USING LINEAR TREND MODELS TO ANALYZE POLICY IMPACT

Marvin B. Mandell, Stuart Bretschneider

Research output: Contribution to journalArticle

Abstract

Policy researchers have become increasingly familiar with a number of improved techniques for analyzing data obtained from interrupted time‐series designs for evaluating public programs and policies. In this paper we contribute to this trend by presenting two groups of data analysis techniques which are not currently widely used by policy researchers, but are likely to be valuable adjuncts to traditional regression techniques for analyzing data obtained from interrupted time‐series designs. First, aids for model specification are presented that enable the analyst to define an appropriate linear trend model—often one which will reduce the degree of multicollinearity and, therefore, produce more precise estimates of the impacts of a public program or policy. Next we consider approaches for point estimation and joint (simultaneous) interval estimation of a policy intervention's total effect at various points in time.

Original languageEnglish (US)
Pages (from-to)476-495
Number of pages20
JournalReview of Policy Research
Volume6
Issue number3
DOIs
StatePublished - 1987
Externally publishedYes

Fingerprint

trend
data analysis
regression
Group
policy
time
programme
public
effect

ASJC Scopus subject areas

  • Geography, Planning and Development
  • Public Administration
  • Management, Monitoring, Policy and Law

Cite this

USING LINEAR TREND MODELS TO ANALYZE POLICY IMPACT. / Mandell, Marvin B.; Bretschneider, Stuart.

In: Review of Policy Research, Vol. 6, No. 3, 1987, p. 476-495.

Research output: Contribution to journalArticle

@article{a370598b255e400f9ddce71a2ea56412,
title = "USING LINEAR TREND MODELS TO ANALYZE POLICY IMPACT",
abstract = "Policy researchers have become increasingly familiar with a number of improved techniques for analyzing data obtained from interrupted time‐series designs for evaluating public programs and policies. In this paper we contribute to this trend by presenting two groups of data analysis techniques which are not currently widely used by policy researchers, but are likely to be valuable adjuncts to traditional regression techniques for analyzing data obtained from interrupted time‐series designs. First, aids for model specification are presented that enable the analyst to define an appropriate linear trend model—often one which will reduce the degree of multicollinearity and, therefore, produce more precise estimates of the impacts of a public program or policy. Next we consider approaches for point estimation and joint (simultaneous) interval estimation of a policy intervention's total effect at various points in time.",
author = "Mandell, {Marvin B.} and Stuart Bretschneider",
year = "1987",
doi = "10.1111/j.1541-1338.1987.tb00762.x",
language = "English (US)",
volume = "6",
pages = "476--495",
journal = "Review of Policy Research",
issn = "0278-4416",
publisher = "Wiley-Blackwell",
number = "3",

}

TY - JOUR

T1 - USING LINEAR TREND MODELS TO ANALYZE POLICY IMPACT

AU - Mandell, Marvin B.

AU - Bretschneider, Stuart

PY - 1987

Y1 - 1987

N2 - Policy researchers have become increasingly familiar with a number of improved techniques for analyzing data obtained from interrupted time‐series designs for evaluating public programs and policies. In this paper we contribute to this trend by presenting two groups of data analysis techniques which are not currently widely used by policy researchers, but are likely to be valuable adjuncts to traditional regression techniques for analyzing data obtained from interrupted time‐series designs. First, aids for model specification are presented that enable the analyst to define an appropriate linear trend model—often one which will reduce the degree of multicollinearity and, therefore, produce more precise estimates of the impacts of a public program or policy. Next we consider approaches for point estimation and joint (simultaneous) interval estimation of a policy intervention's total effect at various points in time.

AB - Policy researchers have become increasingly familiar with a number of improved techniques for analyzing data obtained from interrupted time‐series designs for evaluating public programs and policies. In this paper we contribute to this trend by presenting two groups of data analysis techniques which are not currently widely used by policy researchers, but are likely to be valuable adjuncts to traditional regression techniques for analyzing data obtained from interrupted time‐series designs. First, aids for model specification are presented that enable the analyst to define an appropriate linear trend model—often one which will reduce the degree of multicollinearity and, therefore, produce more precise estimates of the impacts of a public program or policy. Next we consider approaches for point estimation and joint (simultaneous) interval estimation of a policy intervention's total effect at various points in time.

UR - http://www.scopus.com/inward/record.url?scp=84984075193&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84984075193&partnerID=8YFLogxK

U2 - 10.1111/j.1541-1338.1987.tb00762.x

DO - 10.1111/j.1541-1338.1987.tb00762.x

M3 - Article

AN - SCOPUS:84984075193

VL - 6

SP - 476

EP - 495

JO - Review of Policy Research

JF - Review of Policy Research

SN - 0278-4416

IS - 3

ER -