Loading…
On testing marginal versus conditional independence
Summary We consider testing marginal independence versus conditional independence in a trivariate Gaussian setting. The two models are nonnested, and their intersection is a union of two marginal independences. We consider two sequences of such models, one from each type of independence, that are cl...
Saved in:
Published in: | Biometrika 2020-12, Vol.107 (4), p.771-790 |
---|---|
Main Authors: | , |
Format: | Article |
Language: | English |
Subjects: | |
Citations: | Items that this one cites Items that cite this one |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
cited_by | cdi_FETCH-LOGICAL-c301t-249ba8fc6fa9895791a736c71b41195a0e8960c3506e9e45b905440f181023bc3 |
---|---|
cites | cdi_FETCH-LOGICAL-c301t-249ba8fc6fa9895791a736c71b41195a0e8960c3506e9e45b905440f181023bc3 |
container_end_page | 790 |
container_issue | 4 |
container_start_page | 771 |
container_title | Biometrika |
container_volume | 107 |
creator | Guo, F Richard Richardson, Thomas S |
description | Summary
We consider testing marginal independence versus conditional independence in a trivariate Gaussian setting. The two models are nonnested, and their intersection is a union of two marginal independences. We consider two sequences of such models, one from each type of independence, that are closest to each other in the Kullback–Leibler sense as they approach the intersection. They become indistinguishable if the signal strength, as measured by the product of two correlation parameters, decreases faster than the standard parametric rate. Under local alternatives at such a rate, we show that the asymptotic distribution of the likelihood ratio depends on where and how the local alternatives approach the intersection. To deal with this nonuniformity, we study a class of envelope distributions by taking pointwise suprema over asymptotic cumulative distribution functions. We show that these envelope distributions are well behaved and lead to model selection procedures with rate-free uniform error guarantees and near-optimal power. To control the error even when the two models are indistinguishable, rather than insist on a dichotomous choice, the proposed procedure will choose either or both models. |
doi_str_mv | 10.1093/biomet/asaa040 |
format | article |
fullrecord | <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_journals_2476152396</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><oup_id>10.1093/biomet/asaa040</oup_id><sourcerecordid>2476152396</sourcerecordid><originalsourceid>FETCH-LOGICAL-c301t-249ba8fc6fa9895791a736c71b41195a0e8960c3506e9e45b905440f181023bc3</originalsourceid><addsrcrecordid>eNqFkL1PwzAQxS0EEqGwMkdiYkh7F38kHlEFBalSF5gtx3UqV60d7ASJ_55E6c5ypzu9d3r3I-QRYYkg6apx4Wz7lU5aA4MrkiETrKAc4ZpkACAKyhi7JXcpHadRcJERuvN5b1Pv_CE_63hwXp_yHxvTkHIT_N71Lkwr5_e2s2Pxxt6Tm1afkn249AX5env9XL8X293mY_2yLQwF7IuSyUbXrRGtlrXklURdUWEqbBii5BpsLQUYykFYaRlvJHDGoMUaoaSNoQvyNN_tYvgexpDqGIY4pkmqZJVAXlIpRtVyVpkYUoq2VV104yu_CkFNYNQMRl3AjIbn2RCG7j_tH0jDZVI</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2476152396</pqid></control><display><type>article</type><title>On testing marginal versus conditional independence</title><source>Oxford Journals Online</source><creator>Guo, F Richard ; Richardson, Thomas S</creator><creatorcontrib>Guo, F Richard ; Richardson, Thomas S</creatorcontrib><description>Summary
We consider testing marginal independence versus conditional independence in a trivariate Gaussian setting. The two models are nonnested, and their intersection is a union of two marginal independences. We consider two sequences of such models, one from each type of independence, that are closest to each other in the Kullback–Leibler sense as they approach the intersection. They become indistinguishable if the signal strength, as measured by the product of two correlation parameters, decreases faster than the standard parametric rate. Under local alternatives at such a rate, we show that the asymptotic distribution of the likelihood ratio depends on where and how the local alternatives approach the intersection. To deal with this nonuniformity, we study a class of envelope distributions by taking pointwise suprema over asymptotic cumulative distribution functions. We show that these envelope distributions are well behaved and lead to model selection procedures with rate-free uniform error guarantees and near-optimal power. To control the error even when the two models are indistinguishable, rather than insist on a dichotomous choice, the proposed procedure will choose either or both models.</description><identifier>ISSN: 0006-3444</identifier><identifier>EISSN: 1464-3510</identifier><identifier>DOI: 10.1093/biomet/asaa040</identifier><language>eng</language><publisher>Oxford: Oxford University Press</publisher><subject>Asymptotic properties ; Distribution functions ; Independence ; Likelihood ratio ; Nonuniformity ; Signal strength</subject><ispartof>Biometrika, 2020-12, Vol.107 (4), p.771-790</ispartof><rights>2020 Biometrika Trust 2020</rights><rights>2020 Biometrika Trust</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c301t-249ba8fc6fa9895791a736c71b41195a0e8960c3506e9e45b905440f181023bc3</citedby><cites>FETCH-LOGICAL-c301t-249ba8fc6fa9895791a736c71b41195a0e8960c3506e9e45b905440f181023bc3</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids></links><search><creatorcontrib>Guo, F Richard</creatorcontrib><creatorcontrib>Richardson, Thomas S</creatorcontrib><title>On testing marginal versus conditional independence</title><title>Biometrika</title><description>Summary
We consider testing marginal independence versus conditional independence in a trivariate Gaussian setting. The two models are nonnested, and their intersection is a union of two marginal independences. We consider two sequences of such models, one from each type of independence, that are closest to each other in the Kullback–Leibler sense as they approach the intersection. They become indistinguishable if the signal strength, as measured by the product of two correlation parameters, decreases faster than the standard parametric rate. Under local alternatives at such a rate, we show that the asymptotic distribution of the likelihood ratio depends on where and how the local alternatives approach the intersection. To deal with this nonuniformity, we study a class of envelope distributions by taking pointwise suprema over asymptotic cumulative distribution functions. We show that these envelope distributions are well behaved and lead to model selection procedures with rate-free uniform error guarantees and near-optimal power. To control the error even when the two models are indistinguishable, rather than insist on a dichotomous choice, the proposed procedure will choose either or both models.</description><subject>Asymptotic properties</subject><subject>Distribution functions</subject><subject>Independence</subject><subject>Likelihood ratio</subject><subject>Nonuniformity</subject><subject>Signal strength</subject><issn>0006-3444</issn><issn>1464-3510</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><recordid>eNqFkL1PwzAQxS0EEqGwMkdiYkh7F38kHlEFBalSF5gtx3UqV60d7ASJ_55E6c5ypzu9d3r3I-QRYYkg6apx4Wz7lU5aA4MrkiETrKAc4ZpkACAKyhi7JXcpHadRcJERuvN5b1Pv_CE_63hwXp_yHxvTkHIT_N71Lkwr5_e2s2Pxxt6Tm1afkn249AX5env9XL8X293mY_2yLQwF7IuSyUbXrRGtlrXklURdUWEqbBii5BpsLQUYykFYaRlvJHDGoMUaoaSNoQvyNN_tYvgexpDqGIY4pkmqZJVAXlIpRtVyVpkYUoq2VV104yu_CkFNYNQMRl3AjIbn2RCG7j_tH0jDZVI</recordid><startdate>20201201</startdate><enddate>20201201</enddate><creator>Guo, F Richard</creator><creator>Richardson, Thomas S</creator><general>Oxford University Press</general><general>Oxford Publishing Limited (England)</general><scope>AAYXX</scope><scope>CITATION</scope><scope>7QO</scope><scope>8FD</scope><scope>FR3</scope><scope>P64</scope></search><sort><creationdate>20201201</creationdate><title>On testing marginal versus conditional independence</title><author>Guo, F Richard ; Richardson, Thomas S</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c301t-249ba8fc6fa9895791a736c71b41195a0e8960c3506e9e45b905440f181023bc3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Asymptotic properties</topic><topic>Distribution functions</topic><topic>Independence</topic><topic>Likelihood ratio</topic><topic>Nonuniformity</topic><topic>Signal strength</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Guo, F Richard</creatorcontrib><creatorcontrib>Richardson, Thomas S</creatorcontrib><collection>CrossRef</collection><collection>Biotechnology Research Abstracts</collection><collection>Technology Research Database</collection><collection>Engineering Research Database</collection><collection>Biotechnology and BioEngineering Abstracts</collection><jtitle>Biometrika</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Guo, F Richard</au><au>Richardson, Thomas S</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>On testing marginal versus conditional independence</atitle><jtitle>Biometrika</jtitle><date>2020-12-01</date><risdate>2020</risdate><volume>107</volume><issue>4</issue><spage>771</spage><epage>790</epage><pages>771-790</pages><issn>0006-3444</issn><eissn>1464-3510</eissn><abstract>Summary
We consider testing marginal independence versus conditional independence in a trivariate Gaussian setting. The two models are nonnested, and their intersection is a union of two marginal independences. We consider two sequences of such models, one from each type of independence, that are closest to each other in the Kullback–Leibler sense as they approach the intersection. They become indistinguishable if the signal strength, as measured by the product of two correlation parameters, decreases faster than the standard parametric rate. Under local alternatives at such a rate, we show that the asymptotic distribution of the likelihood ratio depends on where and how the local alternatives approach the intersection. To deal with this nonuniformity, we study a class of envelope distributions by taking pointwise suprema over asymptotic cumulative distribution functions. We show that these envelope distributions are well behaved and lead to model selection procedures with rate-free uniform error guarantees and near-optimal power. To control the error even when the two models are indistinguishable, rather than insist on a dichotomous choice, the proposed procedure will choose either or both models.</abstract><cop>Oxford</cop><pub>Oxford University Press</pub><doi>10.1093/biomet/asaa040</doi><tpages>20</tpages></addata></record> |
fulltext | fulltext |
identifier | ISSN: 0006-3444 |
ispartof | Biometrika, 2020-12, Vol.107 (4), p.771-790 |
issn | 0006-3444 1464-3510 |
language | eng |
recordid | cdi_proquest_journals_2476152396 |
source | Oxford Journals Online |
subjects | Asymptotic properties Distribution functions Independence Likelihood ratio Nonuniformity Signal strength |
title | On testing marginal versus conditional independence |
url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-24T17%3A13%3A16IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=On%20testing%20marginal%20versus%20conditional%20independence&rft.jtitle=Biometrika&rft.au=Guo,%20F%20Richard&rft.date=2020-12-01&rft.volume=107&rft.issue=4&rft.spage=771&rft.epage=790&rft.pages=771-790&rft.issn=0006-3444&rft.eissn=1464-3510&rft_id=info:doi/10.1093/biomet/asaa040&rft_dat=%3Cproquest_cross%3E2476152396%3C/proquest_cross%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c301t-249ba8fc6fa9895791a736c71b41195a0e8960c3506e9e45b905440f181023bc3%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2476152396&rft_id=info:pmid/&rft_oup_id=10.1093/biomet/asaa040&rfr_iscdi=true |