Narrative Elements in Social Networks. Which Aspects Determine the Popularity of Photos on Instagram?

Table of Content

1. Executive Summary

2.. Introduction
2.1. Outline

3. Methodology
3.1. Research Design
3.2. Data Collection Method

4... Research Questions
4.1. Narrative elements based research questions
4.2. Graphical elements based research question
4.3. Content based research question
4.4. Further research question

5.. Data Analysis and Findings
5.1. Narrative elements based research questions
5.2. Graphical elements based research question
5.3. Content based research question
5.4. Further research question

6.. Limitations
6.1. Data gathering related limitations
6.2. Further research question

7.. Conclusion
7.1. Managerial Implications

References

Appendix A

Appendix B

Appendix C

l. Executive Summary

The rapidly growing social network Instagram was founded in 2010 and offers users he possibility to share photos with their friends and the whole Instagram community. The community can rate these photos in their popularity by using the tools Likes and Comments. These measures of popularity reflect how the Instagram community perceives the uploaded photos. Besides private persons also businesses can upload photos of their products and use Instagram as a marketing channel. Therefore, for marketers it might be interesting to find out which factors influence the popularity of photos. This is exactly the purpose of the paper at hand. The main dependent variables of the research are the number of Likes and the number of Comments of each photo. The data was gathered by using the tool Statigr.am, which contains all photos that are uploaded on Instagram. The final dataset contained 209 observations that belonged to the two product categories Frozen Yoghurt and Cigarette. After the collection period, statistical tests were executed using descriptive statistics. The main independent variables belonged to the narrative elements that are described by Kress & van Leeuwen, 2006. The research showed that the type of narrative elements used has an influence on the popularity of photos and therefore this finding agrees with the expectations that were based on the literature. Furthermore, the research showed that the content of photos has an influence on the popularity. This was analyzed by the two product categories, of which the photos that contained the product Frozen Yoghurt had higher popularity scores. Additionally, the research showed that the number of Hashtags, which the uploading person can determine, has a strong influence on the popularity. In the sample at hand, photos with more Hashtags had higher popularity measures than photos with less Hashtags. Finally, the research rejected the assumption that graphical elements of photos such as camera angle and distance, as well as the use of filters and different light conditions has an influence on the popularity. In the sample at hand no clear relationship was found. Especially, the findings regarding the narrative elements and the Hashtags allow firms to design their product photos in the most appealing.

All things considered, this paper offers findings that lead to several managerial implications for companies that want to use Instagram as a marketing channel, but in addition unsolved problems with ideas for further research.

2. Introduction

In the last decade the Internet has grown rapidly and became an important part of a lot of peoples everyday life. Due to the development of the Internet many new companies were founded, which provided completely new services. Most of these services were not even imaginable before. These new companies allow the people to purse new activities online, for instance online-shopping, information search and social networking. Especially, the social networking aspect was regarded as a revolutionary idea. Companies such as Facebook, MySpace and Twitter emerged through the social network trend. They connected millions of people all over the world, but also created new methods for measuring “popularity and acceptance”. The most famous of these measures are probably the “Likes” on Facebook and the “Tweets” on Twitter. (Bakardjieva, 2013). Besides the Pioneers of social networking, numerous other social networks were founded. For instance, in 2010 K. Systrom and M. Krieger founded Instagram. Although, Facebook INC bought Instagram in 2012, it is still operating mainly as a separate social network. Instagram focuses on photo sharing and is linked to the Internet and especially Smartphone trend. After registration, every user can upload photos, which can be seen by friends, called Followers, but as well by all other Instagram users and even not registered users. These photos can be “Liked and Commented”. These two aspects can be seen as the main measures for popularity and acceptance of a photo. (Facebook INC, 2013). Therefore, the goal of this research is to determine aspects that have a positive influence on the popularity of a photo on Instagram, which is expressed in the number of Likes and Comments of each photo. Particularly, this research approaches the idea that there is a connection between the measures of popularity and the presence or absence of a narrative process in the picture (Kress & van Leeuwen, 2006). Furthermore, this paper analyses some further factors, which might have an influence on the popularity, such as graphical and content related aspects.

2.1 Outline

The purpose of this paper is to determine factors, which determine the popularity of photos on Instagram, firstly regarding the narrative process and secondly regarding graphical, content related and further factors. This is done first by describing the research design and data collection methods used and second by developing and elaborating on five research questions. These are formulated to conduct research on the purpose statement. Afterwards the design of the different variables is described. Further, the data analysis and findings resulting from the research are shown. These findings come with limitations that are presented. The conclusion points out the main findings and relates them back to the purpose statement. Finally t managerial implications of the research are suggested.

3. Methodology

3.1 Research Design

The first issue that had to be addressed when attempting to answer the purpose statement was to find an appropriate research design, taking into account the timeframe and the resources available for the study. For investigating the content related factors, the researcher decided to choose only photos, which display either a Frozen Yoghurt or a Cigarette. These two products were chosen because on the one hand they are both consumer goods and therefore comparable in their perception and on the other hand very opposing, because Frozen Yoghurts are considered as a new and healthy and Cigarettes as an old and unhealthy product. Due to the two specific products chosen, this research is unique and therefore it was necessary to gather primary data. Although, there is a lot of existing research regarding narrative and graphical elements as well as about photos in social networks, these would have lacked applicability on the purpose statement. A large sample of about 200 photos was needed to properly apply essential statistical tests and to have a roughly normal distribution. Concerning the different possible studies, a descriptive study with quantitative data output seemed to be the most appropriate. An exploratory study was not appropriate, because no generalization about the population could be drawn. Further, a causal study would be beyond the scope of this research. Since the research team only addressed the sample once, they worked with a cross-sectional design (Sekaran & Roger, 2009).

In order to collect the data needed, the research team had to choose an appropriate instrument for data gathering. Since the goal was to gather data from already uploaded photos it was most appropriate to use an observation study. This implies that one directly observe and code the uploaded photos, instead of asking the people in a questionnaire or interview about the photos that they uploaded. The main advantages are cost and time efficiency as well as more accurate results with less bias than questionnaires and interviews. The data was collected by the usage of the tool “statigr.am”. This tool allows accessing all photos from the target platform Instagram and sorting them by “Hashtags”, which can be described as “ a word or phrase preceded by a hash mark (#), used within a message to identify a keyword or topic of interest and facilitate a search for it” (Dictionary.reference.com, 2013)

The research determined Hashtags, which are closely related to the two target product categories and then randomly chose photos. These were then coded into the required variables. While choosing the photos randomly, it was necessary to ensure that the photo was on Instagram for at least twelve hours, so that the popularity, namely the number of Likes and comments, had enough time to develop. Nevertheless, the sample can be regarded as a convenience sample. The final sample consisted of 209 valid observations of which 104 were assigned to the product category Cigarette and 105 to the product category Frozen Yoghurt.

4. Research Questions

To investigate the problem statement, there is the need to formulate research questions, which can help to generate findings that are relatable to the problem statement.

4.1 Narrative elements-based research questions

Research question I: Does the popularity of a photo depend on the presence or absence of narrative elements in the photo?

The first research question aimed to measure if there is a general relationship between the popularity of a photo and the presence and absence of narrative elements in the picture. The popularity was measured by the number of Likes and Comments for the photo. To determine the presence of narrative elements, one had to specify the presence or absence of three different elements in every photo, namely the Actor, the Transaction and the Goal. (Sharpe, De Veaux, & Velleman, 2011).

Research question II: Does the popularity of a photo depend on the type of the narrative elements in the photo?

The second research question attempted to measure if there is a difference in the popularity of a photo between photos with different types of narrative elements. The narrative elements were the same as in research question I. However, the Actor in every photo is classified into four specific groups, namely “Human”, “Object”, “Other” and “No Actor”. The transaction was grouped into “Eye gaze, “Body limb”, “Other” and “No transaction” The goal was grouped into “Human”, “Object”, “Other” and “No Goal”. The collected data showed that for some groups there was no or almost no data collected, so that these groups were disregarded in the analysis.

4.2. Graphical-elements based research question

Research Question III: Does the popularity of a photo depend on the use of graphical elements in the photo?

This research question investigated whether specific graphical elements in the photo have an influence on the popularity on Instagram. There were five graphical elements that were collected, namely Light, Filter, Focus, Distance and Angle. Intuitively, the researcher expected that the presence of these graphical in a photo

Research Question IV: Does the popularity of a photo depend on the type of product used?

This research question attempted to determine if there is a difference in the popularity of photos for different products, which represent the content delivered by the picture. The researcher collected photos with two different products, Cigarettes and Frozen Yoghurts.

4.4 Further research question

Research Question V: Does the popularity of a photo depend on the number of Hashtags for this photo?

The last research question aimed to find out if the number of Hashtags has an influence on the popularity. Intuitively, one expected a positive relationship, since more Hashtags probably lead to the fact that more people find an access the photo and therefore the photo might generate a higher number of Likes and Comments than photos with less Hashtags.

5. Data Analysis and Findings

Before investigating the research questions and testing the hypotheses, it was crucial to check whether specific scientific requirements are met. For a valid research, the dependent variables should be roughly normally distributed. The main dependent variables for this research were the two measures of popularity, the number of likes and the number of comments.

For checking the normality requirement a histogram was most appropriate. The histogram showed that the distribution is skewed and had some huge outliers, which made intuitively sense for the number of likes, since it differs significantly across different photos and users. To solve this problem it was appropriate to use a logarithmic transformation of the dependant variable number of likes. Therefore one applied the natural logarithms on the

Abbildung in dieser Leseprobe nicht enthalten

dependent variable Likes and transferred it to the new variable “LogLikes”. After the transformation, the distribution became roughly normal and therefore all requirements were met. (Sharpe, De Veaux, & Velleman, 2011)

The second dependant variable was the number of comments. As for the number of likes the original distribution was skewed, so that a logarithmic transformation was required. After transferring it in the same way as the number of likes, the new variable was called LogComments. The distribution of this variable was roughly normal so that the requirement was met. Nevertheless, the logarithmic transformation created a new problem for the second dependant variable. Since 97 of 209 photos had zero comments, the new dependant variable only accounts for 112 out of 209 photos. The reason was that it is mathematically not possible to apply the natural logarithms on zeros. Consequently, this technical issue limits the representativeness of the second dependant variable. However since both dependant variables attempted to measure the same, namely the popularity, both are used and the limitations were evaluated afterwards. Furthermore, for a regression analysis it was crucial to implement control variables that account for variation of the dependent variables, which were not of main interest for the research. For this research the variable Followers was intuitively a control variable, since more followers most likely lead to a higher number of likes and comments. During the regression analysis this relationship was proven.

5.1 Narrative elements-based research questions

Research question I: Does the popularity of a photo depend on the presence or absence of narrative elements in photo?

Ho: Neither presence nor absence of narrative elements has an influence on the popularity

Ho: all μ=0

Ha: Either presence or absence of one or more narrative elements has an influence on the popularity

Ha: at least one μ^0

For the analysis of the first research question it was necessary to perform two tests, since the research determined two dependant variables, which together describe the popularity of a photo. Therefore, each test contained one dependant variable and additionally three independent variables, which were presence of actor, transaction and goal. For both tests a linear regression with three independent variables was created. Since the independent variables were nominal it was necessary to create dummy variables for a linear regression. However since the variables only measured the presence or absence of narrative elements they can already be considered as dummies. This implies that one can run the regression without transforming the independent variables. Additionally, both regressions contained the control variable followings. According to Kress & van Leeuwen, 2006 the presence of narrative elements influences the way people perceive a photo and therefore one can expect to have a significant influence on the popularity between photos with different narrative elements. Nevertheless, the outputs showed that the p-values for the three independent variables were clearly higher than five per cent in both regression models. Therefore, one failed to reject the null hypothesis for all three independent variables.

[...]