Identifying causal effects with proxy variables of an unmeasured confounder

We consider a causal effect that is confounded by an unobserved variable, but for which observed proxy variables of the confounder are available. We show that with at least two independent proxy variables satisfying a certain rank condition, the causal effect can be nonparametrically identified, eve...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Veröffentlicht in:	Biometrika 2018-12, Vol.105 (4), p.987-993
Hauptverfasser:	MIAO, WANG, GENG, ZHI, TCHETGEN TCHETGEN, ERIC J.
Format:	Artikel
Sprache:	eng
Schlagworte:	Error analysis Independent variables Miscellanea Null hypothesis Variables
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

container_end_page	993
container_issue	4
container_start_page	987
container_title	Biometrika
container_volume	105
creator	MIAO, WANG GENG, ZHI TCHETGEN TCHETGEN, ERIC J.
description	We consider a causal effect that is confounded by an unobserved variable, but for which observed proxy variables of the confounder are available. We show that with at least two independent proxy variables satisfying a certain rank condition, the causal effect can be nonparametrically identified, even if the measurement error mechanism, i.e., the conditional distribution of the proxies given the confounder, may not be identified. Our result generalizes the identification strategy of Kuroki & Pearl (2014), which rests on identification of the measurement error mechanism. When only one proxy for the confounder is available, or when the required rank condition is not met, we develop a strategy for testing the null hypothesis of no causal effect.
doi_str_mv	10.1093/biomet/asy038
format	Article
fullrecord	<record><control><sourceid>jstor_pubme</sourceid><recordid>TN_cdi_pubmedcentral_primary_oai_pubmedcentral_nih_gov_7746017</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><jstor_id>48546362</jstor_id><oup_id>10.1093/biomet/asy038</oup_id><sourcerecordid>48546362</sourcerecordid><originalsourceid>FETCH-LOGICAL-c470t-494c016fbf89b8e85b0f17a133c82c09b1fdeafd5a136401523c10d28cd591063</originalsourceid><addsrcrecordid>eNqFkc1P3DAQxa0KVLa0R44gS71wCYzjjyQXJISgRSD10p4tx7HBq8Re7Hjp_vcEQreUS0-jmfnp6T09hA4InBBo6GnrwmDGU5U2QOsPaEGYYAXlBHbQAgBEQRlje-hTSsvnVXDxEe1RShmdtgW6ue6MH53dOH-HtcpJ9dhYa_SY8KMb7_Eqht8bvFbRqbY3CQeLlcfZD0alHE2HdfA2ZN-Z-BntWtUn8-V17qNfV5c_L74Xtz--XV-c3xaaVTAWrGEaiLCtrZu2NjVvwZJKEUp1XWpoWmI7o2zHp5NgQHhJNYGurHXHGwKC7qOzWXeV28F0egoQVS9X0Q0qbmRQTv778e5e3oW1rComgFSTwPGrQAwP2aRRDi5p0_fKm5CTLFlFOOUlkAn9-g5dhhz9FE-WFERJ6-bFUTFTOoaUorFbMwTkc01yrknONU380dsEW_pPL38dhrz6r9bhjC7TGOIWZjVngk4GnwDKSahl</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2306238906</pqid></control><display><type>article</type><title>Identifying causal effects with proxy variables of an unmeasured confounder</title><source>Jstor Complete Legacy</source><source>Oxford University Press Journals All Titles (1996-Current)</source><source>JSTOR Mathematics & Statistics</source><creator>MIAO, WANG ; GENG, ZHI ; TCHETGEN TCHETGEN, ERIC J.</creator><creatorcontrib>MIAO, WANG ; GENG, ZHI ; TCHETGEN TCHETGEN, ERIC J.</creatorcontrib><description>We consider a causal effect that is confounded by an unobserved variable, but for which observed proxy variables of the confounder are available. We show that with at least two independent proxy variables satisfying a certain rank condition, the causal effect can be nonparametrically identified, even if the measurement error mechanism, i.e., the conditional distribution of the proxies given the confounder, may not be identified. Our result generalizes the identification strategy of Kuroki & Pearl (2014), which rests on identification of the measurement error mechanism. When only one proxy for the confounder is available, or when the required rank condition is not met, we develop a strategy for testing the null hypothesis of no causal effect.</description><identifier>ISSN: 0006-3444</identifier><identifier>EISSN: 1464-3510</identifier><identifier>DOI: 10.1093/biomet/asy038</identifier><identifier>PMID: 33343006</identifier><language>eng</language><publisher>England: Oxford University Press</publisher><subject>Error analysis ; Independent variables ; Miscellanea ; Null hypothesis ; Variables</subject><ispartof>Biometrika, 2018-12, Vol.105 (4), p.987-993</ispartof><rights>2018 Biometrika Trust</rights><rights>2018 Biometrika Trust 2018</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c470t-494c016fbf89b8e85b0f17a133c82c09b1fdeafd5a136401523c10d28cd591063</citedby><cites>FETCH-LOGICAL-c470t-494c016fbf89b8e85b0f17a133c82c09b1fdeafd5a136401523c10d28cd591063</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktopdf>$$Uhttps://www.jstor.org/stable/pdf/48546362$$EPDF$$P50$$Gjstor$$H</linktopdf><linktohtml>$$Uhttps://www.jstor.org/stable/48546362$$EHTML$$P50$$Gjstor$$H</linktohtml><link.rule.ids>230,314,776,780,799,828,881,1578,27901,27902,57992,57996,58225,58229</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/33343006$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>MIAO, WANG</creatorcontrib><creatorcontrib>GENG, ZHI</creatorcontrib><creatorcontrib>TCHETGEN TCHETGEN, ERIC J.</creatorcontrib><title>Identifying causal effects with proxy variables of an unmeasured confounder</title><title>Biometrika</title><addtitle>Biometrika</addtitle><description>We consider a causal effect that is confounded by an unobserved variable, but for which observed proxy variables of the confounder are available. We show that with at least two independent proxy variables satisfying a certain rank condition, the causal effect can be nonparametrically identified, even if the measurement error mechanism, i.e., the conditional distribution of the proxies given the confounder, may not be identified. Our result generalizes the identification strategy of Kuroki & Pearl (2014), which rests on identification of the measurement error mechanism. When only one proxy for the confounder is available, or when the required rank condition is not met, we develop a strategy for testing the null hypothesis of no causal effect.</description><subject>Error analysis</subject><subject>Independent variables</subject><subject>Miscellanea</subject><subject>Null hypothesis</subject><subject>Variables</subject><issn>0006-3444</issn><issn>1464-3510</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2018</creationdate><recordtype>article</recordtype><recordid>eNqFkc1P3DAQxa0KVLa0R44gS71wCYzjjyQXJISgRSD10p4tx7HBq8Re7Hjp_vcEQreUS0-jmfnp6T09hA4InBBo6GnrwmDGU5U2QOsPaEGYYAXlBHbQAgBEQRlje-hTSsvnVXDxEe1RShmdtgW6ue6MH53dOH-HtcpJ9dhYa_SY8KMb7_Eqht8bvFbRqbY3CQeLlcfZD0alHE2HdfA2ZN-Z-BntWtUn8-V17qNfV5c_L74Xtz--XV-c3xaaVTAWrGEaiLCtrZu2NjVvwZJKEUp1XWpoWmI7o2zHp5NgQHhJNYGurHXHGwKC7qOzWXeV28F0egoQVS9X0Q0qbmRQTv778e5e3oW1rComgFSTwPGrQAwP2aRRDi5p0_fKm5CTLFlFOOUlkAn9-g5dhhz9FE-WFERJ6-bFUTFTOoaUorFbMwTkc01yrknONU380dsEW_pPL38dhrz6r9bhjC7TGOIWZjVngk4GnwDKSahl</recordid><startdate>20181201</startdate><enddate>20181201</enddate><creator>MIAO, WANG</creator><creator>GENG, ZHI</creator><creator>TCHETGEN TCHETGEN, ERIC J.</creator><general>Oxford University Press</general><general>Oxford Publishing Limited (England)</general><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7QO</scope><scope>8FD</scope><scope>FR3</scope><scope>P64</scope><scope>7X8</scope><scope>5PM</scope></search><sort><creationdate>20181201</creationdate><title>Identifying causal effects with proxy variables of an unmeasured confounder</title><author>MIAO, WANG ; GENG, ZHI ; TCHETGEN TCHETGEN, ERIC J.</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c470t-494c016fbf89b8e85b0f17a133c82c09b1fdeafd5a136401523c10d28cd591063</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2018</creationdate><topic>Error analysis</topic><topic>Independent variables</topic><topic>Miscellanea</topic><topic>Null hypothesis</topic><topic>Variables</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>MIAO, WANG</creatorcontrib><creatorcontrib>GENG, ZHI</creatorcontrib><creatorcontrib>TCHETGEN TCHETGEN, ERIC J.</creatorcontrib><collection>PubMed</collection><collection>CrossRef</collection><collection>Biotechnology Research Abstracts</collection><collection>Technology Research Database</collection><collection>Engineering Research Database</collection><collection>Biotechnology and BioEngineering Abstracts</collection><collection>MEDLINE - Academic</collection><collection>PubMed Central (Full Participant titles)</collection><jtitle>Biometrika</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>MIAO, WANG</au><au>GENG, ZHI</au><au>TCHETGEN TCHETGEN, ERIC J.</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Identifying causal effects with proxy variables of an unmeasured confounder</atitle><jtitle>Biometrika</jtitle><addtitle>Biometrika</addtitle><date>2018-12-01</date><risdate>2018</risdate><volume>105</volume><issue>4</issue><spage>987</spage><epage>993</epage><pages>987-993</pages><issn>0006-3444</issn><eissn>1464-3510</eissn><abstract>We consider a causal effect that is confounded by an unobserved variable, but for which observed proxy variables of the confounder are available. We show that with at least two independent proxy variables satisfying a certain rank condition, the causal effect can be nonparametrically identified, even if the measurement error mechanism, i.e., the conditional distribution of the proxies given the confounder, may not be identified. Our result generalizes the identification strategy of Kuroki & Pearl (2014), which rests on identification of the measurement error mechanism. When only one proxy for the confounder is available, or when the required rank condition is not met, we develop a strategy for testing the null hypothesis of no causal effect.</abstract><cop>England</cop><pub>Oxford University Press</pub><pmid>33343006</pmid><doi>10.1093/biomet/asy038</doi><tpages>7</tpages><oa>free_for_read</oa></addata></record>
fulltext	fulltext
identifier	ISSN: 0006-3444
ispartof	Biometrika, 2018-12, Vol.105 (4), p.987-993
issn	0006-3444 1464-3510
language	eng
recordid	cdi_pubmedcentral_primary_oai_pubmedcentral_nih_gov_7746017
source	Jstor Complete Legacy; Oxford University Press Journals All Titles (1996-Current); JSTOR Mathematics & Statistics
subjects	Error analysis Independent variables Miscellanea Null hypothesis Variables
title	Identifying causal effects with proxy variables of an unmeasured confounder
url	https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-19T06%3A53%3A54IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-jstor_pubme&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Identifying%20causal%20effects%20with%20proxy%20variables%20of%20an%20unmeasured%20confounder&rft.jtitle=Biometrika&rft.au=MIAO,%20WANG&rft.date=2018-12-01&rft.volume=105&rft.issue=4&rft.spage=987&rft.epage=993&rft.pages=987-993&rft.issn=0006-3444&rft.eissn=1464-3510&rft_id=info:doi/10.1093/biomet/asy038&rft_dat=%3Cjstor_pubme%3E48546362%3C/jstor_pubme%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2306238906&rft_id=info:pmid/33343006&rft_jstor_id=48546362&rft_oup_id=10.1093/biomet/asy038&rfr_iscdi=true