Addressing Extreme Propensity Scores via the Overlap Weights

The popular inverse probability weighting method in causal inference is often hampered by extreme propensity scores, resulting in biased estimates and excessive variance. A common remedy is to trim patients with extreme scores (i.e., remove them from the weighted analysis). However, such methods are...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:American journal of epidemiology Jg. 188; H. 1; S. 250
Hauptverfasser: Li, Fan, Thomas, Laine E
Format: Journal Article
Sprache:Englisch
Veröffentlicht: United States 01.01.2019
Schlagworte:
ISSN:1476-6256, 1476-6256
Online-Zugang:Weitere Angaben
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract The popular inverse probability weighting method in causal inference is often hampered by extreme propensity scores, resulting in biased estimates and excessive variance. A common remedy is to trim patients with extreme scores (i.e., remove them from the weighted analysis). However, such methods are often sensitive to the choice of cutoff points and discard a large proportion of the sample. The implications for bias and the precision of the treatment effect estimate are unclear. These problems are mitigated by a newly developed method, the overlap weighting method. Overlap weights emphasize the target population with the most overlap in observed characteristics between treatments, by continuously down-weighting the units in the tails of the propensity score distribution. Here we use simulations to compare overlap weights to standard inverse probability weighting with trimming, in terms of bias, variance, and 95% confidence interval coverage. A range of propensity score distributions are considered, including settings with substantial nonoverlap and extreme values. To facilitate practical implementation, we further provide a consistent estimator for the standard error of the treatment effect estimated using overlap weighting.
AbstractList The popular inverse probability weighting method in causal inference is often hampered by extreme propensity scores, resulting in biased estimates and excessive variance. A common remedy is to trim patients with extreme scores (i.e., remove them from the weighted analysis). However, such methods are often sensitive to the choice of cutoff points and discard a large proportion of the sample. The implications for bias and the precision of the treatment effect estimate are unclear. These problems are mitigated by a newly developed method, the overlap weighting method. Overlap weights emphasize the target population with the most overlap in observed characteristics between treatments, by continuously down-weighting the units in the tails of the propensity score distribution. Here we use simulations to compare overlap weights to standard inverse probability weighting with trimming, in terms of bias, variance, and 95% confidence interval coverage. A range of propensity score distributions are considered, including settings with substantial nonoverlap and extreme values. To facilitate practical implementation, we further provide a consistent estimator for the standard error of the treatment effect estimated using overlap weighting.
The popular inverse probability weighting method in causal inference is often hampered by extreme propensity scores, resulting in biased estimates and excessive variance. A common remedy is to trim patients with extreme scores (i.e., remove them from the weighted analysis). However, such methods are often sensitive to the choice of cutoff points and discard a large proportion of the sample. The implications for bias and the precision of the treatment effect estimate are unclear. These problems are mitigated by a newly developed method, the overlap weighting method. Overlap weights emphasize the target population with the most overlap in observed characteristics between treatments, by continuously down-weighting the units in the tails of the propensity score distribution. Here we use simulations to compare overlap weights to standard inverse probability weighting with trimming, in terms of bias, variance, and 95% confidence interval coverage. A range of propensity score distributions are considered, including settings with substantial nonoverlap and extreme values. To facilitate practical implementation, we further provide a consistent estimator for the standard error of the treatment effect estimated using overlap weighting.The popular inverse probability weighting method in causal inference is often hampered by extreme propensity scores, resulting in biased estimates and excessive variance. A common remedy is to trim patients with extreme scores (i.e., remove them from the weighted analysis). However, such methods are often sensitive to the choice of cutoff points and discard a large proportion of the sample. The implications for bias and the precision of the treatment effect estimate are unclear. These problems are mitigated by a newly developed method, the overlap weighting method. Overlap weights emphasize the target population with the most overlap in observed characteristics between treatments, by continuously down-weighting the units in the tails of the propensity score distribution. Here we use simulations to compare overlap weights to standard inverse probability weighting with trimming, in terms of bias, variance, and 95% confidence interval coverage. A range of propensity score distributions are considered, including settings with substantial nonoverlap and extreme values. To facilitate practical implementation, we further provide a consistent estimator for the standard error of the treatment effect estimated using overlap weighting.
Author Li, Fan
Thomas, Laine E
Author_xml – sequence: 1
  givenname: Fan
  surname: Li
  fullname: Li, Fan
  organization: Duke Clinical Research Institute, Durham, North Carolina
– sequence: 2
  givenname: Laine E
  surname: Thomas
  fullname: Thomas, Laine E
  organization: Duke Clinical Research Institute, Durham, North Carolina
BackLink https://www.ncbi.nlm.nih.gov/pubmed/30189042$$D View this record in MEDLINE/PubMed
BookMark eNpNj0tLAzEYRYNU7EM3_gDJ0s3YPCbJDLgppT6gUEHF5ZA0X9qp8zKZqc6_t2IFV_fAPVy4YzSo6goQuqTkhpKUT_UOpu-fPSP0BI1orGQkmZCDfzxE4xB2hFCaCnKGhpzQJCUxG6HbmbUeQsirDV58tR5KwE--bqAKedvj53V9aPE-17jdAl7twRe6wW-Qb7ZtOEenThcBLo45Qa93i5f5Q7Rc3T_OZ8uoZjJtI-G4FUo7LWjKnDNxYphmqeNCK-qoBR4za81aSEkJcOUci43i0hDKlEk4m6Dr393G1x8dhDYr87CGotAV1F3IGP0xlZDJQb06qp0pwWaNz0vt--zvMfsG0MlZIA
ContentType Journal Article
DBID CGR
CUY
CVF
ECM
EIF
NPM
7X8
DOI 10.1093/aje/kwy201
DatabaseName Medline
MEDLINE
MEDLINE (Ovid)
MEDLINE
MEDLINE
PubMed
MEDLINE - Academic
DatabaseTitle MEDLINE
Medline Complete
MEDLINE with Full Text
PubMed
MEDLINE (Ovid)
MEDLINE - Academic
DatabaseTitleList MEDLINE
MEDLINE - Academic
Database_xml – sequence: 1
  dbid: NPM
  name: PubMed
  url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
– sequence: 2
  dbid: 7X8
  name: MEDLINE - Academic
  url: https://search.proquest.com/medline
  sourceTypes: Aggregation Database
DeliveryMethod no_fulltext_linktorsrc
Discipline Medicine
Public Health
EISSN 1476-6256
ExternalDocumentID 30189042
Genre Research Support, U.S. Gov't, Non-P.H.S
Research Support, U.S. Gov't, P.H.S
Journal Article
GroupedDBID ---
-DZ
-E4
-~X
..I
.2P
.I3
.XZ
.ZR
0R~
1TH
23M
2WC
4.4
482
48X
5GY
5RE
5VS
5WA
5WD
6J9
70D
85S
AABZA
AACZT
AAILS
AAJKP
AAMVS
AAOGV
AAPNW
AAPQZ
AAPXW
AARHZ
AAUAY
AAUQX
AAVAP
AAWTL
ABDFA
ABEJV
ABEUO
ABGNP
ABIXL
ABJNI
ABKDP
ABLJU
ABNHQ
ABNKS
ABOCM
ABPTD
ABQLI
ABVGC
ABXVV
ABZBJ
ACGFO
ACGFS
ACGOD
ACPRK
ACUFI
ACUTJ
ACUTO
ADBBV
ADCFL
ADEYI
ADEZT
ADGZP
ADHKW
ADHZD
ADIPN
ADMHG
ADNBA
ADOCK
ADQBN
ADRTK
ADVEK
ADYVW
ADZXQ
AEGPL
AEHKS
AEJOX
AEKSI
AEMDU
AENEX
AENZO
AEPUE
AETBJ
AEWNT
AFFZL
AFIYH
AFOFC
AFRAH
AFYAG
AGINJ
AGKEF
AGSYK
AHMBA
AHMMS
AHXPO
AIAGR
AIJHB
AJEEA
AJNCP
ALMA_UNASSIGNED_HOLDINGS
ALUQC
ALXQX
APIBT
APWMN
ATGXG
AXUDD
BAWUL
BAYMD
BCRHZ
BEYMZ
BHONS
BTRTY
BVRKM
C45
CDBKE
CGR
CS3
CUY
CVF
CZ4
DAKXR
DIK
DILTD
D~K
E3Z
EBS
ECM
EE~
EIF
EJD
EMOBN
F5P
F9B
FLUFQ
FOEOM
FOTVD
FQBLK
GAUVT
GJXCC
GX1
H13
H5~
HAR
HW0
HZ~
IH2
IOX
J21
JXSIZ
KAQDR
KBUDW
KOP
KQ8
KSI
KSN
L7B
M-Z
ML0
N9A
NGC
NOMLY
NOYVH
NPM
O9-
OAWHX
OCZFY
ODMLO
OHH
OJQWA
OJZSN
OK1
OPAEJ
OVD
OWPYF
P2P
P6G
PAFKI
PEELM
PQQKQ
Q1.
Q5Y
R44
RD5
ROL
ROX
ROZ
RUSNO
RW1
RXO
TCURE
TEORI
TJX
TR2
UHB
UPT
W8F
WOQ
X7H
YAYTL
YF5
YKOAZ
YOC
YROCO
YSK
YXANX
ZKX
~91
7X8
ADGHP
AHGBF
AJBYB
ID FETCH-LOGICAL-o269t-5f3d57afa5192ffb48b2a29f35a71f1de342ddbc56610e37ff24b736b0127b832
IEDL.DBID 7X8
ISICitedReferencesCount 537
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000465359900029&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1476-6256
IngestDate Sun Sep 28 05:05:28 EDT 2025
Thu Apr 03 07:06:54 EDT 2025
IsPeerReviewed true
IsScholarly true
Issue 1
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-o269t-5f3d57afa5192ffb48b2a29f35a71f1de342ddbc56610e37ff24b736b0127b832
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
PMID 30189042
PQID 2101277568
PQPubID 23479
ParticipantIDs proquest_miscellaneous_2101277568
pubmed_primary_30189042
PublicationCentury 2000
PublicationDate 2019-01-01
20190101
PublicationDateYYYYMMDD 2019-01-01
PublicationDate_xml – month: 01
  year: 2019
  text: 2019-01-01
  day: 01
PublicationDecade 2010
PublicationPlace United States
PublicationPlace_xml – name: United States
PublicationTitle American journal of epidemiology
PublicationTitleAlternate Am J Epidemiol
PublicationYear 2019
References 33155637 - Am J Epidemiol. 2021 Jan 4;190(1):189-190
References_xml – reference: 33155637 - Am J Epidemiol. 2021 Jan 4;190(1):189-190
SSID ssj0011950
Score 2.69608
Snippet The popular inverse probability weighting method in causal inference is often hampered by extreme propensity scores, resulting in biased estimates and...
SourceID proquest
pubmed
SourceType Aggregation Database
Index Database
StartPage 250
SubjectTerms Bias
Causality
Epidemiologic Methods
Humans
Models, Statistical
Propensity Score
Title Addressing Extreme Propensity Scores via the Overlap Weights
URI https://www.ncbi.nlm.nih.gov/pubmed/30189042
https://www.proquest.com/docview/2101277568
Volume 188
WOSCitedRecordID wos000465359900029&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1LS8NAEB7Uigjio77qixW8hia7STYBQYpYPNha8NVb2M3uShWSamrVf-9sktKTIHgJuQSSyezsN9_MfgNwFrpMRm5qnCDgmKAIvBNaug5GQtxfrZ53OYfs8Yb3-9FwGA9qwq2o2ypnMbEM1CpPLUfeplaIivMgjC7Gb46dGmWrq_UIjUVoMIQy1qv5cF5FsCNOy9NFPHQQ54czedKYtcWLbr9-flPX-x1alltMd-O_L7cJ6zW4JJ3KG7ZgQWdNWOnV5fMmrFUkHanOHm3DeUepsg82eyZXXxNLFZKBZecz26lB7qzEZUGmI0EQJpLbqeX-xuSpZFOLHXjoXt1fXjv1OAUnp2E8cQLDVMCFEQjaqDHSjyQVNDYsENwzntLMp0rJFAGe52rGjaG-5CyU9tskrvxdWMryTO8D0SnjgaCeYiLClEnFKSZGTGidWj1Bn7XgdGanBN3V1iBEpvOPIplbqgV7lbGTcaWrkWCsiWIMIgd_ePoQVvHfxRUZcgQNg4tVH8NyOp2MiveT0g_w2h_0fgA9qLxs
linkProvider ProQuest
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Addressing+Extreme+Propensity+Scores+via+the+Overlap+Weights&rft.jtitle=American+journal+of+epidemiology&rft.au=Li%2C+Fan&rft.au=Thomas%2C+Laine+E&rft.date=2019-01-01&rft.eissn=1476-6256&rft.volume=188&rft.issue=1&rft.spage=250&rft_id=info:doi/10.1093%2Faje%2Fkwy201&rft_id=info%3Apmid%2F30189042&rft_id=info%3Apmid%2F30189042&rft.externalDocID=30189042
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1476-6256&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1476-6256&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1476-6256&client=summon