PE Case 2

explain in what ways you agree or disagree with the permissibility of the identified scenarios. Justify your reaction.

Please provide references
Case 2: Barbara Geddes (1990) points out the pitfalls of selecting cases, units, or observations purely on the basis of the dependent variable.
A. Post by Day 3a brief explanation of the relevance of Barbara Geddes’ argument for program evaluation.
To select on the dependent variable is a description of a researcher confining a set of observations to units of analysis in which an occurrence under study has been noticed, without the inclusion of units in which that phenomenon was not identified. Extrapolations referred to the causes of the occurrence will therefore lack validity. In other words, the outcome would misrepresent the result (Kahan, 2013).
Geddes (1990) asserted that there is a fallacy in construing that the selection of cases based solely on the independent is plausible. Basically, a hypothetical example is evident in an assumption that welfare programs in the states of Connecticut and Massachusetts are successful because they have recipients who are more determined to leave the program, and in contrast programs in Florida and North Carolina are not because the recipients in those states are not determined to leave the program, is purely that, an assumption. It is scientifically flawed to make that determination based on the collection of “one side of the argument”, i.e. data collection based on cases in Massachusetts and Connecticut only, and not on cases in Florida and North Carolina (the other half of the story). Fundamentally, establishing a premise based on only half the information will breed ignorance about “whether or not the factors identified are crucial antecedents of the outcome under investigation” (p. 132). How does this relate to program evaluation? The answer may lie in the emergence of poor generalization. Geddes augmented that there will be fostered, an assumption that the association (or lack of) between factors (variables) within the units of analysis mirror the same associations in the whole set of cases (Geddes, 1990). The author expounded that this type of selection—based solely on the dependent variable—leads to statistical biases, where the result is touted as a lack of association, when in fact there may be a relationship.
B. Then, explain when a scenario of choosing cases solely on the dependent variable might be permissible. Provide a rationale for your explanations.
Though sampling solely on the dependent variable may nurture false inferences, sample selection bias, there are cases when selecting on the dependent variable is not necessarily an adverse thing (Forgues, 2012). The researcher asserted that to totally dismiss its application can in some situations be entirely unfounded. Forgues claimed that selecting on the dependent variable is an alluring technique “when one is looking for the causes of an observed event” (p. 270). It is principally stimulating when the phenomenon under study is unusual, thus resulting in random sampling being impracticable. The use of a case control design can be helpful in proving this (Please see Forgues, 2012).


Forgues, B. (2012). Sampling on the dependent variable is not always that bad: Quantitative
case-control designs for strategic organization research. Strategic Organization, 10(3), 269. Retrieved from
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
Kahan, D. (2013, November 19). Don’t select on the dependent variable in studying the science
communication problem [Blog post]. Retrieved from