Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Case studies I & II

No description
by

Erika Forsberg

on 27 August 2015

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Case studies I & II

Case studies I & II
10 November, 2014

Erika Forsberg
Department of Peace and Conflict Research
erika.forsberg@pcr.uu.se

Outline
theme of week 2
what is a case?
population?
observation?
case study research
strategies for selecting cases in small-N research
What are relevant scope conditions for this causal statement?
From which population would you consider selecting cases?
Mon 10/11 10.15-12.00 Case studies I
Mon 10/11 13.15-1500 Case studies II Erika Forsberg



Tue 11/11 10.15-12.00 Process-tracing Ludvig Norman




Fri 14/11 Seminar Assignment 2
Erika's slides
Gerring (2007)
(Elster)
Samii
Ludvig's slides
Bennett & Checkel
(Elster)
Hedström & Swedberg
Pierson
see assignment 2
Week 2 - overview
This week is dedicated to case studies, i.e. the intense study of one or a few cases. Key terms and themes include: what is a case; what is a case study; types of case studies; the strengths and weaknesses of case studies; selecting cases from a population; process-tracing.
Before lunch
After lunch
https://medium.com/message/that-catcalling-video-and-why-research-methods-is-such-an-exciting-topic-really-32223ac9c9e8
Implicit "research question":
"Do women get verbally harassed?"
or perhaps:
"Do conventionally attractive white women get verbally harassed while walking the streets of NYC?"
Collection of "data" not guided by theory.
As a result, "evidence" is compatible with multiple hypotheses:
H1: Men of color are more likely to catcall.
H2: All men are equally likely to catcall, but the makers of the video were biased, consciously or unconsciously, against men of color

Unsystematic collection of data:
not purposeful --> useless
with a bias --> highly problematic

H3: It's a spurious relationship
Recap from last week
The goal of social science: identifying, describing and explaining (and predicting) phenomena and patterns in the social world.

What is a case study?
little consensus in existing literature
some claim it does not involve quantitative analysis
others claim that it strives towards a comprehensive description of a phenomenon
others equate it with some data-gathering technique
and so on...
Gerring's definition of a case study

Gerring (2007) tries to construct a definition that resonates with established usage:
"A case study investigates the properties of one or a few cases intensively"

Assumptions:
(1) a case is a "case of something"
(2) always involves comparison
(3) causal ambitions
When the cases are so many that it is no longer possible to investigate each intensively
When the focus shifts from the properties of single cases to those of a sample of cases
When this point is passed: cross-case study
When is it no longer a case study?
When do we use case studies?
1. Descriptive studies - compare different cases to better describe phenomena
2. Explanatory (causal) studies
(a) theory (hypothesis) testing studies: compare different cases to examine whether a certain factor explain a phenomenon
(b) theory (hypothesis) generating studies: compare different cases to examine whether additional factors can be identified in order to develop theory further

We will follow Gerring and focus on explanatory studies

1. Typical case (1+)
2. Diverse (2+)
3.Extreme (1+)
4. Deviant (1+)
5. Influential (1+)
6. Crucial (1+)
most likely
least likely
7. Pathway (1+)
8. Most similar/"method of difference" (2+)
9. Most different/"method of agreement" (2+)
Techniques for selecting cases
purposeful/strategic, not random (all case studies)
aims to be representative in some sense; selected due to its (assumed) status within an (assumed) broader population of cases
based on existing large-N/cross-case patterns (e.g. typical, diverse, extreme, deviant, influential)
based on theoretical assumptions and/or expectations (e.g. most likely, least likely)

Observation 1 of case A
(example: political party A at time t1)
Observation 2 of case A
(example: political party A at time t2)
Observation 3 of case A
(example: political party A at time t3)
Case A
(political party in country X)
Observation 1 of case B
(example: political party B at time t1)
Observation 2 of case B
(example: political party B at time t2)
Observation 3 of case B
(example: political party B at time t3)
Case B
(political party in country X)
Population
(political parties)
Population
(organizations)
An illustration using a dataset

Theory Data/cases/empirics




Methods
1. Conceptualize and theorize about patterns and phenomena in the social world (last week)
2. Identify scope conditions and population
3. Select cases from population
4. Collect data regarding your cases (week 3)
5. Analyze data (week 4-7)

The role of the case
Typical case (1+)
2. Diverse cases (2+)
3. Extreme case (1+)
4. Deviant (1+)
5. Influential (1+)
5. Influential (1+)
6. Crucial/critical case (1+)
(a) Most likely
(b) Least likely
7. Pathway (1+)
8. Most similar design/"method of difference"
9. Most different/"method of agreement"
Most-similar/Method of difference with variation in Y
Most-similar/Method of difference with variation in X
Most different/Method of agreement with variation in Y
Most different/Method of agreement with variation in X
What is a case?
"...a spatially delimited phenomenon (unit) observed at a single point in time or over some period of time. It comprises the type of phenomenon that an inference attempts to explain."

Ex: individuals, regions, countries, parties, etc

importance of inference
a "case study" always involves comparison
What is a population?
cases are "cases of" something - a population/universe
populations are thus comprised by cases
theoretical scope conditions define the population
what is a relevant level of analysis?
what is a relevant unit of analysis?
feasibility often has implications for the identification of a population of cases
Example
A student wants to study the use of violence against civilians during armed conflict.

What is a relevant population to select cases from?
Countries? Conflicts? Rebel groups? Individual rebels/soldiers?

Depends on theory!
What is an observation?
A case is often examined within
- across smaller units or
- over time
Recap from last week:
Suppose we have constructed our causal argument
Moreover, we have carried out our causal analysis
How do we know that we have explained an outcome?
One way is checking to what extent our analysis satisfy the‘four causal criteria’
The causal criteria
The causal criteria
1. covariation/counterfactual difference: do X and Y covary?
2. temporal order: does X precede Y?
3. isolation: can you rule of that some variable Z does not cause both X and Y?
4. mechanism: have you specified how a change in X leads to a change in Y?

X: independent variable, presumed cause; Y: dependent variable, presumed effect
If you can confidently say yes to all of the above, then you can confidently say you have explained an outcome
Case studies vs cross-case studies
1. Covariation: large-N advantage over small-N
(especially with probabilistic causal relationships)
2. Temporal order: small-N advantage over large-N
(opportunity to follow each case closely over time)
3. Isolation: large-N advantage over small-N
(control for more potential confounders)
4. Mechanism: small-N advantage over large-N
(opportunity to uncover and explore causal mechanisms at the micro-level)

Other trade-offs and affinities
Hypothesis
generating
of
testing
?
Internal
or
external
validity?
Causal
mechanism
or causal
effect
?
Deep
or
broad
scope of proposition?
Heterogenous
or
homogenous
population?
Strong
or
weak
causal strength?
Rare
or
common
useful variation?
Data availability
concentrated
or
dispersed
?
And some final problems with cross-case studies
Difficult to model complex causal relationships
Influential cases may obscure average causal effects
Arbitrary significance tests
Random measurement errors are likely
Selected to achieve maximum variation along relevant dimensions
Extreme values on X
or extreme values on Y
or a combination of X and Y
easiest to achieve when variables are categorical (on/off, absence/presence, woman/man)
for continuous variables: choose high, low, and perhaps also mean or median
may be combined with typical case strategy
Selected based on having an extreme value on an independent or dependent variable of interest
it is far away from the mean of that variable (much higher, or much lower)
unusual, but not necessarily a deviant case
not so widely used
A crucial case is a case that “must closely fit a theory if one is to have confidence in the theory´s validity or, conversely, must not fit equally well any rule contrary to that proposed “ (Eckstein 1975).
One tests a theory on a case which is the most likely to conform to theory.

If not supported, then not likely to be supported in other cases either.
One tests a theory on a case that is leastlikely to conform to theory.

If supported, then it is likely to find
support in other cases as well.
Example of most-likely logic
Previous research: theory suggesting that the interaction between high education and unemployment will increase the risk of armed conflict.
Evidence is inconclusive
BA thesis which examines cases of lower level violence
If not supported, then it is very unlikely that it will lead to armed conflict.
The goal is to shed further light on the causal mechanism presumably linking X and Y.
Always combined with cross-case analysis
Select a case that:
(1) is not an outlier
(2) its score on Y is strongly linked to the X of theoretical interest, while controlling for confounding variables
minimum 2 cases
similar in "all" respects except the variables of interest (X and Y)
"all" = relevant respects, that is
a way of isolating the causal relationship
reflects a logic, not necessarily successful as case selection strategy
why? if you know the values of both X and Y, you know the result beforehand
minimum 2 cases
different in "all" respects except the
variables of interest (X and Y)
"all" = relevant respects, that is
a way of improving generalizability
reflects a logic, not necessarily successful as case selection strategy
why? if you know the values of both X and Y, you know the result beforehand
Types of case studies
Focus on spatial or temporal variation?
Focus on variation within and/or between cases?
Similar or diverse - a trade-off
If we select different cases: we may generalize our results to more cases/time-periods, etc but we risk comparing cases that are not really comparable

If we select similar cases: the comparison becomes more reasonable but we limit the number of cases that we are able to generalize our findings to

If we are clever in our case selection, we can sometimes achieve both

Exercise
Suppose you are interested in explaining outcome Y. You have developed four hypotheses about variables (X1-X4) that may impact on Y. You only have time to study 2 cases. Which 2 of these 5 would you select if you want to test X3?
Full transcript