{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_53363"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_53363","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"THREE PAPERS ON DESIGN UNCERTAINTY FOR LINEAR ESTIMATORS","abstract":"Thesis (Ph.D.)--Michigan State University. Economics - Doctor of Philosophy, 2026","abstract_html":"Thesis (Ph.D.)--Michigan State University. Economics - Doctor of Philosophy, 2026","abstract_has_math":false,"creators":["Danko, Robin"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Wooldridge, Jeffrey M.","Galvao, Antonio","Vogelsang, Tim","Kachkovskiy, Ilya"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T18:59:47Z","subjects":["Economics"],"languages":["eng"],"rights":["Attribution 4.0 International"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:53363","local:Danko_grad.msu_0128D_22419"],"render_values":[{"text":"etd:53363","href":null,"code":true},{"text":"local:Danko_grad.msu_0128D_22419","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/g6jf-8c55","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/g6jf-8c55","record_label":"Repository record","record_kind":"doi","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://d.lib.msu.edu/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Ad.lib.msu.edu%3Aetd_53363","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Wooldridge, Jeffrey M.","Galvao, Antonio","Vogelsang, Tim","Kachkovskiy, Ilya"]},{"key":"dc:creator","label":"Author","values":["Danko, Robin"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026"]},{"key":"dc:relation","label":"Dc Relation","values":["Electronic Theses & Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["text","Electronic dissertations"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Economics"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution 4.0 International"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:53363","local:Danko_grad.msu_0128D_22419","https://doi.org/doi:10.25335/g6jf-8c55"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (Ph.D.)--Michigan State University. Economics - Doctor of Philosophy, 2026","This dissertation studies design-based uncertainty in linear estimators: uncertainty that arisesnot from sampling variation alone, but from the fact that a researcher observes only one of many possible counterfactual outcomes for each sampled unit. I show that conventional heteroskedasticity and cluster-robust standard errors account only for sampling-based uncertainty, are conservative in this setting when a non-negligible share of a population of interest is sampled. I develop alternative variance estimators that consistently estimate the population variance. The first paper considers heteroskedasticity-robust inference for two-stage least squares (2SLS) estimation when treatment effects are heterogeneous across units and a significant share of the population is sampled. Building on the finite-population framework of Abadie et al. (2020), I show that the usual Eicker-Huber-White standard errors overstate the true sampling variability of the 2SLS estimator in this setting, and I derive the correct asymptotic variance along with an estimator that is less conservative than the Eicker-Huber-White standard errors. I characterize the conditions under which the causal interpretation of the 2SLS coefficient differs from its descriptive counterpart, and show that the two coincide under a constant-treatment-effect restriction. The second paper turns to cluster-robust inference when treatment effects vary across clusters and the number of clusters, while large, is itself finite rather than growing without bound. I model both the population and the sample as containing a finite number of clusters and derive the resulting asymptotic distribution of the OLS estimator. I show that the standard cluster-robust variance estimator is conservative in this setting and propose an alternative that adapts to the observed ratio of sampled to population clusters, nesting the standard estimator as a special case when that ratio is negligible. The third paper extends this design-based, finite-cluster framework to settings with endoge nous treatment, where identification relies on instrumental variables. I derive the asymptotic prop erties of both two-stage least squares and generalized method of moments estimators under finite population, finite-cluster sampling, and show that the same conservativeness result extends to the usual cluster-robust standard errors used with IV estimators. I propose consistent variance estima tors for both the 2SLS and GMM cases and characterize the conditions under which each estimator is preferred. Together, these papers argue that the choice between sampling-based and design-based uncer tainty materially affects the size of standard errors and the resulting inference in applications where a substantial fraction of the population of interest is observed.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf","1 online resource (136 p.) : ill."]},{"key":"dc:title","label":"Title","values":["THREE PAPERS ON DESIGN UNCERTAINTY FOR LINEAR ESTIMATORS"]}]}],"canonical_facts":{"dc:contributor":["Wooldridge, Jeffrey M.","Galvao, Antonio","Vogelsang, Tim","Kachkovskiy, Ilya"],"dc:creator":["Danko, Robin"],"dc:date":["2026"],"dc:description":["Thesis (Ph.D.)--Michigan State University. Economics - Doctor of Philosophy, 2026","This dissertation studies design-based uncertainty in linear estimators: uncertainty that arisesnot from sampling variation alone, but from the fact that a researcher observes only one of many possible counterfactual outcomes for each sampled unit. I show that conventional heteroskedasticity and cluster-robust standard errors account only for sampling-based uncertainty, are conservative in this setting when a non-negligible share of a population of interest is sampled. I develop alternative variance estimators that consistently estimate the population variance. The first paper considers heteroskedasticity-robust inference for two-stage least squares (2SLS) estimation when treatment effects are heterogeneous across units and a significant share of the population is sampled. Building on the finite-population framework of Abadie et al. (2020), I show that the usual Eicker-Huber-White standard errors overstate the true sampling variability of the 2SLS estimator in this setting, and I derive the correct asymptotic variance along with an estimator that is less conservative than the Eicker-Huber-White standard errors. I characterize the conditions under which the causal interpretation of the 2SLS coefficient differs from its descriptive counterpart, and show that the two coincide under a constant-treatment-effect restriction. The second paper turns to cluster-robust inference when treatment effects vary across clusters and the number of clusters, while large, is itself finite rather than growing without bound. I model both the population and the sample as containing a finite number of clusters and derive the resulting asymptotic distribution of the OLS estimator. I show that the standard cluster-robust variance estimator is conservative in this setting and propose an alternative that adapts to the observed ratio of sampled to population clusters, nesting the standard estimator as a special case when that ratio is negligible. The third paper extends this design-based, finite-cluster framework to settings with endoge nous treatment, where identification relies on instrumental variables. I derive the asymptotic prop erties of both two-stage least squares and generalized method of moments estimators under finite population, finite-cluster sampling, and show that the same conservativeness result extends to the usual cluster-robust standard errors used with IV estimators. I propose consistent variance estima tors for both the 2SLS and GMM cases and characterize the conditions under which each estimator is preferred. Together, these papers argue that the choice between sampling-based and design-based uncer tainty materially affects the size of standard errors and the resulting inference in applications where a substantial fraction of the population of interest is observed.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."],"dc:format":["application/pdf","1 online resource (136 p.) : ill."],"dc:identifier":["etd:53363","local:Danko_grad.msu_0128D_22419","https://doi.org/doi:10.25335/g6jf-8c55"],"dc:language":["eng"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["Attribution 4.0 International"],"dc:subject":["Economics"],"dc:title":["THREE PAPERS ON DESIGN UNCERTAINTY FOR LINEAR ESTIMATORS"],"dc:type":["text","Electronic dissertations"]},"updated_at":"2026-11-18T18:59:47Z"}