Lecture 4 - Research Design Flashcards
What is research design?
Aim of research = to expand boundaries of existing knowledge (whether foundational or contextual or interpretative)
- to advance the state of the art (state of existing knowledge, scientific literature)
Getting valid answers to research questions in a reliable and efficient way
applied (operational) epistemology - how do we know
note to the reading
mainly focuses on positivist research design
no mention of normative theory
little attention to interpretivist theory (because this is much more difficult: there are more varieties than in the precise considerations of positivist research design)
three levels of generality
level
- ontological, epistemological, broad theoretical outlook (is often implicit, but we should try to be explicit about the nature of the world and what we can know about it)
- research goal and question, relation with theory, conceptualization, operationalization, class of research methodology (qualitative, quantitative, both)
*this step is always explicit - cases, variables, evidence
with level 2 we can deduct the ontological epistemological approach
criticism schematic representation of the research process (reading)
research questions and research goals
- puzzles and lacunae
- substantive problems
- practical problems (government can ask questions to political scientists, to figure out the reason for a problem)
orientations towards research questions and research goals
- scientific description
- causal explanation and prediction (= aka positivist)
- understanding/interpretation (=aka interpretative)
combinations is possible, but remember that epistemology and ontology are skins, not sweaters
scientific description
often not considered to be scientifically significant
- analytical dimensions (concepts) often theoretically informed (every researcher, whatever approach, must conceptualize)
- implications for both empirical and normative theory
Two different types of description
- comprehensive representation of one/few case(s) (many variables, often unique(ish) phenomena)
- classification of many cases
classification and description
description goes with classification
‘the value of scientific description is in selecting a set of important analytical dimensions on which to project the empirical world’ (LMS)
classification is an important first step in theory
classifications are imposed on the social/political world
classifications can’t be neutral, they are embedded in the way we understand the world, into our own understanding and education, this element of constructivism, is correct for all approaches: classifications and categories don’t exist in the world
difference between foundationalist and anti-foundationalist approaches to classifications don’t have to do with linguistic classifications, the difference is the things they classify: for anti-foundationalists the things they classify themselves are unstable, for foundationalism the things exist independently in the world
foundationalism: no matter what you call it, you mean the same thing
causal explanation
associated with foundationalist ontologies (behaviouralist, rational choice, Marxist and institutionalist approacches)
two types:
- retrospective
- prospective
causal explanation in social sciences
often phrased in probablistic language instead of deterministic
namely:
if the cause had been different, the outcome would have been different
they claim to be able to say … leads to it being more/less likely that ….
requires large statistical research
interpretivist research (goal)
associated with anti-foundationalist ontologies
reject the scientific concept of causation
political actions are the product of subjective reasons, meanings and beliefs -> we should :
- focus on understanding rather than explanation (Weber)
- focus on constitutive rather than causal arguments (Wendt)
theory development stage
theory comes to play in lots of different moments (e.g. research questions, choosing methodologies, making conclusions, conceptualisation)
- text doesn’t highlight this enough according to Theuns
6 types of engagement of research projects
- mere application of theory
- abductive explanations (back and forth between theoretical standardizations and observations, kind of middle ground between induction and deduction)
- theory development (deductive from premises to propositions)
- theory generation (inductive: patterns in empirical data)
- hypothesis testing
research methodology
IT IS NOT ONE STEP IN THE RESEARCH PROCES
formulating a research question is prior to choosing a research methodology, but the choice is not holy shaped by the question (also by e.g. epistemological and ontological position)
experimental vs observational designs
we can’t always mimic laboraty experiments in PS
- phenomena are to complex to mimic in a lab (e.g. Trumpism)
even if we could, we probably shouldn’t create experiments of creating societies because of ethical reasons (e.g. you can’t just pick to groups and expose them to dire conditions)
we often try to work with observational designs
(no choice of treatment group and control group)o
observational methods
researchers in social sciences can’t always assign treatment and control groups
researchers can control:
- which cases to study
- how may cases to observe
- which features of the cases to observe
- what (sort of) data to collect (qualitative or quantitative)
conceptualization
- concepts are abstract and unobservable (until you have operationalized it and found indicators)
- they are contested
- concepts must be judged by the usefulness and coherence
- once we have a concept we try to operationalize it
- concepts can have an inductive element (how the terms are used in politics/by people) but this has limits
how to conceptualize is also contested
case selection and the selection of variables
how many cases? large-N or small-N
- can be applied to all approaches to some degree, generally speaking the discussion is mostly about positivist design
positivists: the more cases, the better
large-N observational
population = individuals, countries, years, written content
- population determines how many cases are possible to study
ideal = randomly selected sample of cases (to be as representative as possible)
practice = often convenience samples, purposive sampling (choosing for a specific feature of a group)
research doesn’t control ‘treatment’ variation
- try to measure, and thus rule out, all potentially confounding variables (which to select = based on theory)
- try to measure, and thus observe, causal mechanisms
fewer cases: comparative method
comparative politics = not the same as the comparative method
comparative method tends to focus on small-N case studies
it is the go-to method for much empirical political science when:
- experimental methods aren’t possible (no control)
- statistical methods are not possible (not enough data/cases)
Most Different Systems Design
we try to find different cases (small-N) that share a feature that we try to understand or explain, but are for the rest really different
Most Similar Systems Design
cases that are very similar, but have a different outcome
we want to explain how similar cases differ in one variable
general limitations to most similar and most different systems designs
- there are a small number of cases (not really confident in hypotheses)
- methods are sensitive to case selection / variable specification (possible selecting cases to fit theory, instead of to test theory)
- difficult to analyse the interactions between variables
one case: single case design
not enough cases, inadequate prior knowledge of relevant variables -> single case studies
comparison still important: compare to concepts, compare to wider universe of cases and concepts
case studies are often descriptive, but:
- contribute to general knowledge
- contribute to theory generating and theory testing (new phenomena can lead to new knowledge and ideas of theories)
- important steps in general approach to further understanding and explanation