Nonprobability sampling and causal analysis

Ulrich Kohler, Frauke Kreuter, Elizabeth Stuart

Research output: Contribution to journalReview article

Abstract

The long-standing approach of using probability samples in social science research has come under pressure through eroding survey response rates, advanced methodology, and easier access to large amounts of data. These factors, along with an increased awareness of the pitfalls of the nonequivalent comparison group design for the estimation of causal effects, have moved the attention of applied researchers away from issues of sampling and toward issues of identification. This article discusses the usability of samples with unknown selection probabilities for various research questions. In doing so, we review assumptions necessary for descriptive and causal inference and discuss research strategies developed to overcome sampling limitations.

Original languageEnglish (US)
Pages (from-to)149-172
Number of pages24
JournalAnnual Review of Statistics and Its Application
Volume6
DOIs
StatePublished - Mar 7 2019

Fingerprint

Causal Inference
Causal Effect
Social Sciences
Usability
Unknown
Necessary
Methodology
Sampling
Strategy
Awareness
Review
Design
Factors
Causal inference
Response rate
Research strategy
Social sciences
Causal effect

Keywords

  • big data
  • causal inference
  • generalizability
  • heterogeneous treatment effects
  • measurement error
  • nonprobability sampling
  • self-selection
  • validity

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Nonprobability sampling and causal analysis. / Kohler, Ulrich; Kreuter, Frauke; Stuart, Elizabeth.

In: Annual Review of Statistics and Its Application, Vol. 6, 07.03.2019, p. 149-172.

Research output: Contribution to journalReview article

@article{3bd3f6a493f1494fa9d1404eb9bf9fb1,
title = "Nonprobability sampling and causal analysis",
abstract = "The long-standing approach of using probability samples in social science research has come under pressure through eroding survey response rates, advanced methodology, and easier access to large amounts of data. These factors, along with an increased awareness of the pitfalls of the nonequivalent comparison group design for the estimation of causal effects, have moved the attention of applied researchers away from issues of sampling and toward issues of identification. This article discusses the usability of samples with unknown selection probabilities for various research questions. In doing so, we review assumptions necessary for descriptive and causal inference and discuss research strategies developed to overcome sampling limitations.",
keywords = "big data, causal inference, generalizability, heterogeneous treatment effects, measurement error, nonprobability sampling, self-selection, validity",
author = "Ulrich Kohler and Frauke Kreuter and Elizabeth Stuart",
year = "2019",
month = "3",
day = "7",
doi = "10.1146/annurev-statistics-030718-104951",
language = "English (US)",
volume = "6",
pages = "149--172",
journal = "Annual Review of Statistics and Its Application",
issn = "2326-831X",
publisher = "Annual Reviews Inc.",

}

TY - JOUR

T1 - Nonprobability sampling and causal analysis

AU - Kohler, Ulrich

AU - Kreuter, Frauke

AU - Stuart, Elizabeth

PY - 2019/3/7

Y1 - 2019/3/7

N2 - The long-standing approach of using probability samples in social science research has come under pressure through eroding survey response rates, advanced methodology, and easier access to large amounts of data. These factors, along with an increased awareness of the pitfalls of the nonequivalent comparison group design for the estimation of causal effects, have moved the attention of applied researchers away from issues of sampling and toward issues of identification. This article discusses the usability of samples with unknown selection probabilities for various research questions. In doing so, we review assumptions necessary for descriptive and causal inference and discuss research strategies developed to overcome sampling limitations.

AB - The long-standing approach of using probability samples in social science research has come under pressure through eroding survey response rates, advanced methodology, and easier access to large amounts of data. These factors, along with an increased awareness of the pitfalls of the nonequivalent comparison group design for the estimation of causal effects, have moved the attention of applied researchers away from issues of sampling and toward issues of identification. This article discusses the usability of samples with unknown selection probabilities for various research questions. In doing so, we review assumptions necessary for descriptive and causal inference and discuss research strategies developed to overcome sampling limitations.

KW - big data

KW - causal inference

KW - generalizability

KW - heterogeneous treatment effects

KW - measurement error

KW - nonprobability sampling

KW - self-selection

KW - validity

UR - http://www.scopus.com/inward/record.url?scp=85062883523&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85062883523&partnerID=8YFLogxK

U2 - 10.1146/annurev-statistics-030718-104951

DO - 10.1146/annurev-statistics-030718-104951

M3 - Review article

AN - SCOPUS:85062883523

VL - 6

SP - 149

EP - 172

JO - Annual Review of Statistics and Its Application

JF - Annual Review of Statistics and Its Application

SN - 2326-831X

ER -