1 ), 131 – 150 W r i t i n g

Respond by Day 5 to two colleagues and
explain in what ways you agree or disagree with the permissibility of
the identified scenarios. Justify your reaction. $5 Post

Barbara Geddes makes very good valid points in her article, ‘How
the Cases you Choose Affect the Answers you get: Selection Bias in
Comparative Politics.’ Geddes starts off by expressing how the selection
of the dependent variable can draw bias conclusions. For example, if
someone wanted to prove that race, gender, and income play a role in
recividism rates, then the researcher would draw conclusions based on
just these three variables instead of looking at the bigger picture. The
major problem with selecting cases for the desired study on the
dependent variable comes from the logic of explanation. “Most graduate
students learn in the statistics courses forced upon them that
selection on the dependent variable is forbidden, but few remember why,
or what the implications of violating this taboo are for their own
work.” (Geddes, 1990). The only thing that can actually be explained
using a sample selected on the dependent variable are the differences
among the selected cases. Instead of drawing on three conclusions that
cause recidivism, one should look at every variable to establish what
the actual problem could come from.

The first mistake that researchers can make is jumping to
conclusions that any characteristic that the selected cases share is
from one cause. The other involves assuming that a relationship between
variables within the selected set of cases reflect relationships in the
entire population of the cases. Since the researcher is looking for some
sort of connection, the research will become biased when in fact there
might not be any connection at all, or vice versa.

Geddes expresses interest in how to go about conducting the
correct hypothesis. “The two tasks crucial to testing any hypothesis are
to identify the universe of cases to which the hypothesis should apply,
and to find or develop measures of the variables.” (Geddes, 1990). The
next step is to conduct a random sample that shows the cases are
uncorrelated with the placement of cases on the dependent variables. If
for example, the number of cases are too large for one study, a random
sample should be taken. However, the random cases need to show that they
are uncorrelated.

One scenario that might be okay to choose cases solely on the
dependent variable is to conduct such research that proves a point or
finding. For example, one could base the research off of countries that
have the most people in the military or the most money. These are easy
to establish since there is only one variable that the researcher needs.

Geddes, B. (1990). How the cases you choose affect the answers you get: Selection bias in comparative politics. Political Analysis, 2(1), 131–150. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.372.5896&rep=rep1&type=pdf