WEEK 2- L4.2 research design overview Flashcards
what’s a research design?
blueprint for;
1. **investigating the research question in a coherent and logical way **(type of research- normative, philosophical, empirical or more exploratory or confirmatory)
2. **including what kinds of data is needed **(for test, not confirmation)
3. how the data is collected (methods of data collection)
4. and what methods of analysis will be used (mode of analysis- logic to draw inferences)
why is the explaratory-confirmatory distinction problematic?
- exploratory research is less common and appreciated- can’t be just descriptive summary needs some theory development. confirmatory research draws on existing theories and applies and tests these- makes it the default choice
- the word confirmatory sounds like you will only look at supporting evidence
what’s research design- list:
- type of research
- evidence or data needed
- method of data collection
- mode of analysis (logic to draw inferences)
- how threats to internal and external validity and measurement reliability are minimized
research design types
- experimental designs
- cross sectional or longtidunial designs (observation)
- comparative designs
- historical designs
idea behind experimental designs…
isolate the explanatory or causal factor, and randomly assigned them to the unit of analysis (radomized intervention)
types of experimental designs
- laboratory ruling out any other influence factors)
-
field done in the real world with a randomized intervention but less control over other factors
3.** survey experiment** different order or phrase leading to difference - natural experiments interventions or treatments can occur naturally- when two similar cities enact different policies- there’s no random assignment or intervention- researchers draw conclusions about the policies. results can’t be conclusively attributed to the intervention
what’s the problem with natural experiments?
results can’t be conclusively attributed to the intervention because there was no random assignment.
it fails the criteron of internal validity
=> it’s not an experiment
what’s crossectional?
comparison of units of analysis at the same time
what’s longtidunial?
implies a change over time
when are observational designs (cross-long) used?
in large n research
quanitative
panel and cohort designs add a…
time dimension to the design
what’s a panel?
same units are observed repeatedly
what’s a cohort?
doesn’t trace the same units, but repeats the same measurements/ observations over time with new samples.
what does a panel study allow?
tracing individual changes over time
comparative designs are used in…
single case studies => in depth process tracing
small n and large n case study