Incomplete Data Matrices and Tests on Randomly Missing Data

  • U. Bankhofer
Conference paper
Part of the Studies in Classification, Data Analysis, and Knowledge Organization book series (STUDIES CLASS)


In a practical analysis of data, the problem of incomplete data matrices is usually solved by estimating and substituting the missing values. Nevertheless, the imputation of missing values is only appropriate if the data are randomly missing. In general, the appropriate use of any missing-data method requires fundamental knowledge of the reasons and the underlying missing-data mechanism. With an analysis of the structure of the incomplete data matrix, the effects of the missing-data mechanism to the data under consideration can be investigated. Regarding the possible relations of dependence concerning the missing data, there are a few methods to test the existence of a non-systematic missing-data mechanism. The results of these tests can be sufficient conditions to reject the acceptance of randomly missing data or necessary conditions to accept a non-systematic missing-data mechanism.


Data Matrix Expected Frequency Inductive Analysis Indicator Matrix Test Method Analysis 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. KIM, J.O., CURRY, J. (1977): The Treatment of Missing Data in Multivariate Analysis. Sociological Methods and Research, 6, 215–239. CrossRefGoogle Scholar
  2. LITTLE, R.J.A. (1988): A Test of Missing Completely at Random for Multivariate Data with Missing Values. Journal of the American Statistical Association, 83, 1198–1202. CrossRefGoogle Scholar
  3. LÖSEL, F., WÜSTENDÖRFER, W. (1974): Zum Problem unvollständiger Datenmatrizen in der empirischen Sozialforschung. Kölner Zeitschrift für Soziologie und Sozialpsychologie, 26, 342–357. Google Scholar
  4. RUBIN, D.B. (1976): Inference and Missing Data. Biometrika, 63, 581–592. CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin · Heidelberg 1996

Authors and Affiliations

  • U. Bankhofer
    • 1
  1. 1.Institut für Statistik und Mathematische WirtschaftstheorieUniversität AusgburgAusburgGermany

Personalised recommendations