with respect to the research question and data cross-section under study. For this, we
constructed displays organized by modularity concept and over time, in which we
structured, restructured, and analyzed the within case displays. Because we selected
our cases for literal replication, we were able to observe constructs and patterns a
number of times and obtain general insights. Thus, based on similarities among the
cases we were able to draw conclusions on the modularity aspects and practices under
study as well as on their relationships to demand-based care.
To verify the general patterns and insights that emerged from the data, we compared
them systematically with the evidence of each case to see whether the results accurately
reflected the case study situations (Eisenhardt, 1989). In addition, we reviewed our
insights against those of existing literature. Reviewing emergent theory involves
asking what is similar, what is different and why (Eisenhardt, 1989). For each research
question under study, we addressed literature that deviated from or conflicted with our
findings to force more creative thinking and deeper insights. In addition, we addressed
literature discussing similar findings to tie together underlying similarities. This aids
the generalizability and validity of the research findings (Eisenhardt, 1989).
4.7 Generalizability of the findings
In the previous sections we presented how we have prepared and conducted the
empirical part of this research. Yin (2003) has identified construct validity, internal
validity, external validity, and reliability as general criteria for evaluating case study
research.
In this section, we address these criteria in more detail.
4.7.1 Construct validity
Construct validity is concerned with establishing correct operational measures for
the concepts being studied (Yin, 2003). In this research, low construct validity would
mean that we have not measured the various aspects and practices of modularity in the
field of long-term care for the elderly as well as their relationships to demand-based
care provision accurately. Potential threats to construct validity are respondent bias,
wrong
conceptualization, and incomplete information.
We employed several tactics to support the construct validity in this study. First,
multiple sources of evidence were used. We used different methods for data collection
to triangulate our findings, and we used multiple respondents with different internal
perspectives to avoid partial and biased views in the data obtained. Second, to guard
against incomplete or incorrect information, all interviews were sent back to the
interviewees for a member check. Third, we combined theoretical sources with
82
Modular
Care Provision
suggestions from academics and industry experts throughout the process of data
collection and analysis. Moreover, the industry experts were used to corroborate our
findings. Fourth, to avoid researcher bias during data reduction, we used multiple
coders. Finally, we have tried to establish a chain of evidence from our initial research
questions to the analysis procedures, as presented in this chapter. A chain of evidence
clarifies from the beginning of the study what concepts will be studied and how. It
illustrates how the research questions guide the information needs for the cases, which
determine the data collection methods and the interview questions. Furthermore,
it shows how the techniques for data reduction such as coding and data display, are
derived from the research questions and how they guided data analysis (Zomerdijk,
2005, Yin, 2003).
4.7.2 Internal validity
Internal validity has to do with establishing a causal relationship, whereby certain
conditions are shown to lead to other conditions, as distinguished from spurious
relationships (Yin, 2003). Internal validity is mainly relevant to explanatory studies
since it is related to the logic of causal relationships. Although this study is not exactly
an explanatory study in the sense that we tried to determine whether event ‘x’ led to
event ‘y’, we assumed there is a causal relationship between modularity on the one
hand and demand-based care on the other hand. Therefore, we have to ensure that it is
really modularity, or its various aspects and practices, that bring about the operational
dimensions of demand-based care.
To uphold the internal validity of this study, we employed the following tactics: we
started with a relatively open conceptual model to prevent bias from predetermined
findings and relationships and to enable the insights on modularity in long-term care
to arise from the case studies. In addition, we looked for and analyzed patterns across
cases and verified our findings with our own data. Finally, we compared and matched
our empirical observations and findings with existing literature in order to explain the
observed
patterns, as such tying our results to existing insights.
4.7.3 External validity
External validity is related to establishing the domain to which a study’s findings can
be generalized beyond the immediate case studies (Yin, 2003). Case study research
is often criticized for offering a poor basis for generalization since it is deemed too
situation specific. Yin (2003) points out that generalizing from cases takes place
according to analytical generalizations instead of statistical generalizations. This means
that one concentrates on the expansion and generalization of theories rather than the
83
Resear
ch desig
n
and methods
Chapt
er 4