# Ragin's comparative method by wuyunyi

VIEWS: 10 PAGES: 7

• pg 1
```									Ragin’s comparative method
Characteristics comparative
method
• Combinations of conditions are attributed causal value
• Cases are studied as unique combinations of conditions
and are thus left intact
• Explanations are absolute in that they cover all instances
of a phenomenon (no exceptions)
• Examines all the cases of a population (does not
generalize from sample to population)
• Makes use of categorical variables with two values:
present or absent; high or low; + or -
Characteristics statistical method
• Analytical (effect of each individual variable is
determined independently of other variables)
• Cases are no more than the bearers of variables
• Generalises from sample to population
• Explanations are not absolute but probabilistic (i.e.
outliers and exceptions are accepted as long as there
are not too many)
• Frequency is important (an explanation is stronger the
more instances it covers)
• Intensity of phenomenon is taken into account (ordinal
and continuous variables)
When is Ragin’s comparative
method a suitable approach?
• If explanatory unit of research is at a group level
(school, municipality, region, country);
• If there are few units (e.g. less than 20);
• If the response variable is categorical with binary
values (or can easily be turned into it);
• If most of the explanatory variables presumed
important are binary or can be turned into binary
ones;
• When you are interested in multiple causation;
• To construct an empirical typology (pp 149-160
of Ragin’s book)
Ragin’s comparative method: how
does it work?
A step by step approach:
• Select cases and variables relevant to research interest
and hypotheses
• Turn selected variables into binary variables and define
values;
• Assign present and absent values to each case on these
variables by using upper and lower case letters
• Compile these values in a data matrix;
• Transform the data matrix into a truth table (p. 88 of
Ragin’s book) A truth table lists all the logically different
combinations of values of the independent variables
found in the sample
• Use Boolean algebra to arrive at ‘primitive’ causal
equation
• Applicable in situations of limited cases;
causes;
• The integrity of cases as (unique) combinations
of properties is respected;
• No problems with generalization from sample to
population;
• Gives powerful explanations that cover all cases;
• Convenient tool for constructing typologies.
• Does not distinguish between lower and higher
level variables
• Difficult to transform higher level variables into
binary ones. Problems:
– distance between original values
– variables with a normal distribution
• Frequency not taken into account in assessing
strength of explanations
• Strict explanations covering all instances may
contain so many combinations that interpretation
becomes difficult (i.e. transparency and
parsimony suffer)

```
To top