• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site
  • HSE University
  • Publications of HSE
  • Articles
  • Модель-ориентированный подход к отсутствующим значениям: множественная импутация в многоуровневой регрессии посредством R (на примере анализа опросных данных по гордости страной)

Article

Модель-ориентированный подход к отсутствующим значениям: множественная импутация в многоуровневой регрессии посредством R (на примере анализа опросных данных по гордости страной)

The article substantiates and describes the multiple imputation technique and its procedure of dealing with missing values in dataset. It presents the model-oriented approach as opposed to design-oriented approach to analyzing survey data. The theory section lists the benefits of multiple imputation and states why it should be preferred to other ways of dealing with missing data, such as listwise or pairwise deletion or substitution by the mean. It states the limitations of the multiple imputation technique and the related conditions that the data must satisfy so that multiple imputation could be performed. The empirical section of the article draws on a specific case of the use of multiple imputation in a cross-cultural research on national pride. It shows the procedure in the form of a ready for use program code for diagnosing data ad imputing missing values by means of R programming packages VIM and mice.