Inferential Precision in Single-Case Time-Series Data Streams: How Well Does the EM Procedure Perform When Missing Observations Occur in Autocorrelated Data?

Justin D. Smith*, Jeffrey J. Borckardt, Michael R. Nash

*Corresponding author for this work

Research output: Contribution to journalArticle

13 Citations (Scopus)

Abstract

The case-based time-series design is a viable methodology for treatment outcome research. However, the literature has not fully addressed the problem of missing observations with such autocorrelated data streams. Mainly, to what extent do missing observations compromise inference when observations are not independent? Do the available missing data replacement procedures preserve inferential integrity? Does the extent of autocorrelation matter? We use Monte Carlo simulation modeling of a single-subject intervention study to address these questions. We find power sensitivity to be within acceptable limits across four proportions of missing observations (10%, 20%, 30%, and 40%) when missing data are replaced using the Expectation-Maximization Algorithm, more commonly known as the EM Procedure (Dempster, Laird, & Rubin, 1977). This applies to data streams with lag-1 autocorrelation estimates under 0.80. As autocorrelation estimates approach 0.80, the replacement procedure yields an unacceptable power profile. The implications of these findings and directions for future research are discussed.

Original languageEnglish (US)
Pages (from-to)679-685
Number of pages7
JournalBehavior Therapy
Volume43
Issue number3
DOIs
StatePublished - Sep 1 2012

Fingerprint

Outcome Assessment (Health Care)
Direction compound

Keywords

  • Autocorrelation
  • Case-based
  • Missing observations
  • Power sensitivity
  • Time-series

ASJC Scopus subject areas

  • Clinical Psychology

Cite this

@article{fcf4c28f9552451bbd1616ea0b99d6cf,
title = "Inferential Precision in Single-Case Time-Series Data Streams: How Well Does the EM Procedure Perform When Missing Observations Occur in Autocorrelated Data?",
abstract = "The case-based time-series design is a viable methodology for treatment outcome research. However, the literature has not fully addressed the problem of missing observations with such autocorrelated data streams. Mainly, to what extent do missing observations compromise inference when observations are not independent? Do the available missing data replacement procedures preserve inferential integrity? Does the extent of autocorrelation matter? We use Monte Carlo simulation modeling of a single-subject intervention study to address these questions. We find power sensitivity to be within acceptable limits across four proportions of missing observations (10{\%}, 20{\%}, 30{\%}, and 40{\%}) when missing data are replaced using the Expectation-Maximization Algorithm, more commonly known as the EM Procedure (Dempster, Laird, & Rubin, 1977). This applies to data streams with lag-1 autocorrelation estimates under 0.80. As autocorrelation estimates approach 0.80, the replacement procedure yields an unacceptable power profile. The implications of these findings and directions for future research are discussed.",
keywords = "Autocorrelation, Case-based, Missing observations, Power sensitivity, Time-series",
author = "Smith, {Justin D.} and Borckardt, {Jeffrey J.} and Nash, {Michael R.}",
year = "2012",
month = "9",
day = "1",
doi = "10.1016/j.beth.2011.10.001",
language = "English (US)",
volume = "43",
pages = "679--685",
journal = "Behavior Therapy",
issn = "0005-7894",
publisher = "Elsevier Inc.",
number = "3",

}

Inferential Precision in Single-Case Time-Series Data Streams : How Well Does the EM Procedure Perform When Missing Observations Occur in Autocorrelated Data? / Smith, Justin D.; Borckardt, Jeffrey J.; Nash, Michael R.

In: Behavior Therapy, Vol. 43, No. 3, 01.09.2012, p. 679-685.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Inferential Precision in Single-Case Time-Series Data Streams

T2 - How Well Does the EM Procedure Perform When Missing Observations Occur in Autocorrelated Data?

AU - Smith, Justin D.

AU - Borckardt, Jeffrey J.

AU - Nash, Michael R.

PY - 2012/9/1

Y1 - 2012/9/1

N2 - The case-based time-series design is a viable methodology for treatment outcome research. However, the literature has not fully addressed the problem of missing observations with such autocorrelated data streams. Mainly, to what extent do missing observations compromise inference when observations are not independent? Do the available missing data replacement procedures preserve inferential integrity? Does the extent of autocorrelation matter? We use Monte Carlo simulation modeling of a single-subject intervention study to address these questions. We find power sensitivity to be within acceptable limits across four proportions of missing observations (10%, 20%, 30%, and 40%) when missing data are replaced using the Expectation-Maximization Algorithm, more commonly known as the EM Procedure (Dempster, Laird, & Rubin, 1977). This applies to data streams with lag-1 autocorrelation estimates under 0.80. As autocorrelation estimates approach 0.80, the replacement procedure yields an unacceptable power profile. The implications of these findings and directions for future research are discussed.

AB - The case-based time-series design is a viable methodology for treatment outcome research. However, the literature has not fully addressed the problem of missing observations with such autocorrelated data streams. Mainly, to what extent do missing observations compromise inference when observations are not independent? Do the available missing data replacement procedures preserve inferential integrity? Does the extent of autocorrelation matter? We use Monte Carlo simulation modeling of a single-subject intervention study to address these questions. We find power sensitivity to be within acceptable limits across four proportions of missing observations (10%, 20%, 30%, and 40%) when missing data are replaced using the Expectation-Maximization Algorithm, more commonly known as the EM Procedure (Dempster, Laird, & Rubin, 1977). This applies to data streams with lag-1 autocorrelation estimates under 0.80. As autocorrelation estimates approach 0.80, the replacement procedure yields an unacceptable power profile. The implications of these findings and directions for future research are discussed.

KW - Autocorrelation

KW - Case-based

KW - Missing observations

KW - Power sensitivity

KW - Time-series

UR - http://www.scopus.com/inward/record.url?scp=84862182493&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84862182493&partnerID=8YFLogxK

U2 - 10.1016/j.beth.2011.10.001

DO - 10.1016/j.beth.2011.10.001

M3 - Article

C2 - 22697454

AN - SCOPUS:84862182493

VL - 43

SP - 679

EP - 685

JO - Behavior Therapy

JF - Behavior Therapy

SN - 0005-7894

IS - 3

ER -