Advertisement

How to Compare Different Social Media: A Conceptual and Technical Framework

  • Jakob Linaa JensenEmail author
  • Peter B. Vahlstrup
  • Anja Bechmann
Living reference work entry

Abstract

Social media research has gained traction during the last 10 years within Internet research and digital sociology. However, due to methodological and technical difficulties studies have mainly focused on analyzing only one platform per study (often Twitter or Facebook), especially when the study involves analysis of large public or private data streams (e.g., Bechmann. J Media Bus Stud 11(1):21–38, 2013; Bechmann. Managing the interoperable self. In: Bechmann A, Lomborg S. The ubiquitous internet: user and industry perspectives. Routledge, New York, pp 54–73, 2015; Boyd and Marwick. Social privacy in networked publics: teens’ attitudes, practices, and strategies. In: Proceedings of a decade in internet time, 21–24 September 2011, University of Oxford, 2011; Fernandes et al. Mass Commun Soc 13(5):653–675, 2010; Marichal. First Monday, 12(2), 2013; Lotan et al. Int J Commun 5:1375–1405, 2011; Wu et al. Who says what to whom on Twitter. In: Proceedings of the international World Wide Web conference (WWW 2011), pp 705–714, 2011; Bruns and Burgess. J Stud 13:801–814, 2012). The fact that such data streams are accessed through different application programming interfaces (APIs) that have their own different logics means that complexity increases at a technical level.

However, this chapter argues that addressing such technical complexities also requires looking at issues at the sociological and conceptual level. How do different social media, for instance, Facebook, Twitter, and Instagram, fundamentally differ? By addressing not only the different social and relational logics but also linking them to processes of data retrieval and analysis, the chapter aims to contribute with new insights into the fundamental character of two seemingly related social media and address questions which need to be posed in order to make solid and comparative academic analyses in the future.

More specifically, we develop a framework for the analysis of use and relations across social media, combining a theoretical, a conceptual, a methodological, and a technical approach. We identify challenges and suggest a specific technical implementation that we in the end evaluate. Technically, we focus on Facebook, Instagram, and to a certain extent Twitter, but we argue that our framework can be easily expanded to encompass other social media. The main argument in the chapter is that bridging is not an easy task as several challenges occur on different levels that the researchers need to account for in greater details. First, in the technical infrastructure and/or database structure relations are simple but represent many different complex sociological relationships and interaction types. Second, the elements of social media as accounted for need to be translated socially. Third, when we study user-centric social media use, we increase complexity in at least two different dimensions, we propose a mixed method design and we propose a cross-service approach.

Keywords

Sociology Social Network Analysis Meta-Methodological Multimethods 

Introduction

Social media has become an important focus of research within the fields of humanities and social sciences during the last 10 years. This has raised important methodological questions. For instance, how do we study a complex and ever-changing phenomenon of social media? How do we study users, interactions, and overall patterns? And is it possible to compare social media to other, more well-known media formats, or make comparative analyses across various social media platforms?

The latter question is to be answered in this chapter. So far social media studies have mainly focused on analyzing only one platform per study (most often Twitter followed by Facebook). This has particularly been the case in quantitative studies involving analysis of large public or private data streams (e.g., Bechmann 2013, 2015; Boyd and Marwick 2011; Fernandes et al. 2010; Marichal 2013; Lotan et al. 2011; Wu et al. 2011; Bruns and Burgess 2012). Good reasons for this rather limited approach are methodological and technical difficulties. For instance, the fact that such data streams are accessed through different application programming interfaces (APIs) that have their own different logics means that complexity increases at a technical level. Further, data from the APIs might have different formats and carry different implicit meanings, making direct statistical comparisons difficult or impossible.

However, this chapter argues that addressing such technical and statistical complexities also requires looking at issues at the conceptual and social level. How do different social media platforms, for instance, Facebook, Twitter, and Instagram, fundamentally differ? What are the main social and interactional logics across platforms, and how do different technical architectures affect user dynamics?

More specifically, in this chapter, a framework is established for the analysis of use and relations across social media, combining a theoretical, a conceptual, a methodological, and a technical approach. We identify challenges and suggest a specific technical implementation that we in the end evaluate. Specifically, we focus on Facebook, Instagram, and to a certain extent Twitter, but we argue that our framework can be easily expanded to encompass other social media.

By addressing not only the different social and relational logics but also linking them to processes of data retrieval and analysis, this chapter contributes with new insights into the fundamental character of two seemingly related social media platforms and provide guidelines and inspiration for those who wish to do their own comparative analyses of social media platforms.

The Outline

The chapter starts by discussing conceptual and methodological frameworks in existing studies on social media. Then follows a theoretical approach to studying content and interactions across social media platforms, based on a symbolic interactionist inspired approach claiming that social interaction is the key element of social relations and social life in general. Next, the theoretical frame is turned into a conceptualization of key features of various social media. Then follows a proposed methodological approach based on such conceptual definitions. Finally, a technical approach is shown and discussed, encompassing the three other approaches. The conclusion has a reflection on the challenges translating social concepts into technical solutions and vice versa and directions for future developments are sketched.

Existing Studies of Use Across Social Media Platforms

In this chapter, social media platforms and social network services will be used as synonyms. Boyd and Ellison (2007) in their original definition described what they call social network sites as

web-based services that allow individuals to (1) construct a public or semi-public profile within a bounded system, (2) articulate a list of other users with whom they share a connection, and (3) view and traverse their list of connections….

Even though this is perhaps the most cited definition of social media services, it still fails to address the scope of this chapter – the nature of cross social media use. First and foremost, social network services are no longer entirely web-based as most contemporary use takes place through apps on mobile phones or tablets; therefore, we suggest to replace “sites” with services. Secondly, various services tend to remediate one another. Facebook’s initial focus on personal profiles and networking has been supplemented by content sharing, of images, videos, and news. In fact, Facebook has become a meta-medium, encompassing a multitude of content forms and media types (Jensen and Tække 2013). On the other hand, content sharing services like YouTube, Flickr, and Instagram have added personal profiling or networking properties to their initial layout. It becomes increasingly difficult to clearly distinguish social network services from other kinds of social media. Ellison and Boyd (2013) acknowledge such developments in their updated definition from 2013, highlighting the increasing role of content and the interaction between content sharing and personal relationships.

For some time, much empirical research on social media has focused on Twitter, due to the fact that hashtag streams have the status of public data and are thus easier to retrieve and analyze without informed user consent. Further, Twitter has allowed for extensive analyses through technologies like TwapperKeeper and DMI-TCAT but has recently limited such possibilities. Facebook, on the other hand, has had a more closed approach towards data retrieval as less data streams are public and therefore require informed user consent in order to retrieve (Bechmann and Vahlstrup 2015).

As already stated, the majority of academic studies have mainly focused on analyzing one platform or social network service per study (e.g., Boyd and Marwick 2011; Bechmann 2015). One reason is that data from different social media are located within different APIs, creating technical complexity. A study of a Norwegian election (Enli and Skogerbø 2013) is one among many examples of studies of time-limited phenomena from a cross-platform perspective, but the focus is not on the methodological challenges in cross-platform analysis. Other examples are Taneja et al. (2012) who studies use across media platforms. However, they rely on observation studies in user-defined repertoires and thus focus on media consumption without an interest in the actual data, social relations, and content use. Hanna et al. (2011) establish a framework for understanding the ecosystem and interplay of various social media but fail to propose more detailed analytical measures.

More relevant in our context, Becker et al. (2012) establish a framework for retrieving event-related social media content across social media YouTube, Flickr, and Twitter but do not include Facebook or Instagram in their approach. Courtois and Merchant (2013) are among the first to describe and use APIs as tools for data retrieval and analysis within social research. Their approach is based on YouTube and does not cross digital platforms.

Most studies have taken either a purely conceptual or a purely technical approach and few have presented truly integrated approaches across social media. The present chapter might include features from previous research, but it is an attempt to establish a mixed method framework for studying interactions across social media platforms, combining a sociological understanding of concepts with specific discussions of data retrieval, analysis, and adherent technical issues.

Theoretical Approach

There has been a diversity of sociological perspectives applied to the analysis of computer-mediated communication. In the early years, a common approach was the postmodern, focusing on the fluent identity online and the possibility of a break with existing social and physical relations (Turkle 1995). However, much of this literature focused on the identity construction and personal performance rather than social relations. Such ideas have also guided some expectations on behalf of social media (see for instance Fraser and Dutta 2010). In general, utopian ideas of a radical new social reality with no borders, races, or social divisions proved to be unrealistic. For instance, specific analyses of social network services have demonstrated how most people prefer to interact with whom they already know (Jensen and Scott Sørensen 2013).

Recognizing the importance of identity, we claim that relations are the core phenomena of social life as they are not only the constituting feature of sociality but also define identity. No human is an island but exists in relation to other humans (Burkitt 2008). We draw on the classic sociological approach of symbolic interactionism (Blumer 1980) and claim that self-presentation and identity construction is primarily maintained through interactions with other users (Gergen 2009).

Another related framework that has been widely used to analyze computer-mediated communication is actor-network theory (ANT). However, ANT studies (see, for instance, Latour 2011; Law 1991) focus on the negotiations/translations happening among actors (humans as well as technology) often with a developmental perspective, not on the human social relations as such. Similar focus is found in social construction of technology (SCOT) perspective that has been fruitful in studies of uses and construction of technologies but not with a main focus on social relations either. Still, these theoretical fields as well as cyborg theories (Haraway 1991 and Clark 2011) share an understanding of technology as integrated with the human being per see that is important to our framework. Integration between human and technology is the premise for cross-social media studies, albeit the focus in our perspective is on the social relations and communication. When claiming that we do not need to focus on the different platforms, but on the social relations across platforms and services, we move our focus away from an anticipation of the platforms’ primary meaning as standalone and exterior technology. Instead, it is assumed that people focus more on what they do on the platforms rather than where they do it – be it online, offline, on specific platforms, or across media formats.

Thus, measuring identity and social relations, one should focus on the interactions among users. Following such a position, what presently characterizes social media like Facebook, Twitter, and Instagram is that even though identity might still be constructed through self-presentation and performance, for instance, by the often delicately elaborated personal profiles, the main social feature is the interactions among users. A coherent analytical framework must focus on relations, which take place through interactions, like friending, connecting, replying, retweeting, and sharing.

Conceptual Approach

First, the focus here is on users’ latent communication. Here, the point of departure is taken in Berelson’s (1952) distinction between manifest and latent communication. As one cannot grasp latent meanings of communication directly but must rely on manifest communication and then interpret the latent meanings, the conceptual approach has to be based on manifest communication. Only visible communication acts are grasped here, as the approach is data driven by nature.

Second, the present concept is user-centric approach rather than media-centric approach. The users and their interactions across media are in focus, rather than the medium itself. What is interesting is exactly to what extent usage resembles and differs across social media platforms. By focusing on users and not least their interactions and relationships with other users, we are able to track and follow interactions across media and platforms rather than, as many existing studies, focusing on content only within one platform.

Third, by analyzing across social media, even though one might be not able to access the total social context facing users, an understanding of the media ecology might be established (Gencarelli 2006).

Now, how can such theoretical understandings of relations and sociality be applied through specific concepts in studies across social media platforms?

For instance, relations differ on Facebook and Twitter. Whereas Facebook was originally dedicated to persons and networking, Twitter was basically a kind of microblogging service with limited possibilities of personal profiling. Thus, the different technical (and social) architectures create different user environments and probably also different patterns of actions.

Some more specific examples: the typical features of Facebook are the status update, the friend, and the “like.” The typical features of Instagram are the image and the heart. The typical features of Twitter are the tweet, the hashtag, and the follower. Now the question is not only whether these features are comparable as such. We also face the challenge that Facebook, for instance, has imported the hashtag, but that it is used differently and more sparsely than on Twitter. On the other hand, the Twitter feature to make a tweet a “favorite” is used differently by early tweeters than by latecomers. Earlier, the “favorite” was used as a tool for later retrieval or reading where it is now typically used equivalent to a Facebook “like.” Even though features are technically comparable across social network services, their meaning might differ. The major challenges might be conceptual first, technical later.

The challenge in establishing general concepts for studying social media use is to measure the socially relevant features, acknowledging the differences in sociality and technical architecture on the various services. Despite Facebook and Twitter copying one another (Facebook has, for instance, copied the hashtag, and Twitter has installed a “like function” to replace their “favorite function”), they still remain fundamentally different, in architecture and in social affordances.

Kietzmann et al. (2011) identify seven building blocks of social media: identity, conversations, sharing, presence, relationships, reputation, and groups. However, identity, presence, and reputation are abstract phenomena, which are hard to study directly and must be investigated through more specific (and observable) measures. Such latent phenomena must be studied through manifest interactions. For social network services, there are several basic elements of interactions. Central are of course the persons (profiles) who relate to other persons through content (updates, posts, likes, things shared, etc.). The point is that relation are content-driven, all relations are signified by some kind of content.

However, identity, presence, and reputation are abstract phenomena that are hard to study directly and must be investigated through more specific (and observable) measures. We believe that such latent phenomena must be studied through manifest interactions. For social media, there are several basic elements of interactions. Central are of course the persons (profiles) who relate to other persons through content (updates, posts, likes, things shared, etc.). The point is that relation are content-driven, all relations are signified by some kind of content.

Acknowledging the close interplay, for practical purposes, it is useful to distinguish between relations and content. In developing the analytical framework, there are at least six forms of interaction which run across the three social network services, albeit their specific meaning and social implications might differ considerably. The first three are: central content unit (CCU), acknowledgment, and hashtag address content, and the latter three are: dissemination, conceptual relation, and social relation address relations.

CCU is the core content feature of each service, which the interactions are centered around. For Twitter, it is the tweet; for Instagram, the media object (most often an image or a video); and for Facebook, it is the status update. By acknowledgment refers to a form of content expressing other relationship to or acknowledgment of the CCU. For Twitter, it is the favorite; for Instagram, the heart; and for Facebook, the like. Hashtag is included here as something in between. Where Twitter and Instagram are born with hashtagging, Facebook included it in 2013, creating the above-mentioned ambivalences.

Dissemination is basically about passing on content. By sharing on Facebook or retweeting on Twitter, one creates relations with fellow users who are presented for the shared content. Sharing (or retweeting) is also about positioning one’s identity in relation to the other. Interestingly, Instagram lacks this function at the time of writing. Finally, we distinguish between content and social relations. The former denotes the social aspect of content sharing; how specific content, for instance, shares and comments contribute to re(shape) relations between users. The social relation takes place as friendship or following. It must be observed that social relations differ: on Twitter and Instagram, they are asymmetric: one can follow non-followers. On Facebook, they are by default mutual/bidirectional: people need to recognize each other as friends. Again, Facebook is remediating other social media as they have now included followers as a supplement to friendships.

This catalogue of social elements aims to include central and relevant theoretical aspects but is not necessarily exhaustive. This framework is rooted in a broad sociological focus taking its point of departure in the symbolic interactionist tradition focused only on interactions and relations. Frameworks more specifically rooted in aesthetics or psychology might have a stronger focus on design, context, and in-depth analysis of user profiles and performances.

Methodological Approach

Moving from a media-centric to a user-centric study of social media usage by analyzing use across platforms does not only leave us with theoretical and conceptual challenges. It also provides us with methodological challenges and calls for practical decisions in the research design that have certain implications for the way we collect data. This section will discuss methodological decisions and challenges when we move from a media-centric to a user-centric perspective when analyzing usage across social media instead of usage on one platform.

In constructing a generic method for analyzing social media usage across platforms that will be implemented in a technical solution, we must return to central issues of how we implement our discussion on usage. What exactly are the different types of usage of social media that such a method might cover, and what are the associated challenges and choices connected to these?

When measuring usage, one should not only look at purely data traces as a representation of the social individual but also taking into consideration the context of the user and, for instance, the sense-making of and motivation for usage by the individuals communicating. Such definition strongly encourages a mixed method design in which both the various data traces and the sense-making and motivations can be collected.

As discussed earlier, a good approach is to interpret latent aspects of relations and social life based on the manifest data collected. When this distinction is applied in API-based research, it is wrong to assume that API data collection is purely a quantitative method in contrast to interview and observations as qualitative methods. Instead API data collection can both be used as an ethnographic (Anderson et al. 2009) methodological tool to explore deep and rich descriptions of usage and/or as a quantitative methodological tool to describe broad usage patterns on a general population by collecting and analyzing public and private streams (see also Bechmann and Klausen 2015). Still, such methodological tools cannot stand alone and must be integrated with other approaches such as interviews, diaries, or observations in order to combine data traces with prompts on why such traces exist thereby contextualizing usage in a sense-making perspective (Bechmann and Vahlstrup 2015).

When studying usage across social media and not only data across social media, one might consider that not all communication is provided with hashtags, neither on Twitter nor Instagram, and especially not on Facebook. This leaves us not only with a mixed method design on the tool level (API data, interviews, observation) but also within the API setup choices. In order to collect the total usage, one must not only rely on public streams made available through hashtags but also the private streams of the users. The private streams allow us to study data and relations that users have not hashtagged and thereby made visible and accessible through hashtag analysis.

Such challenges call for a methodological construct in which both quantitative and qualitative scopes are possible and a collection of both public and private streams are made available. Still, capturing usages in a contextual way are still a challenge for all approaches including classical nondigital methods such as physical observation studies. In data-enriched mixed methods studies, this is not different. There are contextual markers such as GPS coordinates, time codes, platform information, frequency of communication, and social relations, but these markers only provide hints of the actual physical and social context. Diaries, prompts, or interviews can unfold the context, but again these are also flawed with delay challenges and interpretative elements if one operates with an understanding of capturing context in an “objective” matter. Studying use in context is an appealing ideal but not realistic with the present methodological toolbox. Nonetheless, a nonpositivistic perspective with more sources that point to usage will provide a more nuanced understanding of the usage in context.

The suggested framework is generic but takes into consideration that not only does the platforms contain sociological differences. In a user-centric perspective, they are also used for very different purposes. There are at least three aspects that can characterize differences in communication on social media: (a) timely differences, (b) spatial differences, and (c) relational differences.

Sociologically, social media are used for both planned events and spontaneous communication, for events that have a short time span and for communication that stretches over a period without a natural isolated frame (timely), for events and communication that are attached to a physical place and communication that are not (spatial), and for communication between defined social relations (within organizations, event participants, or among friends) or communication that have no predefined social relations except service dependent (the subscribers of a Facebook group or page, followers of Twitter, or Instagram hashtags).

Why is this important for the methodological construction of a generic framework? It is important, because if a generic tool is to be implemented technically, we also collect noise depending on the specific setting that we want to study. Collecting data for communication connected to a specific event through hashtag-streams would, for instance, create a lot of noise in the analysis if the same hashtag has been used for other events or communication. Collecting communication for a research question without no natural time period would mean that the setup would need to be large enough technically to encompass the entire stream. Even though this is possible, there are large limitations in the APIs in terms of unlimited data access (Lomborg and Bechmann 2014) and also obvious research design issues apart from the technical limitations that need to be solved beforehand.

Last but not least, the methodological framework needs to account for ethical aspects of the extensive data collection and interpretation of the use. Computational social science studies might be approved and legal in accordance to national legislation, but this does not make it ethically solid. One of the main analytical challenges in cross usage social media studies is the identification of individuals across platforms in order to understand and analyze social relations and interaction that is the scope of this proposed framework. Such identification is obvious in qualitative studies, but in studies with a quantitative research interest, the analysis of cross-identity most be integrated in a technical bridging construct. The next section will account for a possible development of such a construct, but identifying users and relations across social media again has an ethical dimension. How can we inform the users of the implications of such data mining in the recruiting situation even though this is not the main purpose of the study? We propose an informed consent procedure in which such information is provided several times both on the initial contact email/website, on the social media platform retrieval, and on the bridging software. Also, it is important for the research design of such bridging software that participants are able to withdraw from the project at any given time (see also Bechmann and Vahlstrup 2015).

As the theoretical, conceptual, and methodological sections of the framework have accounted for, the social relations are more complex and multidimensional than the limited APIs and the social media database design in general. Despite classical methodological issues, computational social sciences will additionally loose even more complexity on the sociological level in terms of relations but gain complexity in terms of different data input sources as pointers towards usage.

Technical Approach

To be able to collect and analyze any data from the different social media platforms, one first needs to gain access to their APIs. Afterwards data have to be bridged allowing to identify the user across the different platforms and not just isolated on each platform individually. For instance, a user named John Thompson on Facebook may name himself Unicorn123 on Twitter and Tennisfan on Instagram. How do we know in fact that this is the same person and thereby analyze the actual social interaction patterns across services?

As we had already built a running software for collecting Facebook data, we expanded this to be able to collect Instagram data as well (see Bechmann and Vahlstrup 2015) and thereby making it possible to bridge Facebook and Instagram users. Because of time and economical limitations, we have chosen not to include Twitter in this initial attempt of bridging the different platforms. To access Twitter data, there are already several tools available (e.g., “yourTwapperKeeper,” “TCAT – Twitter Capture and Analysis Toolset”), so even though they cannot integrate into the software, we would at least be able to collect the Twitter data related to a given case. At the time of implementation, there were no public available software tools for collecting Instagram data, so this seemed to be the obvious choice to implement as proof-of-concept.

The biggest challenge in bridging the data in the technical setup between Facebook and Instagram was to find a way in which we could make sure that we were actually dealing with data from the same user coming from multiple platforms. To be able to identify a user, a unique identifier is needed such as an ID, a username, or an email address (Gross and Churchill 2007).

The Facebook API exposes the user’s ID, username, and email address. The username is optional for the user and will be removed from v2.0 of the API and onwards. Furthermore, Facebook does not guarantee that the email-address will be present and if it is, it is possible for the user to choose a Facebook generated proxy email address instead of their actual email address. As of v2.0 of the Facebook API, user IDs are changed for each third-party application the user installs, so the same user will have multiple IDs, one for each application context. This is executed to protect the user against tracking across applications of different origin (“Graph API User,” “Facebook Platform Changelog”).

Instagram makes it possible to access the user’s ID and username but does not expose the user’s email address. The user ID would be an easy solution if we were to link user data coming from the same platform, but when the user data is delivered from two different platforms, it is very unlikely that the user should have been given the same ID, even by coincidence. The user’s email address is only present on Facebook and we don’t have any way of knowing if it is proxied or not, so we cannot use the email address either. Even if both platforms were to expose the user’s actual email address, there would still be problems by using the email address to link the data. Users often have multiple email addresses (Gross and Churchill 2007), so if different email addresses were used on Facebook and Instagram, we would not be able identify the user using the email address even if it was accessible. The username is not suitable either. First of all, only Instagram users are required to have a username where it is only a possibility on Facebook and again only accessible through the API until v2.0, but even if Facebook required the user to have a username, we would still face the same challenges as with the user’s email address. Users may use different usernames in different contexts (Liu et al. 2013) so using the username to bridge the platforms, if it was possible, would not be a very safe bet either.

Our solution to the problem was to use the software UserModel object to collect the different pieces of information from Facebook and Instagram in a unified signup flow and then if everything succeeds store the Instagram and Facebook user data in our database along with an association between the data.

In our software, all data related to a user of the system, logged in or not, is saved in a UserModel object which is stored in the user’s HTTP-session. As long as the user has not been inactive for more than 30 min, cleared their cookie cache, or switched browsers, we can use the UserModel object to identify the user even though she is not logged in. Both Facebook and Instagram makes it possible to create a custom callback URL which the user will be returned to. When access to their account has been granted or denied, we can create a combined signup flow for both Facebook and Instagram as seen in the illustration below (Fig. 1).
Fig. 1

Software participant signup flow

After we receive each callback, we test if all the permissions have been granted and only if so, we move on to the next step. If the user in the middle of the process should deny access, no data will be saved about the user at all. During the signup flow, data is saved temporarily in a ParticipantSignupInfo object which is set on the UserModel object at the beginning of the signup flow. At the last step, when all permissions have been granted, we create a new FbUser object to hold the Facebook user data, an IgUser object to hold the Instagram data, and finally a Participant object that will bind the FbUser- and IgUser objects together thereby creating the link between the data from the two platforms (Fig. 2).
Fig. 2

From single to multiple social media platforms support

This solution has proven to work very well and has only few downsides that almost all can be prevented. First, the user session expires due to user inactivity for more than 30 min. This is handled by checking for the presence of the ParticipantSignupInfo object in the UserModel. If not present, this means that the user is trying to access the signup flow in an unorderly fashion and the user is presented with an error page and asked to try again. Second, the user copies an URL from the signup flow to another browser. The solution is the same as above because the new browser will result in a new session context. Third, the user has already signed up once – we need to prevent the same user from signing up more than once so we don’t get redundant data. Because we have access to the user’s ID on both Facebook and Instagram, we can compare the ID from the user who is trying to sign up with the IDs we already have stored in our database. If we already have an FbUser object with the same Facebook ID or an IgUser with the same Instagram ID as the user trying to sign up, we can just skip to the end of the signup flow without saving anything new. As the same user can have several Facebook and Instagram accounts, we implement that a user is already signed up if just one of the Facebook or Instagram IDs are present.

The technical implementation of bridging Facebook and Instagram data in our software is made in such a way that it makes further expansion in relation to cross-service analyses easy, so the same user’s profiles from other services such as Twitter, LinkedIn, and YouTube can be linked together with the Facebook and Instagram data. A loosening of the current restriction requiring all services being present and granted access to would probably be needed if more services are added, thus making it easier to recruit users to sign up even though they do not have an account on all the different social media included in the research project.

Conclusion

This chapter has proposed a methodological framework for the study of usage across social media in a user-centric rather than media-centric perspective that encompasses to bridge between sociological interpretations and technical attributes. In doing so, the chapter has discussed four aspects of such a framework: the theoretical, conceptual, methodological, and the technical approach. The framework contributes with suggestions on how researchers can move from a sociological research interest to a technical level and back to a sociological analysis through conceptual bridging. The framework further has contributed with a fully implemented sociological informed technical design that bridge the data collection and analysis of users on different social network services.

The main argument in the chapter is that bridging is not an easy task as several challenges occur on different levels that the researchers need to account for in greater details. First, in the technical infrastructure and/or database structure relations are simple but represent many different complex sociological relationships and interaction types. Second, the elements of social media as accounted for in Table 1 in this chapter (CCU, acknowledgement, hashtag, dissemination, conceptual and social relation) need to be translated socially. In other words, even though @mentions and comments might be two different things, computational social science needs to discuss if they are treated equally in the cross-service analysis and what the research implications of doing this are on the overall research interest in line with the discussions in this chapter. Third, when we study user-centric social media usage, we increase complexity in at least two different dimensions: we propose a mixed method design and we propose a cross-service approach. This means a large increase in different data points both from different services but also from both in-depth data methods (e.g., interviews, personal streams) and general knowledge methods (public streams). This again challenges the researcher in finding patterns in the datasets if analytical and visual overviews are not properly implemented in the bridging software solution as well. Future work lies in addressing more thoroughly how this increase in complexity challenges the cross-service researcher in greater details in the analytical phase on case studies. Further, the framework has accounted for sociological differences in studying different types of communication on timely, spatial, and relational parameters. Future work also implies to account for potential conceptual differences in different case studies on these three parameters. Last, but not least, future work needs to focus on how the analysis of different CCUs are supported by different and isolated scientific fields within deep learning and how a bridging on this level might take place. This is especially the case with the comparison of mainly textual content units on Twitter with visual content units on Instagram. Natural language processing algorithms and picture recognition algorithms are two different machine learning disciplines. Comparing these elements from a sociological perspective therefore requires an extensive discussion on the differences on algorithmic levels in order to reflect on the comparability of the outcome on a sociological informed analytical level. This chapter has suggested solutions and different challenges in doing sociological cross-service user studies but the perspective calls for even more future research.
Table 1

Social elements of the three social network services

 

Twitter

Instagram

Facebook

CCU

Tweet

Media object

Status update

Acknowledgment

Favorite

Heart

Like

Hashtag (#)

Yes

Yes

Yes (from 2013)

Dissemination

Retweet

Share

Conceptual relation

@mention, mention, reply

@mention, comment

Mention, wall post, comment

Social relation

Follower

Follower

Friend, follower

References

  1. Anderson K, Nafus D, Rattenbury T, Aipperspach R (2009) Numbers have qualities too: experiences with ethno-mining. Ethnogr Prax Ind Conf 2009(1):123–140Google Scholar
  2. Bechmann A (2013) Non-informed consent cultures: privacy policies and app contracts on Facebook. J Media Bus Stud 11(1):21–38CrossRefGoogle Scholar
  3. Bechmann A (2015) Managing the interoperable self. In: Bechmann A, Lomborg S (eds) The ubiquitous internet: user and industry perspectives. Routledge, New York, pp 54–73Google Scholar
  4. Bechmann A, Klausen HB (2015) Measuring impact across social media. Rethink Impacts Rep 2:1–34Google Scholar
  5. Bechmann A, Vahlstrup PB (2015) Studying Facebook and Instagram data: the Digital Footprints software. 20(12)Google Scholar
  6. Becker H, Iter D, Naaman M, Gravano L (2012) Identifying content for planned events across social media sites. In: Proceedings of the fifth ACM international conference on Web search and data mining, pp 533–542. ACMGoogle Scholar
  7. Berelson B (1952) Content analysis in communication research. Free Press, GlencoeGoogle Scholar
  8. Blumer H (1980) Mead and Blumer: the convergent methodological perspectives of social behaviorism and symbolic interactionism. Am Sociol Rev 45(3):409–419CrossRefGoogle Scholar
  9. Boyd D, Ellison NB (2007) Social network sites: definition, history, and scholarship. J Comput-Mediat Commun 13(1):210–230CrossRefGoogle Scholar
  10. Boyd D, Marwick A (2011) Social privacy in networked publics: teens’ attitudes, practices, and strategies. In: Proceedings of a decade in internet time, 21–24 September 2011, University of OxfordGoogle Scholar
  11. Bruns A, Burgess J (2012) Researching news discussion on Twitter: new methodologies. Journal Stud.  https://doi.org/10.1080/1461670X.2012.664428
  12. Burkitt I (2008) Social selves: theories of self and society. Sage, Thousand OaksGoogle Scholar
  13. Clark A (2011) Supersizing the mind: embodiment, action, and cognitive extension. Oxford University Press, New YorkGoogle Scholar
  14. Courtois C, Merchant P (2013) An evaluation of the potential of Web 2.0 APIs for social research. In: Patriarche G, Bilandzic H, Jensen JL, Jurišić J (eds) Audience research methodologies: between innovation and consolidation. Routledge, London, pp 212–226Google Scholar
  15. Ellison NB, Boyd D (2013) Sociality through social network sites. In: Dutton B (ed) Oxford handbook of internet studies. Oxford University Press, Oxford, pp 151–172Google Scholar
  16. Enli GS, Skogerbø E (2013) Personalized campaigns in party-centred politics: Twitter and Facebook as arenas for political communication. Inf Commun Soc 16(5):757–774CrossRefGoogle Scholar
  17. Facebook Platform Changelog. Retrieved from https://developers.facebook.com/docs/apps/changelog
  18. Fernandes J, Giurcanu M, Bowers KW, Neely JC (2010) The writing on the wall: a content analysis of college students’ Facebook groups for the 2008 presidential election. Mass Commun Soc 13(5):653–675CrossRefGoogle Scholar
  19. Fraser M, Dutta S (2010) Throwing sheep in the boardroom: how online social networking will transform your life, work and world. Wiley, New YorkGoogle Scholar
  20. Gencarelli TF (2006) Perspectives on culture, technology, and communication: the media ecology tradition. Hampton Press, CresskillGoogle Scholar
  21. Gergen KJ (2009) Relational being: Beyond self and community. Oxford University PressGoogle Scholar
  22. Gross BM, Churchill EF (2007) Addressing constraints: multiple usernames task spillage and notions of identity. In: CHI EA ’07 CHI ’07 extended abstracts on human factors in computing systems, pp 2393–2398Google Scholar
  23. Hanna R, Rohm A, Crittenden VL (2011) We’re all connected: The power of the social media ecosystem. Business horizons 54(3):265–273CrossRefGoogle Scholar
  24. Haraway D (1991) A cyborg manifesto: science, technology, and socialist-feminism in the late twentieth-century. In: Simians, cyborgs, and women: the reinvention of nature. Routledge, New YorkGoogle Scholar
  25. Jensen JL, Scott Sørensen A (2013) Nobody has 257 friends. Nordicom Rev 34(1):49–62CrossRefGoogle Scholar
  26. Jensen JL, Tække J (2013) Indledning: Facebook i den danske hverdag. In: Fra socialt netværk til metamedie. SamfundslitteraturGoogle Scholar
  27. Kietzmann JH, Hermkens K, McCarthy IP, Silvestre BS (2011) Social media? Get serious! Understanding the functional building blocks of social media. Bus Horiz 54(3):241–251CrossRefGoogle Scholar
  28. Latour B (2011) Network theory, networks, societies, spheres: reflections of an actor-network theorist. Int J Commun 5:15Google Scholar
  29. Law J (1991) Introduction: monsters, machines and sociotechnical relations. In: Law J (ed) A sociology of monsters: essays on power, technology and domination. Routledge, LondonGoogle Scholar
  30. Liu J, Zhang F, Song X, Song Y-I, Lin C-Y, Hon H-W (2013) What’s in a name?: an unsupervised approach to link users across communities. In: WSDM ’13 Proceedings of the sixth ACM international conference on Web search and data mining, pp 495–504Google Scholar
  31. Lomborg S, Bechmann A (2014) Using APIs for data collection on social media. Inf Soc 30(4):256–265CrossRefGoogle Scholar
  32. Lotan G, Graeff E, Ananny M, Gaffney D, Pearce I, boyd d (2011) The revolutions were tweeted: information flows during the 2011 Tunisian and Egyptian revolutions. Int J Commun 5:1375–1405Google Scholar
  33. Marichal J (2013) Political Facebook groups: micro-activism and the digital front stage. First Monday 12(2). https://doi.org/10.5210%fm.v18i12.4653Google Scholar
  34. Rogers R (2013) Digital methods. MIT Press, Cambridge, MAGoogle Scholar
  35. Taneja H, Webster JG, Malthouse EC, Ksiazek TB (2012) Media consumption across platforms: Identifying user-defined repertoires. New media & society 14(6):951–968CrossRefGoogle Scholar
  36. Turkle S (1995) Life in the screen: identity in the internet. Simon, New YorkGoogle Scholar
  37. Wu S, Mason WA, Hofman JM, Watts DJ (2011) Who says what to whom on Twitter. In: Proceedings of the international World Wide Web conference (WWW 2011), pp 705–714Google Scholar
  38. YourTwapperKeeper – Archive Your Social Media. Retrieved from https://github.com/540co/yourTwapperKeeper

Copyright information

© Springer Science+Business Media B.V., part of Springer Nature 2018

Authors and Affiliations

  • Jakob Linaa Jensen
    • 1
    Email author
  • Peter B. Vahlstrup
    • 2
  • Anja Bechmann
    • 3
  1. 1.Danish School of Media and JournalismAarhus CDenmark
  2. 2.Information StudiesAarhus UniversityAarhus CDenmark
  3. 3.Aarhus Institute of Advanced StudiesAarhus UniversityAarhus CDenmark

Personalised recommendations