A test of missing completely at random for longitudinal data with missing observations

Stat Med. 1997 Aug 30;16(16):1859-71. doi: 10.1002/(sici)1097-0258(19970830)16:16<1859::aid-sim593>3.0.co;2-3.

Abstract

Liang and Zeger proposed a generalized estimating equations approach to the analysis of longitudinal data. Their models assume that missing observations are missing completely at random in the sense of Rubin. However, when this assumption does not hold, their analysis may yield biased results. In this paper, we develop a simple and practical procedure for testing this assumption. The proposed procedure is related to that of Park and Davis.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Aged
  • Child
  • Coronary Disease / prevention & control
  • Data Interpretation, Statistical
  • Effect Modifier, Epidemiologic*
  • Female
  • Humans
  • Longitudinal Studies*
  • Male
  • Models, Statistical
  • Obesity / epidemiology
  • Regression Analysis*
  • Risk Factors
  • Sample Size
  • Urinary Incontinence / epidemiology