3. Research design: The role of theory Flashcards
What is scientific theory?
“Theory” represents an established understanding of the World - inferred explanation of observed phenomena
How can theory be a tool?
- Frameworks for organising and categorising relevant
information - Schemata for interpreting situations, cases and
statements - Complexity reductions
- A basis for developing hypotheses that can be tested
- Tools for communicating ideas and interpreting
research results
How can theory be like a “lens”?
Theory allows for a different view of empirical
phenomena which help researchers escape
common sense perception and find new aspects
of the phenomenon
Example. Weick’s theory of organizational sense making. Core proposition: Organizations are tied together by shared narratives and meaning
How can theory be used as a lens?
Deductive (largely quantitative) research
Theories used for formulating research
questions,adding coherence to designs;
developing hypotheses; interpreting
results
Inductive (largely qualitative) research
Theories used for formulating research questions,
developing interview guides/selecting data and
above all: interpreting (qualitative) data
What are the four main functions of theory?
- Asking the right questions (design phase)
- Developing relevant testable hypotheses
- Analysing data
- Interpreting results
What is theory?
A theory is a statement of relationships among
concepts within a set of boundary assumptions
and constraints (Bacharach 1989: 496)
An explanatory statement used to help explain
and understand relations among variables, how
they operate and the processes involved
(Kurtines & Silverman 1999 in Kawulich 37)
Name components in theory
Constructs and variables
In between those propositions and hypotheses
Boundary = assumptions about values, time and space
What is a construct?
A construct is a hypothesised property or entity in terms of which the elements of a system or population
are assumed to differ
What is measurement rule?
Once a measurement rule for a construct has been defined (= operationalisation) and the relations
between the elements on the property have been mapped onto a system of numbers or symbols (= scaling), we speak of a variable
Relationships between constructs
The core of a theory are the assumed causal relationships between the constructs. Each causal
relationship (generally) consists of
* The effect that the theory tries to explain
* One or more causes (either contributory, necessary, sufficient, or necessary and sufficient) that
the theory assumes to be ultimately responsible for the effect
* A mechanism (a set of interconnected mediating processes) that transforms the cause(s) into
the effect
Additional boundary conditions may specify where and when the relationship is assumed to
become stronger, weaker or disappear altogether
Name criterias for evaluating theories
Falsifiability
Logical adequacy
Empirical addequacy
What is falsifiability?
The predictions of a theory (i.e., the hypotheses that can logically be deduced from it) should be
falsifiable. This is the case when its predictions are:
* Non-contradictory: if, for example, the theory assumes that X is a necessary and sufficient
condition for Y, it must also assume that ⌐X is a necessary and sufficient condition for ⌐Y
* Non-tautological: a predicted effect must not already be included in the assumed cause
(e.g., this would be the case in ”we predict that companies with a higher level of market
orientation respond more rapidly to competitors’ actions”)
* Measurable: it must be possible to validly operationalise all constructs involved in the
predictions
Explanatory potential and predictive adequacy
Consistency: a theory must not contradict established empirical fact and should, as much as possible,
also be able to predict established empirical fact
Interestingness: a theory should make new and useful predictions
Corroboration: the ratio of empirically corroborated to empirically falsified hypotheses should be as
high as possible
Accuracy: quantitative predictions made by a theory should fit independently collected empirical
data as closely as possible
Parsimony: The theory should be as simple as possible.
What is range?
Boundary conditions should be clear.
More wider ranging theories more abstract
More narrow ranging theories more specific
