Professional Documents
Culture Documents
11 - Exploring Social Relationships in
11 - Exploring Social Relationships in
Structural Analysis
Dominik E. Froehlich
Abstract
While the concept of mixed method social network analysis (MMSNA) is gaining in
popularity, there is a notable lack of specific mixed research designs that guide the
implementation of MMSNA. In this chapter, I draw from qualitative social network
analysis, specifically, qualitative structural analysis, and expand it towards a mixed
research design. This change, which requires relatively little additional input, fulfills
several important purposes at the same time, and hence may be conducive in increasing
the overall quality of a study.
1.1 Introduction
Contemporary social network research is largely quantitatively driven (e.g., Freeman, 2004;
Froehlich, Rehm, & Rienties, 2020). However, in the last decade, qualitative methods have re-
emerged in social network research. In a field that has been focused too much and too long on
abstract notions of networks, qualitative approaches are becoming increasingly popular and
important (Bernhard, 2018; Herz, Peters, & Truschkat, 2015; Hollstein, 2011). Specifically, Crossley
(2010) points to the weakness of a quantitative approach in social network analysis (SNA) when it
comes to learning about the meaning (i.e., why did actor X learn from actor Y) and content of
relationships (i.e., what has actor X actually learned from actor Y), as well as their variability across
time (i.e., how did the relationship between actor X and actor Y change over time).
At the same time, qualitative methods come with their very own set of limitations. Most importantly,
they often lack oversight to discuss structural phenomena in networks, such as the identification of
clusters and other forms of subgroups, positions of brokerage and the associated idea of structural
holes (Burt, 1992; Froehlich, 2018), and network-level features like density or centralization
(Freeman, 1978). These structural properties, however, are an integral part of much SNA, so it
becomes quite difficult to embed the qualitative findings into the ongoing scientific debate about
social networks.
One such qualitative approach towards SNA is qualitative structural analysis (QSA; Herz et al., 2015).
In QSA, qualitative SNA is conducted using (some form of) interviews as a data collection procedure,
which is enriched with network maps (Kahn & Antonucci, 1980) that the interviewer creates ad hoc,
together with the interviewee during the interview. This is also exemplified by the concentric circle
approach discussed by Van Waes and Van den Bossche (2020). The rationale of the network maps is
to allow for a relational perspective when analyzing the interviews (a more detailed account of the
procedure will be reported below). While these maps do allow for an informative qualitative analysis
of the interview data and add social network thinking to the analysis, I proffer that replacing the ad
hoc network maps with a quantitative pre-survey has important advantages.
In this chapter, I propose an improvement to the QSA and its reliance on ad hoc network maps. By
adding a relatively small quantitative research strand to the (dominant) qualitative strand, we will be
able to efficiently improve the research output in many research settings. I present my argument in
three steps. First, I will briefly summarize the nature and process of QSA. Second, I will outline how
the transition from qualitative SNA design to a qualitatively-oriented, MMSNA design can produce
desirable effects in terms of research quality. I then focus on the three major components of mixed
structural analysis (MSA): the quantitative research strand, the qualitative research strand, and the
integration of both. Third, the chapter finishes with a short reflection on the main features of MSA.
To show the process of conducting QSA, Herz et al. (2015) report a study among career transition
services employees and managers. Narrative interviews (Alheit, 2007) were conducted in conjunction
with the ad hoc creation of network maps (Kahn & Antonucci, 1980). In other words, the (narrative)
interviews were the dominant data source, but whenever a name or organization was mentioned,
network maps were used as a visual tool to specify the nature of that additional actor and the
relationship between that actor and the interviewee. This allowed for some standardization across
interviews, as the network map provided the interviewee with some structure (four concentric circles
and the center) (Herz et al., 2015).
While the information about the data collection is somewhat limited, the authors gave an extensive
account concerning the analysis of the interview transcript. Specifically, they took two steps in
analyzing the interview (see Figure 11.1), in accordance with the two sets of data they have available
for each interview: the network map and the interview transcript.
Figure 11.1
In the first step, researchers should have a look at the network map based on what is known from
formal network analysis. Specifically, Herz et al. (2015) gave a list of exemplary questions to guide the
researcher’s reading of the network maps in three categories: structure-focused (which zeros in on
cohesion and nodal equivalence), actor-focused (which focuses, for instance, on questions of an
individual actor’s centrality and composition of alters), and tie-focused (which centers around the
meaning of individual ties). In other words, the network researcher is led through the major
components of any network (Froehlich & Brouwer, forthcoming; Wasserman & Faust, 1994): the
nodes (actors-focused interpretation), their relationships with each other (tie-focused
interpretation), and the network that the former two components create (structure-focused
interpretation). At the same time, this categorization may be mapped onto the discussion of units of
analysis within (mixed methods) SNA (Bolíbar, 2015; Froehlich, Rehm, & Cornelissen, forthcoming
Froehlich, Mejeh, et al., 2020; Rehm et al., 2020).
The interpretation of the network maps and the questions that arise from that interpretation are
then used as sensitizing concepts to inform the analysis of the interview transcript. For this second
step, the transcript is analyzed sequentially using the coding procedures recommended by grounded
theory methodology (i.e., open, axial, and selective coding; Glaser, 1992).
It is important to point out that this QSA workflow is just one possible arrangement; Herz et al.
(2015) present QSA as a very flexible approach, and remain open to alternative processes. For
example, they pose the question about the sequence of steps. Also, they state (but do not show) that
QSA may also be applied to sociocentric (and not only to egocentric) networks. These reflections, in a
way, tie into the changes that I propose in this chapter, and it may be argued that the method to be
presented is just a variant of QSA. However, given that my suggestions transcend the interpretative
paradigm (Guba & Lincoln, 1994) that is the backdrop of QSA, I do believe that a new label is
warranted.
In this section, I will describe how the introduction of a quantitative research strand prior to the QSA
may be implemented and used. Therefore, this chapter features a sub-section that discusses this
quantitative strand. However, since this addition makes it a mixed study, the mixing and the
potential for integration also need to be discussed; this discussion is featured in another sub-section.
Last, the integration leads to changes in the original approach of conducting QSA so that the
approach to the qualitative strand of research needs to be modified.
Second, the instruments for quantitative data collection need to be designed. While any method of
quantitative data collection is suitable, it is probably a sociometric survey that gives the network
researcher most flexibility. Best practice of quantitative data collection has been discussed elsewhere
(Längler et al., ; Marsden, 1990; Marsden & Campbell, 1984) and will not be a focus of this chapter.
However, it is, of course, important to align the questions of the quantitative data collection to the
goals of the overall research project and the questions of the qualitative phase of MSA.
Third, it is important to also think of ethical considerations when conducting MSA. Individuals taking
part in the quantitative data collection may not be selected for an interview in the end, but they may
still appear in the networks of the interviewed persons (e.g., Korir et al., 2020). Therefore, it is
important to clarify from the beginning how to safeguard the personal interests of the study
participants and their alters. Most importantly, anonymity in many cases cannot be guaranteed, as
interviewees might refer to and evaluate their alters. The network researcher needs to tackle this
delicate issue (e.g., Korir et al., 2020).
As mentioned before, the main purpose of the quantitative research strand is to further develop and
prepare the qualitative strand of research. However, as illustrated by the exemplary list of purposes
of mixing developed by Schoonenboom et al. (2018), multiple points of integration may be
considered: first, the quantitative results may inform the sampling process. By knowing about the
structural features of a network and learning about the positions individual actors take, the sampling
procedure for the qualitative research strand may be executed in a more purposeful and thoughtful
manner (e.g., Murphy et al., 2020). For instance, depending on the research question, actors in
similar or in dissimilar positions may be sampled.
Notions such as structural determinism (Emirbayer & Goodwin, 1994) and most network theories
(Borgatti & Halgin, 2011; Monge & Contractor, 2003) suggest that positioning within a network plays
an important role in the individuals’ opportunities and limitations, and hence the modes of actions
available to the actors. Thus, taking the different positions already in the sampling procedure into
account should positively impact the quality of data generated during the interviews. In the terms of
Schoonenboom et al. (2018), the purpose here is the development of sampling. A strategy that
samples contrasting cases may also be used to extend the purpose of mixing to what Schoonenboom
et al. (2018) refer to as subgroup analysis, where an effect is studied in different subgroups. In this
case, the groups would be defined by their network position (e.g., the notion of structural
equivalence in network literature; Wasserman & Faust, 1994).
Second, the quantitative results may inform the guidelines used in the interview. This may work
through different ways. For example, unlike in QSA, where the (egocentric) network map is produced
alongside the interview, the final (sociocentric) network map may already be presented during the
interview as a prompt (if this is desirable for answering the research question), such as done in
Exemplar 4 of Rienties (2020). Alternatively, as shown in Figure 11.2, the interviewer may also extract
only the ego-networks from the data. Hence the interviewer can already prepare questions about
specific constellations in the (ego-)network of the interviewee. Alternatively, the MM approach to
network data collection presented by Rice et al. (2014) may be implemented without additional
effort. Rice et al. (2014) suggested combining a free recall name generator (Bailey & Marsden, 1999;
Burt et al., 2012) – which may be implemented as part of the quantitative survey – with the relations
mentioned during an (semi-structured) interview, which is done in the qualitative strand of MSA.
Another approach regarding how the quantitative results may inform the guidelines used in the
interview is implemented by triggering a very different set of questions that had not seemed relevant
before the quantitative results. For instance, these could be questions about the historic
development of ties, questions about odd features of the network, or follow-up questions to any
psychometric questions that may be included in the survey alongside the sociometric part.
Additionally, questions may arise from the sheer availability of multiple raters (“You mentioned that
you work a lot with A and B, but neither A nor B mentioned you in their respective networks. Can you
explain this?”), which also allows for analysis of, for example, reciprocity (Gouldner, 1960). Last, the
interview guidelines may change simply because some questions were asked already in the
quantitative data collection. In the terms of Schoonenboom et al. (2018), the purpose of any variant
presented here is interview schedule development.
Figure 11.2
Third, the purpose of mixing may also be one of method triangulation. For instance, and relating back
to the mixed data collection procedure proposed by Rice et al. (2014), networks and relationships
could be generated independently in both a quantitative and a qualitative fashion. This would allow a
comparison of both measurements.
Fourth, I agree with the general procedure proposed by Herz et al. (2015) to connect the relational
information from the (quantitative or qualitative) network data. The questions provided by Herz et
al. (2015) about structural features, actors, and ties, represent a good starting point for thinking
about the network map. However, the quantitative data will usually allow for some more in-depth
analysis of the dataset. To begin with, a sociometric network will normally be wider in diameter
(Wasserman & Faust, 1994) than an egocentric network. As such, the interpretation may also include
other structural phenomena such as clusters (Wasserman & Faust, 1994) or structural holes (Burt,
1992; Froehlich, 2018). Relatedly, this will allow studying the phenomenon at hand in different units
of analysis (Bolíbar, 2015; Froehlich, Rehm Cornelissen, et al., forthcoming; Froehlich &
Schoonenboom, forthcoming). Next, the quantitative data may not only be analyzed and interpreted
in qualitative terms (“qualitizing”, Tashakkori & Teddlie, 1998), but also by means of formal SNA. At
the very least, basic metrics may be calculated – centrality measures (Freeman, 1978), clusters, etc. –
or different layout and formatting options may be applied to the visualization procedure in order to
increase the effectiveness of the visual interpretation. Indeed, empirical research suggests that
layout (Blythe, McGrath, & Krackhardt, 1996) and formatting (Pfeffer & Schönfeld, 2018) are
important factors to consider when it comes to interpreting network maps. Last, the quantitative
data collection may allow for the gathering of additional data to test other network theories, such as
homophily (Frieling & Froehlich, 2017; McPherson, Smith-Lovin, & Cook, 2001). For these network
theories, other sets of guiding questions may be developed on top of questioning the structural
features, the actors, and the nature of ties.
Fifth, by having quantitative and qualitative components, the potential for common method bias
(where the measurement instrument causes variation in the data; Podsakoff, Mackenzie, &
Podsakoff, 2012) is reduced. This is even truer when the quantitative component is executed long
before the potential interviews, allowing for the data to triangulate across time.
For the data analysis, Herz et al. (2015) proposed deriving a set of questions from the generated
network maps. These questions should then be answered based on the interview transcripts (e.g.,
using open coding procedures). For MSA, this is no different. However, MSA generates more data
before the actual interview, so more data may be used as a basis upon which to pose questions. This
is especially true when it comes to data from the sociometric network that does not exist in a QSA.
The advantages – previously discussed under the banner of purposes of mixing – therefore lie in the
potential for triangulating the data, reducing common method bias and measurement error, and
analyzing the wider structural properties of a network (and hence also including other units of
analysis).
1.4 Conclusion
In this chapter, I have set out to develop a specific MMSNA research design called mixed structural
analysis (MSA). My point of departure was the qualitative structural analysis (QSA) developed by
Herz et al. (2015). I then presented what could be called a sequential explanatory research design
(Creswell, 2009), where a quantitative pre-study is used to further develop (e.g., Schoonenboom et
al., 2018) the QSA. I then dissected the process and discussed the quantitative research strand, the
qualitative research strand, and the added value and purpose of mixing. Like Herz et al. (2015), I
conclude this chapter with a reflection upon the features, and open questions about the present
MSA method.
In introducing the method, I have also made the claim that its benefits may be reaped without too
much effort. This statement is based on two tenets. First, surveys are a cheaper form of data
collection than interviews. While a purposeful sampling procedure, for instance, could be informed
purely by qualitative methods, quantitative methods – and especially surveys – generally allow for
more of an overview at lower cost. Second, there are also economic influences. Given the dominance
of quantitative methods of SNA in the last century (e.g., Freeman, 2004; Froehlich, Rehm, & Rienties,
2020), these methods have also become more accessible, for instance, through an increase in
availability of workshops or university courses at any level, which often focus exclusively on
quantitative social network research.
In this article, I have written about MSA in purely conceptual terms. Future research may focus more
on the practical side of conducting MSA, in a similar way that Herz et al. (2015) focused on explicating
how to actually “do” QSA. For instance, while I have written about the formal features of MSA, I have
not explained how to approach the planning of an MSA study. Certainly, as more components are
involved in the research design, planning gets more complex and more potential risks need to be
managed. Another example would be the design of the intersection of the quantitative and the
qualitative parts: What information to extract from the quantitative data collection (e.g., which
relationships are relevant)? Should the qualitative phase seek to confirm these networks, or should
the focus be entirely different? While these questions are relevant and need to be discussed
thoughtfully by the researchers executing MSA, I also want to reiterate at this point that MSA comes
in a rather flexible form and may be utilized to achieve a range of research objectives.
References
Alheit, P. (2007). Geschichten und Strukturen. Methodologische Überlegungen zur Narrativität
[Stories and structures. Methodological thoughts on narration]. Zeitschrift für Qualitative Forschung,
8(1), 23.
Bailey, S., & Marsden, P. V. (1999). Interpretation and interview context: examining the General
Social Survey name generator using cognitive methods. Social Networks, 21(3), 287–309.
https://doi.org/10.1016/S0378-8733(99)00013-1
Blythe, J., McGrath, C., & Krackhardt, D. (1996). The effect of graph layout on inference from social
network data. In F. J. Brandenburg (Ed.), Graph Drawing (Vol. 1027, pp. 40–51).
https://doi.org/10.1007/BFb0021789
Bolíbar, M. (2015). Macro, meso, micro: Broadening the “social” of social network analysis with a
mixed methods approach. Quality & Quantity, 3, 2217–2236. https://doi.org/10.1007/s11135-015-
0259-0
Borgatti, S. P., & Halgin, D. S. (2011). On network theory. Organization Science, 22, 1168–1181.
https://doi.org/10.1287/orsc.1110.0641
Burt, R. S. (1992). Structural holes: the structure of social capital competition. Cambridge, MA:
Harvard University Press.
Burt, R. S., Meltzer, D. O., Seid, M., Borgert, A., Chung, J. W., Colletti, R. B., … Margolis, P. (2012).
What’s in a name generator? Choosing the right name generators for social network surveys in
healthcare quality and safety research. BMJ Qual Saf, 21(12), 992–1000.
https://doi.org/10.1136/bmjqs-2011-000521
Creswell, J. W. (2009). Research Design: qualitative, quantitative, and mixed methods approaches
(3rd ed.). https://doi.org/10.4135/9781849208956
Crossley, N. (2010). The Social World of the Network. Combining Qualitative and Quantitative
Elements in Social Network Analysis. Sociologica, (1). https://doi.org/10.2383/32049
Emirbayer, M., & Goodwin, J. (1994). Network Analysis, Culture, and the Problem of Agency. The
American Journal of Sociology, 99(6), 1411–1454.
Freeman, L. C. (1978). Centrality in social networks conceptual clarification. Social Networks, 1(3),
215–239. https://doi.org/10.1016/0378-8733(78)90021-7
Frieling, M., & Froehlich, D. E. (2017). Homophilie, Diversity und Feedback: Eine soziale
Netzwerkanalyse [Homophily, diversity, and feedback: A social network analysis]. In Internationales
Personalmanagement: Rollen – Kompetenzen – Perspektiven. Implikationen für die Praxis
[International human resource management: Roles – competencies – perspectives]. SpringerGabler.
Froehlich, D. E. (2018). Burt (2005): Brokerage & Closure. In B. Holzer & C. Stegbauer (Eds.),
Schlüsselwerke der Netzwerkforschung. VS Verlag für Sozialwissenschaften.Froehlich, D. (2020b).
Exploring Social Relationships in “a Mixed Way”: Mixed Structural Analysis In D. Froehlich, M. Rehm,
& B. Rienties (Eds.), Mixed Methods Approaches to Social Network Analysis (pp. XX-XX). London:
Routledge.
Froehlich, D. (2020a). Mapping Mixed Methods Approaches to Social Network Analysis in Learning
and Education. In D. Froehlich, M. Rehm, & B. Rienties (Eds.), Mixed Methods Approaches to Social
Network Analysis (pp. XX-XX). London: Routledge.
Froehlich, D. E., & Brouwer, J. (forthcoming). Social Network Analysis as Mixed Analysis. In A. J.
Onwuegbuzie & R. B. Johnson (Eds.), Reviewer’s Guide for Mixed Methods Research Analysis.
Routledge.Froehlich, D., Mamas, C., & Schneider, H. (2020). Automation and the journey to mixed
methods social network analysis. In D. Froehlich, M. Rehm, & B. Rienties (Eds.), Mixed Methods
Approaches to Social Network Analysis (pp. XX-XX). London: Routledge.
Froehlich, D., Mejeh, M., Galey, S., & Schoonenboom, J. (2020). Integrating Units of Analysis. In D.
Froehlich, M. Rehm, & B. Rienties (Eds.), Mixed Methods Approaches to Social Network Analysis (pp.
XX-XX). London: Routledge.
Froehlich, D. E., Rehm, M., & Cornelissen, F. (forthcoming). Investigating Informal Learning via Social
Media using Mixed Methods Social Network Analysis: A Case Study. MedienPädagogik: Zeitschrift für
Theorie und Praxis der Medienbildung.
Froehlich, D. E., & Schoonenboom, J. (forthcoming). Multi-level qualitative social network research
for integrating different units of analysis.
Froehlich, D. E., Van Waes, S., Schoonenboom, J., & Schäfer, H. (forthcoming). Linking Quantitative
and Qualitative Network Approaches: A Review of Mixed Methods Social Network Analysis in
Education Research. Review of Research in Education.
Geertz, C. (1973). Thick Description: Toward an Interpretive Theory of Culture. In The Interpretation
of Cultures: Selected Essays (pp. 3–30). New York, NY: Basic Books.
Glaser, B. G. (1992). Basics of Grounded Theory Analysis: Emergence Vs. Forcing. Mill Valley, CA:
Sociology Press.
Herz, A., Peters, L., & Truschkat, I. (2015). How to do Qualitative Structural Analysis: The Qualitative
Interpretation of Network Maps and Narrative Interviews. Forum Qualitative Sozialforschung /
Forum: Qualitative Social Research, 16(1). https://doi.org/10.17169/fqs-16.1.2092
Hesse-Biber, S. (2010). Qualitative Approaches to Mixed Methods Practice. Qualitative Inquiry, 16(6),
455–468. https://doi.org/10.1177/1077800410364611
Hollstein, B. (2011). Qualitative Approaches to Social Reality: The Search for Meaning. In J. Scott & P.
J. Carrington (Eds.), The Sage Handbook of Social Network Analysis (pp. 404–416). London: Sage
Publications.
Kahn, R. L., & Antonucci, T. C. (1980). Convoys over the life course: attachment, roles, and social
support. In P. B. Baltes & O. Brim (Eds.), Life Span Development and Behavior (Vol. 3). New York, NY:
Academic Press.
Korir, M., Mittelmeier, J., & Rienties, B. (2020). Is Mixed Methods Social Network Analysis Ethical? In
D. Froehlich, M. Rehm, & B. Rienties (Eds.), Mixed Methods Approaches to Social Network Analysis
(pp. XX-XX). London: Routledge.Laumann, E. O., Marsden, P. V., & Prensky, D. (1983). The Boundary
Specification Problem in Network Analysis. In R. S. Burt & E. Minor (Eds.), Applied network analysis: A
methodological introduction (pp. 18–87). Beverly Hills, CA: Sage.
Marsden, P. V. (1990). Network Data and Measurement. Annual Review of Sociology, 16(1990), 435–
463.
Marsden, P. V., & Campbell, K. E. (1984). Measuring tie strength. Social Forces, 63(2), 482–501.
https://doi.org/10.2307/2579058
McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a Feather: Homophily in Social
Networks. Annual Review of Sociology, 27, 415–444. https://doi.org/10.1146/annurev.soc.27.1.415
Monge, P. R., & Contractor, N. S. (2003). Theories of Communication Networks. New York, NY: Oxford
University Press.
Pfeffer, J., & Schönfeld, M. (2018). Towards an Understanding of Perception of Network Visualization.
Presented at the Sunbelt XXXVIII Conference, Utrecht.
Podsakoff, P. M., Mackenzie, S. B., & Podsakoff, N. P. (2012). Sources of Method Bias in Social Science
Research and Recommendations on How to Control It. Annual Review of Psychology, 63(1), 539–569.
https://doi.org/10.1146/annurev-psych-120710-100452Rehm, M., Cornelissen, F., Notten, A., Daly,
A., & Supovitz, J. (2020). Power to the People?! Twitter Discussions on (Educational) Policy Processes.
In D. Froehlich, M. Rehm, & B. Rienties (Eds.), Mixed Methods Approaches to Social Network Analysis
(pp. XX-XX). London: Routledge.
Rice, E., Holloway, I. W., Barman-Adhikari, A., Fuentes, D., Brown, C. H., & Palinkas, L. a. (2014). A
Mixed Methods Approach to Network Data Collection. Field Methods, 26(3), 252–268.
https://doi.org/10.1177/1525822X13518168
Rienties, B. (2020). Powers and limitations of MMSNA: Critical reflections and moving forwards In D.
Froehlich, M. Rehm, & B. Rienties (Eds.), Mixed Methods Approaches to Social Network Analysis (pp.
XX-XX). London: Routledge.
Schoonenboom, J., Johnson, R. B., & Froehlich, D. E. (2018). Combining Multiple Purposes of Mixing
Within a Mixed Methods Research Design. International Journal of Multiple Research Approaches,
10(1), 271–282. https://doi.org/10.29034/ijmra.v10n1a17
Smith, S. S. (2013). Social network boundaries and tricky to access populations: a qualitative
approach. International Journal of Social Research Methodology, 17(6), 613–623.
https://doi.org/10.1080/13645579.2013.820076
Tashakkori, A., & Teddlie, C. (1998). Mixed methodology: Combining qualitative and quantitative
approaches. Thousand Oaks, CA, US: Sage Publications, Inc.Van Waes, S., & Van den Bossche, P.
(2020). Around and around: The concentric circles method as powerful tool to collect mixed method
network data. In D. Froehlich, M. Rehm, & B. Rienties (Eds.), Mixed Methods Approaches to Social
Network Analysis (pp. XX-XX). London: Routledge.
Wasserman, S., & Faust, K. (1994). Social Network Analysis: Methods and Applications. New York, NY:
Cambridge University Press.