Heatherton, T. F., & Polivy, J. (1991). Development and validation of a scale for measuring state self-esteem. Journal of Personality and Social Psychology, 60(6), 895-910. doi:
10.1037/0022-3514.60.6.895
Hellen, K., Saaksjarvi, M. C. (2012). Celebrity endorsement: The effects of social comparisons on women’s self-esteem and purchase intentions.
Proceedings of the 2nd Nordic Conference on Consumer Research, Gothenborg,
Sweden.
Hendrickse, J., Arpan, L. M., Clayton, R. B., & Ridgway, J. L. (2017). Instagram and college women’s body image: Investigating the roles of appearance-related comparisons and intrasexual competition. Computers in Human Behavior, 74, 92 – 100. doi:
10.1016/j.chb.2017.04.027
Hersey, J. C., Niederdeppe, J., Evans, W. D., Nonnemaker, J., Blahut, S., Holden, D., ... & Haviland, M. L. (2005). The theory of" truth": how counterindustry campaigns affect smoking behavior among teens. Health Psychology, 24(1), 22(10).
Ho, S. S., Lee, E. W. J., & Liao, Y. (2016). Social network sites, friends, and celebrities: The roles of social comparison and celebrity involvement in adolescents’ body image dissatisfaction. Social Media + Society, 2(3), 1 – 11. doi: 10.1177/2056305116664216 Lin, R., & Utz, S. (2015). The emotional responses of browsing Facebook: Happiness, envy,
and the role of tie strength. Computers in human behavior, 52, 29-38. doi: 10.1016/j.chb.2015.04.064
24 Lup, K., Trub, L., & Rosenthal, L. (2015). Instagram #Instasad? Exploring associations
among Instagram use, depressive symptoms, negative social comparison and strangers followed. Cyberpsychology, Behavior, and Social Networking, 18(5), 247 – 252. doi: 10.1089/cyber.2014.0560
Marwick, A. E. (2015). Instafame: Luxury selfies in the attention economy. Public Culture,
27(1 75), 137-160. doi: 10.1215/08992363-2798379
McGuire, W. J. (1961). The effectiveness of supportive and refutational defenses in
immunizing and restoring beliefs against persuasion. Sociometry, 24(2), 184 – 197. doi: 10.2307/2786067
Messeri, P., Haviland, M., & Lyndon Stone, A. A. (2005). The theory of “truth”: How counterindustry media campaigns affect smoking behavior among teens. Health
Psychology, 24(1), 22 – 31. doi: 10.1037/0278-6133.24.1.22
Nesi, J., & Prinstein, M. (2015). Using social media for social comparison and feedback-seeking: Gender and popularity moderate associations with depressive symptoms.
Journal of Abnormal Child Psychology, 43(8), 1427-1438. doi:
10.1007/s10802-015-0020-0
Niederdeppe, J., Heley, K., & Barry, C. L. (2015). Inoculation and narrative strategies in competitive framing of three health policy issues. Journal of Communication, 65(5), 838-862. doi: 10.1111/jcom.12162
Parker, K. A., Ivanov, B., & Compton, J. (2012). Inoculation's efficacy with young adults' risky behaviors: can inoculation confer cross-protection over related but untreated issues? Health Communication, 27(3), 223-233. doi: 10.1080/10410236.2011.575541
Stapleton, P., Luiz, G., Chatwin, H. (2017). Generation validation: The role of social comparison in use of Instagram among emerging adults. Cyberpsychology, Behavior
and Social Networking, 20(3), 142 – 149. doi: 10.1089/cyber.2016.0444
Szybillo, G. J., & Heslin, R. (1973). Resistance to persuasion: inoculation theory in a marketing context. Journal of Marketing Research, 10(4), 396-403.
25 Vogel, E. A., Rose, J. P., Roberts, L. R., & Eckles, K. (2014). Social comparison, social
media, and self-esteem. Psychology of Popular Media Culture, 3(4), 206. doi:
10.1037/ppm0000047
Wang, J. ; Wang, H., Gaskin, J., Hawk, S. (2017). The mediating roles of upward social comparison and self-esteem and the moderating role of social comparison orientation in the association between social networking site usage and subjective well-being.
Frontiers in Psychology, 8. doi: 10.3389/fpsyg.2017.00771
Xu, A. J., & Wyer, R. S. (2012). The role of bolstering and counterarguing mind-sets in persuasion. Journal of Consumer Research, 38(5), 920 – 932. doi: 10.1086/661112 Yang, C. (2016). Instagram use, loneliness, and social comparison orientation: Interact and
browse on social media, but don’t compare. Cyberpsychology, Behavior, and Social
26
Appendix 1: manipulations
Text:
English: Next, you will see a photo of an Instagram user. I want to ask you to look at it as you
would normally do when you use Instagram. After that, you can click through.
Dutch: Je krijgt nu een foto van een Instagram gebruiker te zien. Ik wil je vragen ernaar te
kijken zoals je dat normaal gesproken ook zou doen als je Instagram gebruikt. Daarna kun je doorklikken.
28
Unknown peers’ photos
29
Inoculation message
English: I want to ask you to read the text below. Only click through when you have read the
complete text.
Instagram is full of photos of female Instagram users in nice places, in which their clothes and make-up always seems to look good. Sometimes it can appear like Instagram users are
continuously traveling to pretty countries. Next to that, many Instagram users seem content with their body and appearance and show this with proud in their photos and selfies. When you see such interesting photos, you might start to question if you are interesting, pretty or good enough.
However, Instagram users often position their body in such a way that it looks as good as possible in the photo. Next to that, many photos are photoshopped with the help of Instagram filters or external photo editing apps: from light photoshopping with a color filter, to the removal of pimples or wrinkles and the editing of the body shapes. Therefore, do not think too fast that Instagram photos show reality. Also, photos only give a limited view into the daily life of the Instagram user and therefore it is not realistic to compare yourself with others on Instagram. It is unlikely that the person in the photo continuously travels to pretty places and always has a good time.
30
Dutch: Ik wil je vragen onderstaande tekst te lezen. Klik pas door wanneer je de hele tekst
hebt gelezen.
Instagram staat vol met foto’s van vrouwelijke Instagram gebruikers op leuke plekken, waarbij hun kleding en make-up er altijd goed uit lijkt te zien. Soms kan het overkomen alsof Instagram gebruikers continu op reis naar mooie landen gaan. Daarnaast lijken veel Instagram gebruikers tevreden over hun lijf en uiterlijk en laten ze dat trots zien in hun foto's en selfies. Wanneer jij zulke interessante foto's ziet, begin je jezelf misschien wel af te vragen of jij interessant, knap of goed genoeg bent.
Instagram gebruikers positioneren hun lichaam, echter, vaak op zo’n manier dat het zo goed mogelijk op de foto staat. Daarnaast zijn veel foto’s gefotoshopt met de hulp van Instagram filters of externe foto editing apps: van lichte fotoshop met een kleurenfilter, tot het
wegwerken van puistjes of rimpels en het bewerken van de lichaamsvormen. Denk daarom niet te snel dat Instagram foto's de realiteit weergeven. Foto's geven ook maar een beperkte kijk in het dagelijkse leven van de Instagram gebruiker en daarom is het niet realistisch om jezelf met anderen te vergelijken op Instagram. Het is onwaarschijnlijk dat de persoon in de foto continu naar mooie plekken reist en het altijd goed heeft.
31
Appendix 2: Measurements 2.1 State measurements
State self-esteem scale (Heatherton & Polivy, 1991)
English:
Current Thoughts
“This is a questionnaire designed to measure what you are thinking at this moment. There is, of course, no right answer for any statement. The best answer is what you feel is true of yourself at this moment. Be sure to answer all of the items, even if you are not certain of the best answer.
Again, answer these questions as they are true for you RIGHT NOW.”
Dutch:
Gedachten op dit moment
“Deze vragen zijn ontworpen om te meten wat je op dit moment denkt. Er is uiteraard geen juist of onjuist antwoord voor de stellingen. Het beste antwoord is wat jij denkt dat waar is voor jouzelf op dit moment. Zorg ervoor dat je alle stellingen beantwoordt, zelfs als je niet zeker bent van het beste antwoord.”
Nogmaals, beantwoord de vragen met antwoorden die op dit moment juist zijn voor jou.
English Dutch
Performance items
I feel confident about my abilities. Ik voel mij zeker over mijn vaardigheden. I feel as smart as others. Ik voel mij net zo slim als anderen. I feel good about myself. Ik voel mij goed over mijzelf.
I feel like I’m not doing well. Ik heb het gevoel dat ik niet goed presteer.R
32 I am worried about whether I am regarded
as a success or failure.
Ik maak mij druk over of ik gezien word als een succes of mislukking.R
I feel inferior to others at this moment. Ik voel mij minderwaardig in vergelijking tot anderen op dit moment.R
I am worried about what other people think of me.
Ik maak mij druk om wat andere mensen van mij denken.R
I feel concerned about the impression I am making.
Ik maak mij druk over de indruk die ik maak.R
I am worried about looking foolish. Ik maak mij druk over of ik gek overkom.R
I feel displeased with myself. Ik voel mij ontevreden met mijzelf.R
Appearance items
I feel satisfied with the way my body looks right now.
Ik voel mij tevreden met de manier waarop mijn lichaam er op dit moment uitziet. I am dissatisfied with my weight. Ik ben ontevreden met mijn gewicht.R
I am pleased with my appearance right now. Ik ben tevreden met mijn uiterlijk op dit moment.
33
Upward social comparison: state measurement (Vogel, Rose, Roberts & Eckles, 2014)
English Dutch
State When you looked at the
Instagram post, to what extend did you think you were worse off than the person in the photo?
In welke mate dacht je dat je slechter af was dan de persoon in de foto toen je naar de Instagram post keek?
2.1 Control variables Trait self-esteem
Instructions (Rosenberg, 1965, in Fetzer Institute, n.d., p. 3):
English: “Below is a list of statements dealing with your general feelings about yourself. Please indicate how strongly you agree or disagree with each statement.”
Dutch: “Hieronder vind je een lijst met stellingen omtrent jouw algemene gevoelens over jezelf. Geef graag aan hoe sterk je het eens of oneens bent met elke stelling.”
Items (Rosenberg, 1965, in Fetzer Institute, n.d.):
1. On the whole, I am satisfied with myself. 2. At times I think I am no good at all. R 3. I feel that I have a number of good qualities. 4. I am able to do things as well as most other people. 5. I feel I do not have much to be proud of. R
6. I certainly feel useless at times. R
7. I feel that I'm a person of worth, at least on an equal plane with others. 8. I wish I could have more respect for myself. R
34 10. I take a positive attitude toward myself.
allan upward? Extent less more?
Trait upward social comparison (Vogel, Rose, Roberts & Eckles, 2014)
English Dutch
Trait When comparing
yourself to others on Instagram, to what extent do you focus on people who are better off than you?
Wanneer je jezelf met anderen vergelijkt op Instagram, in welke mate focus jij je dan op mensen die het beter hebben dan jij?
Attention to Instagram photos
English: What was the color of the top or dress of the person in the picture?
A: Blue B: White C: Orange D: Black
Dutch: Wat was de kleur van het topje of de jurk van de persoon in de foto?
A: Blauw B: Wit C: Oranje D: Zwart
2.2 Manipulation checks
1. English: Please make clear to which extent you agree with the following statement, on a scale from 1 (not at all) to 7 (very much)?:
Dutch: Kun je aangeven in welke mate je het eens bent met de volgende stelling, op een
schaal van 1 (helemaal mee oneens) tot 7 (helemaal mee eens)?:
Exposure to different sources
35
Dutch: De persoon in de foto is vergelijkbaar met mij.
2. English: Did you already know the person in the photo before this study?
Dutch: Kende je de persoon in de foto al voorafgaand dit onderzoek?
Inoculation messages
English: Please make clear to which extent you agree with the following statement, on a scale
from 1 (not at all) to 7 (very much)?:
Dutch: Kun je aangeven in welke mate je het eens bent met de volgende stelling, op een
schaal van 1 (helemaal mee oneens) tot 7 (helemaal mee eens)?:
1. English: The pictures on Instagram do not represent reality. Dutch: de foto’s op Instagram geven geen realiteit weer.
2. English: By seeing Instagram photos of others, my self-confidence can be reduced. Dutch: Door het zien van Instagram foto´s van anderen kan mijn zelfvertrouwen verminderd
36
Appendix 3: Tables
Table 1
Conditions
Condition Mean SD Minimum Maximum N
Non-inoculation x influencer 21.75 1.94 18 25 24 Non-inoculation x unknown peer 22.73 1.72 19 25 22 Inoculation x influencer 22.26 1.77 18 25 31 Inoculation x unknown peer 22.19 1.82 18 25 27
37 Table 2
Mediation analysis with overall self-esteem
Step Predictor B SE p R2 F p 1 State comparison Source condition -.2392 -.1437 .0544 .1172 .0000 .2232 .1609 9.6840 .0001 2 State comparison Source condition Trait comparison Trait self-esteem -.1149 -.0161 -.0064 .8847 .0379 .0693 .0337 .0635 .0031 .8170 .8499 .0000 .7175 62.8755 .0000
Note. The steps were separate regression analyses using the Process Macro for SPSS. N = 104. Dependent
variable = overall self-esteem. SE = standard error of B. Source condition: 1 = unknown peer, 0 = influencer. State and trait upward comparison, measured on a 5-point scale (1 = not at all – 5 = a great deal), trait self-esteem measured on a 5-point scale (1 = strongly disagree – 5 = strongly agree), overall state self-self-esteem measured on a 5-point scale (1 = not at all – 5 = extremely).
38 Table 3
Moderation analysis with overall state self-esteem
Step Predictor B SE p R2 F p 1 State comparison Inoculation condition Interaction -.1372 .2560 -.1648 .0761 .2970 .1045 .0744 .3908 .1177 .1887 7.7513 .0001 2 State comparison Inoculation condition Interaction Trait comparison Trait self-esteem -.1351 -.1048 .0435 -.0051 .8989 .0501 .1797 .0642 .0341 .0663 .0082 .5613 .4995 .8812 .0000 .7187 50.0887 .0000
Note. The steps were separate regression analyses using the Process Macro for SPSS. N = 104. Dependent
variable = overall self-esteem. SE = standard error of B. Inoculation condition: 1 = inoculation message, 0 = no inoculation message. State and trait upward comparison, measured on a 5-point scale (1 = not at all – 5 = a great deal), trait esteem measured on a 5-point scale (1 = strongly disagree – 5 = strongly agree), overall state self-esteem measured on a 5-point scale (1 = not at all – 5 = extremely).
39 Table 4
Moderation analysis with social state self-esteem
Step Predictor B SE p R2 F p 1 State comparison Inoculation condition Interaction -.2004 .2972 -.1801 .0988 .3854 .1356 .0451 .4425 .1871 .1812 7.3774 .0002 2 State comparison Inoculation condition Interaction Trait comparison Trait self-esteem -.1280 .0065 .0094 -.1168 .8750 .0854 .3064 .1094 .0581 .1130 .1370 .9831 .9316 .0473 .0000 .5105 20.4382 .0000
Note. N = 104. The steps were separate regression analyses using the Process Macro for SPSS. Dependent
variable = social state self-esteem. SE = standard error of B. Inoculation condition: 1 = inoculation message, 0 = no inoculation message. State and trait upward comparison, measured on a 5-point scale (1 = not at all – 5 = a great deal), trait self-esteem measured on a 5-point scale (1 = strongly disagree – 5 = strongly agree), social state self-esteem measured on a 5-point scale (1 = not at all – 5 = extremely).
40 Table 5
Moderation analysis with appearance state self-esteem
Step Predictor B SE p R2 F p 1 State comparison Inoculation condition Interaction -.2218 -.1360 -.0654 .0759 .1642 .1519 .0043 .4095 .6678 .0842 3.0642 .0315 2 State comparison Inoculation condition Interaction Trait comparison Trait self-esteem -.1428 .0461 .1713 .0823 .9758 .0732 .1354 .1262 .0670 .1303 .0540 .7342 .1777 .2224 .0000 .4202 14.2044 .0000
Note. The steps were separate regression analyses using the Process Macro for SPSS. N = 104. Dependent
variable = appearance state self-esteem. SE = standard error of B. Inoculation condition: 1 = inoculation
message, 0 = no inoculation message. State and trait upward comparison, measured on a 5-point scale (1 = not at all – 5 = a great deal), trait self-esteem measured on a 5-point scale (1 = strongly disagree – 5 = strongly agree), appearance state self-esteem measured on a 5-point scale (1 = not at all – 5 = extremely).
41 Table 6
Moderation analysis with performance state self-esteem
Step Predictor B SE p R2 F p 1 State comparison Inoculation condition Interaction -.1551 -.1640 -.2422 .0564 .1221 .1129 .0071 .1821 .0343 .1169 4.4115 .0059 2 State comparison Inoculation condition Interaction Trait comparison Trait self-esteem -.0796 -.0065 -.0440 .0541 .8252 .0497 .0919 .0857 .0455 .0884 .1124 .9442 .6088 .2376 .0000 .5335 22.4158 .0000
Note. The steps were separate regression analyses using the Process Macro for SPSS. N = 104. Dependent
variable = performance state self-esteem. SE = standard error of B. Inoculation condition: 1 = inoculation message, 0 = no inoculation message. State and trait upward comparison, measured on a 5-point scale (1 = not at all – 5 = a great deal), trait self-esteem measured on a 5-point scale (1 = strongly disagree – 5 = strongly agree), performance state self-esteem measured on a 5-point scale (1 = not at all – 5 = extremely).