6. Reflection on the method design
6.4. Development and implementation of practical indicators
The development process of the practical indicators differed per analysis method. There were already existing indicators for social network measures. Although these statistical indicators were originally developed for the analysis of offline social networks, the researcher could adapt these measures to the context of online social media networks. Table 9 summarizes all practical indicators derived from the online social network analysis and describes possible inferences.
Table 9 Practical indicators derived from the online social network analysis
Practical indicators derived from the online social network analysis
Indicator Short description
Network metrics
Graph density Indicates how well connected a network is internally Modularity Indicates the internal community structure of a network Connected components Indicates potential isolation of sub-networks within a network Diameter Indicates the longest shortest path between all pairs of nodes
Edge metrics
(average) Path lengths Indicates the (average) distance between all pairs of nodes
Node metrics
Degree Indicates a node´s (general, in- and out-going) level of activity Betweenness centrality Indicates how often a node appears on shortest paths between
nodes and potential control over information flow in a network (nodes as brokers)
Closeness centrality Indicates a node´s (average) distance to all other nodes in the network and geographic position within the network
Eccentricity Indicates the maximum distance of a node to its most distant node in the network
Eigenvector centrality Indicates a node´s importance based on connections to other important nodes in the network
Clustering coefficient Indicates how complete the neighborhood of a node is
The difference the researcher had to outline before analyzing the network data was that the online social media networks were defined in terms of communication networks. Actually, for the first time, the intangible concept of social capital became visible. This was earlier indicated in the literature review, and therefore, integrated in the
conceptualization of social capital in online social media networks. The researcher defined relationship types according to this conceptualization. As soon as the indicators of the online social network analysis were defined, this technique was implemented by using the software tool Gephi. This tool offered state of the art network visualization and
96 all relevant network measurement facilities. The different measures were, thus,
computed within this software tool.
The relationships mentioned above also formed the basis for the triad census used in this research. Therefore, the indicators derived from the triad configurations implied the definitions of relationship types transcribed to the context of online social media networks. These indicators were, thus, also applicable to online social networks. There are 16 different possible triad configurations. Table 10 shows the practical indicators derived from the triad census and it describes their general levels of (inter)activity. Table 10 Practical indicators derived from the triad census
Practical indicators derived from the triad census
Indicator Short description Level of
activity
MAN low
003 Empty triad; indicates three null dyads 012 Indicates only one asymmetric dyad 102 Indicates only one mutual dyad
021D Indicates two downward asymmetric dyads 021U Indicates two upward asymmetric dyads 021C Indicates two cyclic asymmetric dyads
111D Indicates one mutual dyad and one downward dyad 111U Indicates one mutual dyad and one upward dyad 030T Indicates three asymmetric transitive dyads 030C Indicates three asymmetric cyclic dyads 201 Indicates two mutual dyads
120D Indicates one mutual dyad and two downward asymmetric dyads
120U Indicates one mutual dyad and two upward asymmetric dyads
120C Indicates one mutual dyad and two cyclic asymmetric dyads
210 Indicates two mutual dyads and one asymmetric dyad 300 Indicates three mutual dyads
high
These 16 triad configurations remained valid, even if the interpretation of the relationship types relied on communication relationships. Therefore, the implementation of the indicators was a rather simple process. It was the conceptualization of social capital which had to be previously defined.
97 The researcher used content analysis as third analysis method. Content analysis can be a very flexible and adaptable approach to investigate many different kinds of research data. The development of practical indicators for this particular research integrated both conventional and directed approaches to content analysis in this research. The researcher used a bottom-up approach to the analysis of the LinkedIn content, thus, starting with reading and categorizing the content of the discussions. In a further step, the researcher tried to bring the categories in line with existing literature. This was an iterative process, resulting in an appropriate coding scheme for the LinkedIn content. For the Twitter content analysis, the researcher based the categories on an existing coding scheme but adapted it to the actual content of the tweets. This was also an iterative development process. Table 11 describes the indicators derived from the content analysis methods. Table 11 Practical indicators derived from the content analysis
Practical indicators derived from the content analysis
Indicator Short description
LinkedIn content analysis
Information Indicates that the post or comment mainly contains information and was primarily created to share ideas, knowledge, or expertise
Identity Indicates that the post or comment mainly contains supporting content and was primarily created to build or strengthen a common identity
Action Indicates that the post or comment mainly contains appealing messages and was primarily created to motivate or activate others to participate and engage
Twitter content analysis
Content Indicates which information is actually shared through the tweets; resources, events, or personal expressions
Sentiment Indicates the intention and the temper of the tweets; informative, supportive, appealing, questioning or neutral
Link Indicates which links were added to extent the content of the tweets; news or project websites, social networking sites, the Google map, or
government websites
Referring to the characteristics of different approaches to content analyses, the researcher recommends appropriate methodological adaptation in future research, according to contingently varying contents, depending on different cases. As the
development of the indicators referring to the content analysis was much more complex than the implementation of the quantitative methods, the implementation of these indicators was also a more sophisticated process. Categories and codes had to be constantly evaluated and if necessary, adjusted during implementation.
98 Additionally, the researcher had to test the coding schemes by asking a second coder to apply the coding schemes to the research data.