Appetite 99 (2016) 121e129

Contents lists available at ScienceDirect

Appetite
journal homepage: www.elsevier.com/locate/appet

Adolescents' presentation of food in social media: An explorative

study
Christopher Holmberg a, *, John E. Chaplin b, Thomas Hillman c, Christina Berg a
a
Department of Food and Nutrition, and Sport Science, University of Gothenburg, Laroverksgatan 11, 405 30, Gothenburg, Sweden
b
Sahlgrenska Academy at Gothenburg University, 416 85, Gothenburg, Sweden
c
Department of Education, Communication and Learning, University of Gothenburg, Laroverksgatan 5, 405 30, Gothenburg, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: The study aimed to explore how adolescents communicate food images in a widely used social media
Received 31 July 2015 image-sharing application. We examined how and in what context food was presented and the type of
Received in revised form food items that were frequently portrayed by following a youth related hashtag on Instagram. The
23 October 2015
hashtag #14år (“14 years”) was used to find adolescent users on Instagram: these users public photo
Accepted 8 January 2016
Available online 11 January 2016
streams were then searched for food items they had shared with others. Food items were identified and
categorized based on type of food and how the food items were presented. Most of the adolescent users
(85%) shared images containing food items. A majority of the images (67.7%) depicted foods high in
Keywords:
Adolescents
calories but low in nutrients. Almost half of these images were arranged as a still life with food brand
Health communication names clearly exposed. Many of these images were influenced by major food marketing campaigns.
Instagram Fruits and vegetables occurred in 21.8% of all images. This food group was frequently portrayed zoomed
Online food communication in with focus solely on the food, with a hashtag or caption expressing palatability. These images were
Photo food presentation often presented in the style of a cook book. Food was thus presented in varied ways. Adolescents
Social media themselves produced images copying food advertisements. This has clear health promotion implications
since it becomes more challenging to monitor and tackle young people's exposure to marketing of un-
healthy foods in these popular online networks because images are part of a lifestyle that the young
people want to promote. Shared images contain personal recommendations, which mean that they may
have a more powerful effect than commercial advertising.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction unique possibility to observe adolescents' dietary communication.

Food practices reflect social and cultural values (Neely, Walton,
& Stephens, 2014; Roper & La Niece, 2009), and as food plays an
essential role in everyday life, there is much scope to consider the
social aspects of young people's food practices as important ele-
ments of their wellbeing and health (Neely et al., 2014). In order to
gain a better understanding of what food means to adolescents it is
necessary to study their communication regarding food, particu-
larly what food-related messages they choose to present to one
another. Adolescents today communicate to a large extent via social
media applications and services (McBride, 2011), and the Internet
has become a growing part of children's foodscape (Brembeck &
Johansson, 2010). These newly emerged channels give us a
* Corresponding author.
E-mail address: christopher.holmberg@gu.se (C. Holmberg).
Drawing on Goffman's (1959) work on self-presentation, Boyd
(2014) argues that what adolescents convey to others in social
media is a matter of what they choose to share in order to make a
good impression and also what they unintentionally communicate
as a result of the medium and the interaction.
Experimental studies have demonstrated that peers in a social
media setting can influence children and adolescents to adjust their
candy intake (Bevelander, Anschutz, Creemers, Kleinjan, & Engels,
2013) and choose unfamiliar foods (Bevelander, Anschutz, &
Engels, 2012). Research in neuroscience also indicates that images
of food can affect appetite-related brain activity (Beaver et al.,
2006; Killgore & Yurgelun-Todd, 2006; Schienle, Scha €fer, Her-
mann, & Vaitl, 2009).
With the increase in the possibilities for image based commu-
nication, Swedish adolescents have moved away from text based
communication (Findahl, 2014). It has been reported that most
Swedish adolescents (91%) aged between 13 and 16 years use social

http://dx.doi.org/10.1016/j.appet.2016.01.009
0195-6663/© 2016 Elsevier Ltd. All rights reserved.
122 C. Holmberg et al. / Appetite 99 (2016) 121e129
media and that Instagram, a mobile-based image based application,
was the most used service, utilized by 75% (85% girls and 64% boys)
of young Swedish people (Swedish Media Council, 2015). A similar
smartphone-facilitated social media trend has been noted in the
US, where 52% of teens aged 13e17 years old are reported to use
Instagram (Pew Research Center, 2015). As of December 2014, only
four years after its launch, Instagram had around 300 million
monthly active users worldwide (Instagram, 2014b).
When it comes to analyzing food related content in social media
specifically (as opposed to online media in general, such as web-
sites), there has been limited research. Some researchers have
investigated how food is expressed and communicated in online
communities and blogs (e.g. Lynch, 2010; Simunaniemi, Sandberg,
Andersson, & Nydahl, 2011), while others have nutritionally
analyzed recipes on popular food blogs (Schneider, McGovern,
Lynch, & Brown, 2013). When analyzing images shared on Insta-
gram, researchers observed that a substantial proportion of these
were of food items (Hu, Manikonda, & Kambhampati, 2014). Pre-
vious research has used Instagram tags (Sharma & Choudhury,
2015) or likes (Mejova, Haddadi, Noulas, & Weber, 2015) to study
food communication on shared images on Instagram, but to our
knowledge there are no studies that have analyzed how food is
presented visually.
The present study aims to explore how adolescents portray food
in a widely used image-sharing application. We examine how and
in what context food was presented, how the uploader describe the
images, and the type of food items that were frequently portrayed
by following a youth related hashtag on Instagram.
2. Material and methods
A content analytic approach to the analysis of Instagram images
was employed. Krippendorff (2004) contends that content analysis
is a research technique for the systematic study of communication.
Riffe, Lacy, and Fico (2005), argue that content analysis is a repli-
cable method for analyzing both visual and written content. This
approach is suitable because Instagram images are multimodal
productions containing both visual and written content. We
considered both the images and written descriptions provided by
the uploaders. In the present study, almost all (94.6%) of the images
contained hashtags or captions. See Fig. 2 for fabricated examples
that typify the Instagram images. Content analysis is regarded as an
unobtrusive method, because materials are examined after they are
produced (Krippendorff, 2004). By using already available data we
avoided the risk that our data collection method intruded in the
research context. Although content analysis, by itself, does not
establish how viewers apprehend or value what they see, it in-
dicates what is given importance or salience (Leeuwen & Carey,
2001). Content analysis thus offers a quantified dimensional
description of fields of representation.
Data derived from social media and the Internet is considered to
be in the public realm if it is not password protected or when a
subscription is not required to access the content (Hookway, 2008).
The Association of Internet researchers (AoIR) (Markham &
Buchanan, 2012), recommends that when conducting research
into online social network services, researchers should investigate
how content privacy is defined in the terms of service (TOS)
agreement. Instagram (2014a) state in their privacy policy that
information or content that the users voluntarily disclose for
posting to the service becomes available to the public and that user
content may be used and altered by others. In other words,
uploaders provide a copyright release by using the public section of
the service, there is no expectation of privacy. However, despite this
adolescents can be viewed as a vulnerable group in society and
adolescents may not be fully aware of the consequences of sharing
personal information publically online. Personal information
transmitted online can quickly be shared with a wider audience
than was anticipated by the uploader. Which can have unforeseen
and long-term consequences (Fritz, 2014). Thus, following recom-
mendations by Moreno, Goniu, Moreno, and Diekema (2013), we do
not present unique combinations of images that were linked to one
individual. Instead, in the following presentation, fabricated ex-
amples using the first author's Instagram account were created to
illustrate the results, see Fig. 2. The study protocol was evaluated by
the regional ethical review board in Gothenburg (DNR: 468-14).
2.1. Instagram
Instagram is a free mobile application which permits users to
upload images and videos, edit the media by adding different filters
and share the images within the user's social network and via a
metadata hashtag which functions as a label and as searchable
keyword (Schlesselman-Tarango, 2013), enabling other users to
find the material. The user can also assign a caption to the image.
Similar to other online social networks, such as Facebook, users of
Instagram are able to comment and ‘like’ each other's images.
Instagram account holders must be at least 13 years of age
(Instagram, 2014c).
2.2. Sample
We utilized a user generated age-related hashtag in Instagram
to identify adolescents' photo streams for analysis. Following a
hashtag label allowed us to explore adolescent communication
patterns without interfering with the production of the images. The
hashtag #14år (Swedish, Danish and Norwegian for “14 years”) was
used as the searchable label. We tested several different hashtags to
identify adolescent users. Hashtag 14år (#14år) was chosen as it is a
good indicator of the age of the user, mostly predicting adolescents
who recently turned 14 years.
The third party web-application Statigram was used to track,
manage and analyze the images generated via Instagram, as
Instagram is primarily a mobile-based service. The selection of
accounts was based on identifying images appended with the
hashtag. In Statigram the hashtag was found to have been applied
to 3479 images as of March 2014. However, as users change their
privacy settings from public to private, delete their accounts, or
change their user names, 1358 images were not retrievable. As
some accounts contained several images appended with this
hashtag, 409 images were removed because the uploader had
applied the hashtag to multiple images. Thus, 1712 Instagram ac-
counts were available. When these accounts were identified, we
excluded 711 accounts since they, based on written and visual
profile information, were judged not to belong to an adolescent
user. For example, the hashtag could relate to a user's pet turning 14
years or images of couples celebrating 14 years of marriage. Only
users who could be identified as adolescents (i.e. including infor-
mation about their age in their profile) were included in the study.
Thus, 1001 Instagram accounts were eligible for analysis (see Fig. 1.).
Each account was searched for images visibly showing food for
human consumption. In accounts with such images, the first food
image prior to the hashtag was selected for analysis. To avoid an
overrepresentation of birthday related images, the selected image
had to appear at least two days before the tagged image (#14år).
Estimation of Instagram users' gender, and language, was done
using textual profile information (bio-information), profile photos,
and uploaded images in their stream (e.g. ‘selfies’).
 C. Holmberg et al. / Appetite 99 (2016) 121e129 123

Fig. 1. Sampling procedure.

2.3. Categorization and analyses
We focused on categorizing what kind of food items were
depicted in the images; how food was displayed; in which context
food was presented, and how the uploader described the image
with captions and hashtags, see Tables 1 and 2.
Food items identified in the photos were categorized and we
also used captions and hashtags if they provided additional infor-
mation about the food item than was clearly visible in the image.
For example, the category “fruits” was applied when the hashtag
#banana was used in conjunction with an image depicting
smoothies. Similarly, for complex foods and meals, items were
coded individually if clearly visible (e.g. a sandwich with tomato
slices was categorized as “Bread” and “Vegetables”).
To describe food presentation, we created categories reflecting
the context of the situation, how food was displayed, and how the
food was described by the uploader. Context referred to which
setting food was present in and how food was prepared. Food
display concerned how food was arranged and how the photo was
taken. Description of food referred to how the uploader described
the image through hashtags and captions.
The categories were chosen to cover different aspects of the
images; therefore they were not mutually exclusive. One Instagram
image could be assigned to several categories. See Fig. 2a for an
illustration, this Instagram image is both categorized as “Eating out
meal” as well as “People portrayed with food”. The objective of the
coding was to provide a descriptive assessment of photo content,
not to hypothesize on the motivation of the uploader.
We used a general inductive approach for analysis as there were
few earlier studies concerning this phenomenon. Coding was per-
formed in four steps. First, images with food were identified, then
principles for categorization were developed and evaluated within
the research group, a coding protocol was agreed, and finally the
materials were coded. The first author manually classified the
124 C. Holmberg et al. / Appetite 99 (2016) 121e129

Fig. 2. Fabricated Instagram photos that illustrate the five typical categories of a) Eating out meal and People portrayed with food, b) Food is arranged and Branding c) Zoomed-in
food image.

Table 1
Categories reflecting how food was presented in the Instagram images.

Category Description of category

Food context, displayed visually or in text

Eating out meal Food is situated in or carried as take-away from a commercial food establishment (e.g. restaurant, bistro, cafe ). See Fig. 2a for an example.
Homemade Food is made at home (e.g. caption or hashtags states that the uploader cooked the food) or the image is clearly signifying that it is homemade (e.g.
buns on a baking tray or in an oven).
Special occasion Food is presented during a special occasion (an occasion or event not occurring very often such as a holiday or birthday party).
People portrayed with Food is presented in company with a person. The face and a large portion of the person(s) body are visible. See Fig. 2a for an example.
food
Food eaten Food is consumed actively in the image. Food on a plate is not sufficient; e.g. a person drinking a beverage in the image or bite marks on a
sandwich.
Food display, visually or in text
Food is arranged Food items have been arranged artistically before the photo was taken. See Fig. 2b for an example.
Food is handheld Food items are displayed by being held up in front of the camera by the uploader
Zoomed-in food image Food items are the only thing in focus in the image. Image is zoomed in directly on food, similar to images found in cook books. The food is not in
containers such as bottles and cans. See Fig. 2c for an example.
Branding Food related brands clearly displayed centrally in the Instagram image (e.g. names of restaurants, food manufacturers' logos). Similar to adverts.
See Fig. 2b for an example.
Textual description of situation or food
Tonality Instagram is positive, negative or neutral in tone. Usually with the presence of expressive emoji's and captions, such as hashtags or captions
describing feelings and emotions. See Fig. 2aec for examples.
Expression of Adjectives related to palatability in the caption or hashtags (e.g. #yummy, delicious). See Fig. 2c for an example.
palatability

images into categories. Confirmatory classification was carried out
by the other authors who collectively agreed on which categories to
use. The last author independently coded a random 10% (n ¼ 85) of
the sample in order to establish inter-rater reliability. A Cohen's
kappa analysis was used to take into account chance agreement
(Sun, 2011). The analysis showed that agreement for food group
classification was between k ¼ 0.71 and k ¼ 1.0. For categories
reflecting the food context, food display, and textual description,
values between k ¼ 0.73 and k ¼ 1.0 were reached.
To analyze if the way food was depicted was linked to the type of
food presented we created two aggregated contrasting food cate-
gories of importance to health that were prevalent in the sample.
The two food categories were foods high in calories but low in micro
nutrients (HCLN; e.g. candy, soda, ice cream, cake), see Table 2 for a
complete list, and fruits and vegetables (F&V; i.e. fruits, fruit juice,
vegetables, and berries). These two categories did not cover all food
items identified.
Chi-squared tests were applied to test if the two food categories
were presented differently compared to other food groups.
3. Results
The median time between tag and image was 36 days (min ¼ 2
days, max ¼ 610 days). Of the 1001 Instagram accounts food images
 C. Holmberg et al. / Appetite 99 (2016) 121e129
Table 2
Frequency of food groups in Instagram images (total food images, n ¼ 854).
Food groups
Cookies and pastrya (e.g. buns, desserts, cakes, cupcakes)
Soda and lemonadea
Chocolatea (including chocolate sauce)
Unidentifiable beverages
Ice creama (including popsicles and milkshakes)
Candya (including caramel sauce)
Fruitsb
Vegetablesb
Berriesb
Sauce (e.g. dressing, mayonnaise, ketchup, soy sauce)
High-fat dairya (e.g. cream on beverages or next to desserts)
Fast food (i.e. french fries, pizza, hamburger, hot dog)
Meat (e.g. red meat, poultry, cured meat)
Chocolate milk (e.g. hot chocolate)
Bread
Savory snacksa (e.g. potato chips, cheese doodles, popcorn)
Coffee and tea
Unidentifiable dishes
Energy drinksa
Milk and yoghurt (including sweetened yoghurt)
Water (e.g. tap water, sparkling water)
Pancakes and waffles
Rice
Fish and shellfish
Cheese (including cottage cheese)
Pasta
Potato (e.g. potato gratin, roesti, boiled, fried, mashed)
Food fats (e.g. butter, margarine, oil)
Sugara (e.g. syrup, honey)
Fruit juiceb
Other cereal (e.g. porridge, formula, muesli)
Jam and marmalade
Quiche, pierogi, tacos, wrap
Alcohola
Egg (including egg dishes such as omelette)
a Most of the images containing food brands were of sodas, and the
Included in HCLN category.
b
Included in fruits and vegetables category (F&V).
were found in 854 (85.3%) accounts. Of these, 625 (73.2%) belonged
to users who were identified as girls, while 228 (26.7%) to boys, one
user was not identified as either a boy or girl. A majority of the users
wrote in Swedish 475 (55.6%). Two other languages were used,
Norwegian 258 (30.2%), Danish 119 (13.9%), and two users used a
combination of these. All of the accounts used one of these three
languages, however, English phrases and words were used
repeatedly in many profiles. Food was central in 95.8% of the 854
images, that is, the food was centrally placed in the photo or spe-
cifically referred to in the description. Images which included food
in the periphery of the photo were found in 4.2% of cases.
3.1. Types of foods
Food items were coded into 35 food groups (Table 2). The most
commonly identified food groups were cookies and pastry, found in
20.0% of the images. Soda and lemonade were found in almost as
many, 18.5%, of the images. Other foods that frequently appeared in
the images were chocolates (15.8%), ice cream (11.2%), and candy
(11.0%). Also, more healthy food groups such as fruits were identi-
fied in 9.4% of the images, vegetables in 9.3%, and berries in 8.5%.
3.2. Food presentation
Table 3 shows how the foods are portrayed in the images, i.e. in
which social context, how it is displayed in the image, and how it is
described by the uploader. In one quarter of the images, outside
s, restaurants, bistros were shown. Fig. 2a
eateries, e.g. cafe
125
illustrates a typical Eating out meal. Many of these images depicted
high-fat dairy beverages (e.g. Frappuccinos from Starbucks, hot
% chocolate) and desserts with cream which is why high-fat dairy is
20.0%
the biggest food group in the category. Another reoccurring cate-
18.5% gory identified in ten percent of the images included food items
15.8% that were homemade. Cookies and pastry were dominant when it
11.9% comes to food prepared at home, constituting 63.2% of these
11.2%
homemade foodstuffs.
11.0%
9.4% On 22.2% of the images a special occasion was portrayed such as
9.3% a party or holiday. The biggest food groups identified in these im-
8.5% ages were of sodas and lemonades, and cookies and pastry.
8.1%
People were portrayed in a quarter of the images. It could be
8.0%
7.6%
themselves (e.g. taking a selfie while eating) or of a peer sitting next
6.7% to them in a cafe, see Fig. 2a for an example of a typical image from
6.6% this category. However, only 15.5% of the Instagram images depic-
5.6% ted partly consumed food items, thus, in most images, where
5.5%
people were portrayed, food was not seen to be consumed.
5.2%
4.2% Around one fifth of the images contained food items that were
3.7% displayed as in an exhibition before the photo was taken, e.g. food
3.7% items were lined-up, arranged in a pictorial fashion. These food
3.4% items were usually packaged. See example in Fig. 2b. Many of these
3.2%
2.0%
images were of chocolate, candy, and soda. These often included
2.0% captions indicating the anticipation of a gaming night or a party.
1.9% Another way of displaying the food items was to hold it up before
1.9% the camera to showcase it. These food items usually consisted of
1.8%
soda bottles and ice cream cones. In 19.2% of the images, the
1.8%
1.8% uploader had zoomed in on the food items, as illustrated in Fig. 2c.
1.5% Although the most frequent food items in this category were
1.2% cookies and pastry, the second, third and fourth largest food groups
1.1% in this category were fruits, berries, and vegetables.
1.1%
0.8%
In as many as 37.2% of the images a brand name was promi-
0.5% nently displayed as it would be in an advertisement, see Fig. 2b.
most frequent brand depicted was Coca-Cola. See Table 4 for a list of
frequent food related brands. ther prominent brands appearing in
the images were the American ice cream company Ben and Jerry's
and the coffee company Starbucks. These brands sometime con-
tained captions or hashtags indicating that the uploader was
participating in a campaign, wherein the brands encourage con-
sumers to post pictures to promote their products. Almost half
(n ¼ 24) of the images with Coca-Cola were clearly influenced by
the “Share a Coke” campaign where uploaders took photos of Coca-
Cola bottles with an individual's name on it. It could be the
uploader's own name or someone in their social network which the
uploader then tagged in the Instagram image. Some images pro-
moting Ben & Jerry's (n ¼ 4) and Starbucks (n ¼ 5) also contained
these type of campaign-related hashtags, endorsing different fla-
vors of the companies' products. Several labels of energy drinks and
chocolate were also depicted in the images. Food brand names
were also shown more peripherally in other images, and in total,
almost half (44.7%) of all images contained a brand name.
Regarding tonality, we found that the uploader assigned positive
adjectives and symbols to most of the images containing food
(74.8%). The positive captions, hashtags or symbols (e.g. emojis)
were related to the food or the situation in general, such as
“#havinganicetime” or “#Love”, see Fig. 2a and c. Negative ex-
pressions were rarer and only identified in 1.3% of the images. The
uploader expressed palatability on 20.5% of the images while ex-
pressions of unpleasant taste were found in only two Instagram
images.
3.3. Distinguishing the presentation of HCLN and F&V foods
In total, high calorie low nutrient (HCLN) food items such as
cookies and pastry, soda and lemonade, chocolates, ice cream, and
126 C. Holmberg et al. / Appetite 99 (2016) 121e129

Table 3
Frequency of images in different food presentation categories together with the most common food groups.

Category n (% of total Most frequent food groupsa (Threshold set at 10%)

images)

Food context, displayed visually or in text

Eating out meal 214 (25.1%) High-fat dairy (22.0%), Soda and lemonade (19.6%), Unidentifiable beverages (19.1%), Fast food (18.7%), Chocolate (17.8%), Ice
cream (16.4%), Coffee and tea (14.5%), Sauce (14.5%), Cookies and pastry (14.0%) Chocolate milk (13.6%), Vegetables (11.2%)
Homemade 76 (8.9%) Cookies and pastry (63.2%), Fruits (10.5%)
Special occasion 190 (22.2%) Soda and lemonade (23.7%), Cookies and pastry (21.1%), Ice cream (15.8%), Chocolate (15.8%), Candy (15.3%), Unidentifiable
beverages (14.2%), Savory snacks (12.1%), Fruits (11.1%), Berries (10.0%)
People portrayed with 224 (26.2%) Soda and lemonade (24.1%), Unidentifiable beverages (23.7%), Fast food (13.4%), Cookies and pastry (11.2%), Candy (10.3%), High-
food fat dairy (10.3%)
Food eaten 129 (15.1%) Soda and lemonade (27.1%), Fast food (17.8%), Cookies and pastry (11.6%), Ice cream (10.1%)
Food display, visually or in text
Food is arranged 183 (21.4%) Chocolate (28.6%), Candy (25.8%), Soda and lemonade (24.2%), Cookies and pastry (23.6%), Ice cream (13.7%), Fruits (13.2%), Savory
snacks (12.6%), High-fat dairy (10.4%)
Food is handheld 138 (16.2%) Soda and lemonade (23.2%), Ice cream (15.9%), Chocolate (14.5%), Chocolate milk (10.1%)
Zoomed-in food image 164 (19.2%) Cookies and pastry (34.8%), Fruits (18.3%) Berries (17.7%), Vegetables, (15.2%), Chocolate (13.4%), Fast food (11.2%), Sauce (10.4%)
Branding 318 (37.2%) Soda and lemonade (33%), Chocolate (21.4%), Ice cream (16%), Candy (15.1%), High-fat dairy (12.6%), Unidentifiable beverages
(11.9%), Cookies and pastry (10.7%)
Textual description of situation or food
Positive tone 639 (74.8%) Cookies and pastry (21.7%), Soda and lemonade (17.2%), Chocolate (17.1%), Ice Cream (12.5%), Candy (11.5%), Unidentifiable
beverages (11.3%)
Negative tone 11 (1.3%) Soda and lemonade (27.3%), Candy (18.2%), Sauce (18.2%), Unidentifiable food (18.2%)
Expression of 175 (20.5%) Cookies and pastry (26.0%), Chocolate (18.1%), Berries (13.0%), Ice cream (11.9%), Vegetables (10.7%), Fruits (10.7%)
palatability
Expression of 2 (0.2%) Soda and lemonade (50%), Candy (50%)
unpleasant taste
a
As several food groups can be depicted on the same image, total percentage can be more than 100%.

candy were depicted in 67.7% of the images. By comparing images 4. Discussion

containing HCLN food items with all other food images, we aimed
to explore in which settings this food group were particularly Results from this study show that many adolescent Instagram
frequent, see Table 5. HCLN foods were more frequent in a context users share images of food, and that food is portrayed in different
of a special occasion than other foods. These types of foods were but often aesthetic ways. Although the images in our study are not
more often displayed in an arranged photo of food items, a still life, likely to be representative of the adolescents' everyday eating
and with the brand in focus, as in an advertisement. These images patterns, it is interesting to know what the adolescents chose to
also carried more overall positive expressions in captions or share and how they present and frame food to each other.
hashtags. However, HCLN foods were less frequently zoomed in Food was only seen actively consumed in 15.1% of the images.
compared to other foods. The majority of photos could be interpreted as the anticipation of a
Overall, fruits, vegetables, and berries (F&V) were present in meal, where food items could be seen functioning as props high-
21.8% of the images. Fruit and vegetables occurred less often in lighting an occasion, an activity, or a feeling. The fact that 74.8% of
images where people were portrayed compared to images con- descriptions by the uploaders referred to positive general feelings
taining other foods. Fruit and vegetables were less frequently dis- and emotions, while only 20.5% of the descriptions focused on the
played with brands and held in front of the camera, but more taste of the food items, reinforces the assumption that foods served
frequently shown zoomed in and were given captions that as integrated components of an activity. Drawing on Boyd's (2014)
expressed positive palatability. work, we suggest that food function as objects in adolescents
identity creation on Instagram. In our study many of the images
were arranged and displayed in a way which indicates great care
and attention in the presentation which could represent a part of an
adolescent's online self-representation.
Table 4 We found two main ways that food had been presented in
Most frequent brands in the category Branding (n ¼ 318). Instagram; one that focuses on either the aesthetical features or
Brand N % of Branding images home-made qualities of the food; and the other where the focus is
Coca-Cola 55 17.3
on food being a part of a lifestyle or a situation.
Ben & Jerry's 29 9.1 Images depicting fruits and vegetables were commonly por-
Starbucks 27 8.5 trayed as if from a cook book and these images were significantly
McDonald's 15 4.7 associated with the user expressing palatability. The positive
Marabou 14 4.4
framing indicates that fruits and vegetables contained a certain
Espresso House 14 4.4
Fanta 11 3.5 status and were worth sharing in the social networks, which can be
Pepsi 9 2.8 seen as positive from a health promotion perspective. This way of
Red Bull 9 2.8 portraying fruits and vegetables is something that could be used in
Cloetta 6 1.9
dietary health promotion activities targeting adolescents. As these
Oreo 5 1.6
Monster 5 1.6
images often were colorful and aesthetically pleasing they may
Urge 4 1.3 invite further consumption of these food groups. Images that solely
Sprite 4 1.3 focused on the food items, when the image was zoomed in on the
Freia 4 1.3 foods, often portrayed food that were homemade where the
Tine 4 1.3
intention of the uploader appears to show their cooking skills. As
 C. Holmberg et al. / Appetite 99 (2016) 121e129 127

Table 5
Frequency of images in each food presentation category that depict HCLN food items and fruits and vegetables (F&V) respectively, compared to images not depicting these food
groups.a

Category HCLN Non-HCLN Pb F&V Non-F&V Pc

n ¼ 578 n ¼ 276 n ¼ 186 n ¼ 668
% (n) % (n) % (n) % (n)

Eating out meal 23.7 (137) 27.9% (77) ns 22.6% (42) 25.7% (172) ns
Not eating out meal 76.3% (441) 72.1% (199) 77.4% (144) 74.3% (496)
Homemade 9.0% (52) 8.7% (24) ns 5.9% (11) 9.7% (65) ns
Not homemade 91.0% (526) 91.3% (252) 94.1% (175) 90.3% (603)
Special occasion 24.4% (141) 15.6% (49) <0.05 24.7% (46) 21.6% (144) ns
Not a special occasion 75.6% (437) 84.4% (265) 75.3% (140) 78.4% (524)
People portrayed with food 23.4% (135) 32.2% (89) <0.01 18.3% (34) 28.4% (190) <0.01
People not portrayed with food 76.6% (443) 67.8% (187) 81.7% (152) 71.6% (478)
Food eaten 13.5% (78) 18.5% (51) ns 10.8% (20) 16.3% (109) ns
Food not eaten 86.5% (500) 81.5% (225) 89.2% (166) 83.7% (559)
Food is arranged 27.0% (157) 9.8% (27) <0.001 19.9% (37) 21.9% (147) ns
Food is not arranged 73.0% (421) 90.2% (249) 80.1% (149) 78.1% (521)
Food is handheld 17.0% (98) 14.5% (40) ns 6.5% (12) 18.9% (126) <0.001
Not handheld 83.0% (480) 85.5% (236) 93.5% (174) 81.1% (542)
Zoomed-in food image 16.1% (93) 25.7% (71) <0.01 34.4% (64) 15.0% (100) <0.001
Food not in zoom 83.9% (485) 74.3% (205) 65.6% (122) 85.0% (568)
Branding 46.5% (269) 17.8% (49) <0.001 10.2% (19) 44.8% (299) <0.001
No branding 53.5% (309) 82.2% (227) 89.8% (167) 55.2% (369)
Positive tone 79.1% (457) 69.9% (193) <0.01 78.0% (145) 75.6% (505) ns
Not a positive tone 20.9% (121) 30.1% (83) 22.0% (41) 24.4% (163)
Expression of palatability 20.4% (118) 20.7% (57) ns 26.3% (49) 18.9% (126) <0.05
Not expression of palatability 79.6% (460) 79.3% (219) 73.7% (137) 81.1% (542)
a
Omitting the images containing both HCLN and fruits and vegetables (76, 8.8%) (e.g. chocolate dipped strawberries) did not substantially change the results displayed in
the table.
b
Differences between images containing HCLN food items compared to images not containing HCLN food items, according to Pearson Chi-Square tests.
c
Differences between images containing fruits and vegetables compared to images not containing fruits and vegetables, according to Pearson Chi-Square tests.

cookbooks are popular and cooking programs on TV are abundant,
these images would appear to be inspired by this phenomenon
which have flooded the markets in recent years (Collins, 2009;
Counihan & Van Esterik, 2013). Although this category mostly
contained cookies and pastry in our study, this stylistic approach
could be used when promoting home cooked meals for adolescents.
For example, it could be used in the school subject Home and
Consumer Studies, to encourage adolescents to photograph and
share their endeavors in the kitchen with fellow classmates. This
might inspire other adolescents to cook at home.
The second approach to presenting food was a more lifestyle
and situation oriented approach. The uploaders focused on partic-
ular situations and associated the images with major food brands
and food establishments. Images were arranged, and images were
zoomed out so that people and locations could be included in the
image. Best (2014) argues that a useful way to understand young
peoples' food consumption is to not only focus on what young
people eat but also where they eat. The places in which young
people consume and interact with food are embedded into inter-
personal relationships thus structuring how young people interact
with food. It can be understood therefore that location plays an
essential role in the food event and it is this that is communicated
to other young people (Best, 2014). In our study, eating out meals
were common, and these outside eateries were mostly coffee shop
chains. This is also reflected in the foods presented in the images. In
addition to the food groups coffee and tea (5.2%) and cookies and
pastry (20.0%), a major portion of the beverages that were cate-
gorized as unidentifiable (11.9%) were beverages that were placed
in cups from these coffee shops. The consumption of coffee and
coffee shops was thus a popular motif.
The food images contained a high amount of high calorie low
nutrient food items (67.7%). Many of these foods were associated
with brands, such as beverages containing high-fat dairy products
(e.g. Frappuccinos with cream from Starbucks), but also sodas (e.g.
Coca-Cola), and ice cream (e.g. Ben & Jerry's). These foods were
displayed as would be found in an advertisement and it is
reasonable to assume that food items such as these represent a
symbolic meaning. Roper and La Niece (2009) found that the
symbolic meaning of food brands becomes more intricate during
adolescence, therefore food presented on social media becomes a
part of the identity of the uploader and conclusions can be made
about other uploaders based on their consumption practices.
Williams (2013) raises concern that food marketing through
social media services is dominated by soft drink and fast food
franchises, with further concern arising due to the direct targeting
to 13 to 17-year-olds. For example, Starbucks was one of the biggest
advertisers on Facebook in 2013 reaching many adolescent con-
sumers (Boyland & Whalen, 2015). The frequency of these brands
on display in the Instagram images of adolescents in the present
study would seem to indicate that these uploaders' were not aware,
or simply agreed with the persuasive intent of food manufacturers.
In other words the adolescents participated in the advertising.
Previous research suggests that children who recognize the
persuasive intent of commercial messages are more critical towards
advertisements, for instance by developing counter argumentation
(Verhellen, Oates, De Pelsmacker, & Dens, 2014). In our study it is
the adolescent Instagram users that promote these types of brands
by mimicking adverts and displaying food items in personalized
settings. For example, some of the images in the sample were
influenced by the “Share a Coke” campaign (Esterl, 2014). These
users thus adopted food manufacturer's existing marketing logic
while promoting the sugar-rich beverages in their own social
networks.
Assessing exposure to food marketing among adolescents is
challenging and only scarce research exists concerning the extent
and impact of food marketing to young people in online media
(Kelly, Vandevijvere, Freeman, & Jenkin, 2015). Previous research
has indicated that food brands capitalize on users' online social
networks to expand the reach and personal application of their
advertising, and that adolescents and young adults are the most
receptive to engage with such content (Freeman et al., 2014). Our
study shows that these marketing techniques seem to have made
128 C. Holmberg et al. / Appetite 99 (2016) 121e129
an impact in adolescents' online food presentations, and dietary
health promotion initiatives should thus make adolescents aware
of how they themselves contribute to food manufacturers'
persuasive marketing. The ability to comprehend and evaluate
health information from online sources is commonly defined as
eHealth literacy (Norman & Skinner, 2006). Findings from our
study imply that eHealth literacy initiatives need to take into ac-
count food marketing transmitted by the adolescents themselves in
social media.
Our study shows how adolescents communicate in social media
and the findings could therefore have implications on how to frame
teen-oriented messages in health communication practice. The
framing of health messages is important since depending on how
health-related messages are outlined they differentially affect in-
dividuals' health-related choices and behavior (Gallagher &
Updegraff, 2012; Rothman, Bartels, Wlaschin, & Salovey, 2006).
Our study indicates that food messages encapsulated in the Insta-
gram images often were positioned in specific situations and set-
tings. Therefore, the situation should not be neglected when
communicating dietary health advice to teens, as adolescents
might view food differently within different situations and
contexts.
The concern over adolescents food consumption and the place
of the commercial market in the lives, health and wellbeing of
young people has grown in recent years (Best, 2014). We should
make better use of social media to endorse countervailing messages
that promote healthy dietary habits among adolescents. Adoles-
cents increasingly communicate with visual rather than textual
messages, and use symbols and emojis to express reactions. It is
therefore important that health communicators are aware of how
to frame messages to reach the right audience. This is especially
significant as images have been seen to beneficially complement
textual-based health information (Houts, Doak, Doak, & Loscalzo,
2006).
Conducting this type of online research poses methodological
issues which are important to address. Several of the Instagram
accounts were private and were thus inaccessible to us. Our
research is therefore only based on the publicly available data and
we are unable to evaluate how representative this is of all com-
munications between adolescents on Instagram. However, there is
no reason to suppose that the private accounts should be any
different to the public ones. Another limitation is that social media
permits individuals to select how to present themselves online and
there can be a discrepancy between their online and offline iden-
tities (Blinka & Smahel, 2009). Although we chose to identify the
users based on their profile information, we cannot be certain who
operated the accounts and if they were in fact adolescents. How-
ever, this method had specific advantages because the unobtrusive
approach made it possible to investigate online dietary communi-
cation without interfering with it. This is important since studies
measuring food intake have noted frequent misreporting of foods
of low social desirability (Lissner, 2002) as adolescents generally
tend to view certain foods as “bad” or “good” (Stevenson, Doherty,
Barnett, Muldoon, & Trew, 2007).
Due to the sampling technique used in this study, it is difficult to
know if we can generalize our findings to other corpuses of Insta-
gram data. There might, for example, be an unknown bias in
initiating the sampling procedure by following a specific hashtag.
The use of hashtags are not only platform-specific but also con-
textually specific to the individual user (Highfield & Leaver, 2014).
However, by systematically selecting one food image per account,
we were able to collect a wide range of food images. This enabled us
to analyze a variety of food presentations and to generate a
multifaceted array of categories.
In conducting research into social media it is important to keep
up-to-date with the rapidly changing practices in this field. These
include changes applied by the services themselves, new platforms
being developed, and changes in the practices of users (Highfield &
Leaver, 2014). An area of future research would be to investigate the
differences between how adolescents perceive food marketing
transmitted by their peers, in online social networks, compared to
conventional online advertising. It is likely that the impact of food
marketing campaigns which utilize several digital media platforms,
and that engage individually with consumers, will be greater than
that of traditional marketing (Kelly et al., 2015).
Our study was explorative in nature, thus the findings were
mainly descriptive and aimed to set the ground for future work in
this area. Building on this research, we plan to interview a specific
group of adolescents about their online food communication in
order to explore qualitative aspects of these practices. Our study
was also linguistically bounded in nature, focusing on adolescents
who post on Instagram in the Scandinavian languages. There are
important opportunities for future research to examine other lan-
guages and possible cultural differences in the ways adolescents
portray food on social media. Recent discussion in the English
speaking media of a trend where young women post images of
themselves with high calorie foods and the hashtag #iatethis in the
caption (e.g. Davis, 2014) is one example of a practice that might
indicate a cultural difference that could be examined.
In conclusion, this study shows that food images were found in
most Instagram accounts of adolescents. Food was portrayed in
different ways but it was often centrally placed and framed in a
positive way. Images containing high calorie low nutrient food
items were framed as if they were advertisements, with brands
clearly exposed, in the context of coffee shops or other outside
eateries. Images depicting fruits and vegetables were often in zoom,
similar to images found in cook books. Considering that health
behaviors, such as food habits, can spread though social media, it
would be interesting to see if these different ways of presenting and
framing food could have an impact on how adolescents perceive
food and if it significantly affects their consumption practices.
Acknowledgments
This study was supported by grants from Formas e The Swedish
Research Council for Environment, Agricultural Sciences and
Spatial Planning (grant number 259-2012-38) and is a research
activity within EpiLife, Gothenburg's center for epidemiologic
studies (Forte 2006-1506). We would also like to affirm our respect
for Instagram users and their publically shared images which made
it possible for us to conduct this research.
References
Beaver, J. D., Lawrence, A. D., van Ditzhuijzen, J., Davis, M. H., Woods, A., &
Calder, A. J. (2006). Individual differences in reward drive predict neural re-
sponses to images of food. Journal of Neuroscience, 26(19), 5160e5166. http://
dx.doi.org/10.1523/JNEUROSCI.0350-06.2006.
Best, A. L. (2014). Youth consumers and the fast-food market. Food, Culture & Society,
17(2), 283e300. http://dx.doi.org/10.2752/175174414X13871910531980.
Bevelander, K. E., Anschutz, D. J., Creemers, D. H., Kleinjan, M., & Engels, R. C. (2013).
The role of explicit and implicit self-esteem in peer modeling of palatable food
intake: a study on social media interaction among youngsters. PLoS ONE, 8(8),
e72481. http://dx.doi.org/10.1371/journal.pone.0072481.
Bevelander, K. E., Anschutz, D. J., & Engels, R. C. (2012). The effect of a fictitious peer
on young children's choice of familiar v. unfamiliar low- and high-energy-dense
foods. British Journal of Nutrition, 108(6), 1126e1133. http://dx.doi.org/10.1017/
S0007114511006374.
Blinka, L., & Smahel, D. (2009). Fourteen is fourteen and a girl is a girl: validating the
identity of adolescent bloggers. CyberPsychology & Behavior, 12(6), 735e739.
http://dx.doi.org/10.1089/cpb.2009.0044.
Boyd, D. (2014). It's complicated: The social lives of networked teens. New Haven: Yale
University Press.
Boyland, E. J., & Whalen, R. (2015). Food advertising to children and its effects on
diet: review of recent prevalence and impact data. Pediatric Diabetes, 16(5),
 C. Holmberg et al. / Appetite 99 (2016) 121e129
331e337. http://dx.doi.org/10.1111/pedi.12278.
Brembeck, H., & Johansson, B. (2010). Foodscapes and children's bodies. Culture
Unbound, 2, 797e818. doi:10.3384/
cu.2000.1525.10242797#sthash.oZypzMJS.dpuf.
Collins, K. (2009). Watching What We Eat: The Evolution of Television Cooking Shows.
New York: Continuum.
Counihan, C., & Van Esterik, P. (2013). Food and culture: A reader (Vol. 3). New York ;
London: Routledge.
Davis, P. A. (2014). Seriously, stop pretending you ate that. Retrieved from http://
nymag.com/thecut/2014/05/seriously-stop-pretending-you-ate-that.html.
Esterl, M. (2014). ‘Share a coke' creates pop in sales. Wall Street Journal - Eastern
Edition, 264(74). B5eB5. Retrieved from http://search.ebscohost.com/login.
aspx?direct¼true&db¼buh&AN¼98520556&site¼ehost-live.
Findahl, O. (2014). Svenskarna och Internet 2014. Retrieved from https://www.iis.se/
docs/SOI2014.pdf.
Freeman, B., Kelly, B., Baur, L., Chapman, K., Chapman, S., Gill, T., et al. (2014). Digital
junk: food and beverage marketing on facebook. American Journal of Public
Health, 104(12), e56e64. http://dx.doi.org/10.2105/ajph.2014.302167.
Fritz, G. K. (2014). Social media use and adolescents: a guide for parents. The Brown
University Child and Adolescent Behavior Letter, 30(S3), IeII. http://dx.doi.org/
10.1002/cbl.20208.
Gallagher, K., & Updegraff, J. (2012). Health message framing effects on attitudes,
intentions, and behavior: a meta-analytic review. Annals of Behavioral Medicine,
43(1), 101e116. http://dx.doi.org/10.1007/s12160-011-9308-7.
Goffman, E. (1959). The presentation of self in everyday life. New York: Anchor Books
Doubleday.
Highfield, T., & Leaver, T. (2014). A methodology for mapping instagram hashtags.
Hookway, N. (2008). ‘Entering the blogosphere': some strategies for using blogs in
social research. Qualitative Research, 8(1), 91e113. http://dx.doi.org/10.1177/
1468794107085298.
Houts, P. S., Doak, C. C., Doak, L. G., & Loscalzo, M. J. (2006). The role of pictures in
improving health communication: a review of research on attention, compre-
hension, recall, and adherence. Patient Education and Counseling, 61(2),
173e190. http://dx.doi.org/10.1016/j.pec.2005.05.004.
Hu, Y., Manikonda, L., & Kambhampati, S. (2014). What we instagram: a first
analysis of instagram photo content and user types. In Paper presented at the
international AAAI conference on weblogs and social media.
Instagram. (2014a). Privacy policy: Section 3, sharing of your information; Parties with
whom you may choose to share your user content. Retrieved from http://
instagram.com/about/legal/privacy/.
Instagram. (2014b). Stats. Retrieved from https://instagram.com/press/.
Instagram. (2014c). Terms of use. Retrieved from http://instagram.com/about/legal/
terms/.
Kelly, B., Vandevijvere, S., Freeman, B., & Jenkin, G. (2015). New media but same old
tricks: food marketing to children in the digital age. Current Obesity Reports,
4(1), 37e45. http://dx.doi.org/10.1007/s13679-014-0128-5.
Killgore, W. D. S., & Yurgelun-Todd, D. A. (2006). Affect modulates appetite-related
brain activity to images of food. International Journal of Eating Disorders, 39(5),
357e363. http://dx.doi.org/10.1002/eat.20240.
Krippendorff, K. (2004). Content analysis. An introduction to its methodology. Cali-
fornia: Sage Publications.
Leeuwen, V. T., & Carey, J. (2001). The handbook of visual analysis. London, England:
SAGE Publications Ltd.
Lissner, L. (2002). Measuring food intake in studies of obesity. Public Health Nutri-
tion, 5(6a), 889e892. http://dx.doi.org/10.1079/PHN2002388.
Lynch, M. (2010). From food to fuel: perceptions of exercise and food in a com-
munity of food bloggers. Health Education Journal, 71(1), 72e79. http://
dx.doi.org/10.1177/0017896910386284.
Markham, A., & Buchanan, E. (2012). Ethical decision-making and internet research:
Recommendations from the AoIR ethics working committee (Version 2.0).
Retrieved from http://aoir.org/reports/ethics2.pdf.
McBride, D. L. (2011). Risks and benefits of social media for children and adoles-
cents. Journal of Pediatric Nursing, 26(5), 498e499. http://dx.doi.org/10.1016/j.
129
pedn.2011.05.001.
Mejova, Y., Haddadi, H., Noulas, A., & Weber, I. (2015). #FoodPorn: Obesity patterns in
culinary interactions. arXiv:1503.01546v3. Retrieved from http://arxiv.org/abs/
1503.01546v3.
Moreno, M. A., Goniu, N., Moreno, P. S., & Diekema, D. (2013). Ethics of social media
research: common concerns and practical considerations. Cyberpsychology,
Behavior, and Social Networking, 16(9), 708e713. http://dx.doi.org/10.1089/
cyber.2012.0334.
Neely, E., Walton, M., & Stephens, C. (2014). Young people's food practices and social
relationships. A thematic synthesis. Appetite, 82(0), 50e60. http://dx.doi.org/
10.1016/j.appet.2014.07.005.
Norman, D. C., & Skinner, A. H. (2006). eHealth literacy: essential skills for consumer
health in a networked world. Journal of Medical Internet Research, 8(2), e9.
http://dx.doi.org/10.2196/jmir.8.2.e9.
Pew Research Center. (2015). Teens, social media & technology overview 2015.
Retrieved from http://www.pewinternet.org/files/2015/04/PI_TeensandTech_
Update2015_0409151.pdf.
Riffe, D., Lacy, S., & Fico, F. (2005). Analyzing media messages: Using quantitative
content analysis in research. London: Lawrence Erlbaum.
Roper, S., & La Niece, C. (2009). The importance of brands in the lunch-box choices
of low-income British school children. Journal of Consumer Behaviour, 8(2/3),
84e99. http://dx.doi.org/10.1002/cb.275.
Rothman, A. J., Bartels, R. D., Wlaschin, J., & Salovey, P. (2006). The strategic use of
gain- and loss-framed messages to promote healthy behavior: how theory can
inform practice. Journal of Communication, 56, S202eS220. http://dx.doi.org/
10.1111/j.1460-2466.2006.00290.x.
Schienle, A., Scha €fer, A., Hermann, A., & Vaitl, D. (2009). Binge-eating disorder:
reward sensitivity and brain activation to images of food. Biological Psychiatry,
65(8), 654e661. http://dx.doi.org/10.1016/j.biopsych.2008.09.028.
Schlesselman-Tarango, G. (2013). Searchable signatures: context and the struggle
for recognition. Information Technology & Libraries, 32(3), 5e19. Retrieved from
http://search.ebscohost.com/login.aspx?
direct¼true&db¼afh&AN¼90292134&site¼ehost-live.
Schneider, E. P., McGovern, E. E., Lynch, C. L., & Brown, L. S. (2013). Do food blogs
serve as a source of nutritionally balanced recipes? an analysis of 6 popular
food blogs. Journal of Nutrition Education and Behavior, 45(6), 696e700. http://
dx.doi.org/10.1016/j.jneb.2013.07.002.
Sharma, S. S., & Choudhury, M. D. (2015). Measuring and characterizing nutritional
information of food and ingestion content in instagram. In Paper presented at
the proceedings of the 24th international conference on World Wide Web Com-
panion, Florence, Italy.
Simunaniemi, A.-M., Sandberg, H., Andersson, A., & Nydahl, M. (2011). Laypeople
blog about fruit and vegetables for self-expression and dietary influence. Health
Communication, 26(7), 621e630. http://dx.doi.org/10.1080/
10410236.2011.561520.
Stevenson, C., Doherty, G., Barnett, J., Muldoon, O. T., & Trew, K. (2007). Adolescents'
views of food and eating: identifying barriers to healthy eating. Journal of
Adolescence, 30(3), 417e434. http://dx.doi.org/10.1016/j.adolescence.2006.04.
005.
Sun, S. (2011). Meta-analysis of Cohen's Kappa. Health Services and Outcomes
Research Methodology, 11(3e4), 145e163. http://dx.doi.org/10.1007/s10742-011-
0077-3.
Swedish Media Council. (2015). Ungar & Medier 2015. Fakta Om Barns och Ungas
Anva€ndning och Upplevelser av Medier. Retrieved from http://www.
statensmedierad.se/upload/_pdf/Ungarochmedier2015_ver1.pdf.
Verhellen, Y., Oates, C., De Pelsmacker, P., & Dens, N. (2014). Children's responses to
traditional versus hybrid advertising formats: the moderating role of persua-
sion knowledge. Journal of Consumer Policy, 37(2), 235e255. http://dx.doi.org/
10.1007/s10603-014-9257-1.
Williams, S. (2013). Action needed to combat food and drink companies' social
media marketing to adolescents. Perspectives in Public Health, 133(3), 146e147.
http://dx.doi.org/10.1177/1757913913484871.