Population and Sample Uncertainty
Year: 2012
Type: preprint
Abstract: Causal inference seeks to identify a causal mechanism and the set of cases for which the causal claim makes sense. It is always based on a specific sample analyzed. However, whether results can be reliably generalized to the population depends on whether the sample represents a true random draw from the population. In principle, the population can be deductively derived as the set of cases for which a theory claims validity or inductively derived as the set of cases for which the studied sample could have represented a random draw. However, researchers are typically uncertain about what constitutes the relevant population for testing their hypotheses as well as uncertain about what population their studied sample might represent. Both uncertainties render the underlying assumption of econometrics – that one studies a random draw from the population – fictional. We explore the principal sources of population and sample uncertainty and discuss the inferential threats they pose. We suggest tests for probing the robustness of inferences toward population and sample uncertainty, illustrating our suggestions with an example from the study of civil war onset.
Source: SSRN Electronic Journal
Authors: Thomas Pluemper, Eric Neumayer
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2291419730