Are missing data lost data?

• Pages: 131-140
• Date: 01 January 2021
• Published date: 01 January 2021
• Author: Blerina Metanj (Subashi)
• Subject Matter: Derecho Público y Administrativo

European Journal of Economics, Law and Social Sciences

IIPCCL Publishing, Graz-Austria

Vol. 5 No. 1

January, 2021

ISSN 2519-1284

Acces online at www.iipccl.org

131

Are missing data lost data?

Blerina Metanj (Subashi)

Senior Research Analyst

IDRA Research & Consulting

Abstract

Data analysis is sometimes compromised by missing data. In recent years the development of

statistical methods to address missing data has been an active area of research. In this context,

multiple imputationsis a general-purpose method for analyzing datasets with missing data

that is broadly applicable to a variety of missing data. In this paper rstly, issues on missing

data will be presented with a brief theoretical introduction to the multiple imputations as an

analytic strategy under this eld. Besides,a discussion on multiple imputation techniques with

a data example for illustration and clarity will be presented. In the end, some conclusions on a

particular analysis will be discussed.

Keywords: multiple imputations, missing data, data analysis.

Introduction

Missing data can be a cause of problems and biases in many datasets. These can be

a challenge to many researchers of dierent elds that have to deal with data. In the

last years, a lot of methodologies handle these kinds of problems. Neglecting missing

data can harm the results and bring to wrong conclusions (Lile & Rubin, 1987;

Graham, Hofer, Donaldson, MacKinnon, & Schafer, 1997; Schafer & Graham, 2002).

Multipleimputationsin this context, are a methodology for handling missing data

which can be used by researchers on many analytic levels. Many research studies

have used multiple imputations (e.g., Graham et al., 1997; Wayman, 2002a) and good

general reviews on multiple imputations have been published (Graham, Cumsille,

&Elek-Fisk, 2003; Graham & Hofer, 2000; Schafer & Olsen, 1998; Sinharay, Stern,

& Russell, 2001). However, many researchers still do not use it due to the lack of

information on the benets that it can bring to their work.

This paper aims to be additional material to the existing literature giving a concrete

example of using multiple imputations and showing its benets. Many researchers

can understand it and might use it in their work.

1. Background

Item missing data has long been recognized as a problem for data analysts. Early

solutions to the problem of missing data were directed to specic distributions for the

variables of interest and paerns of missing data. For example, Buck’s (1960) method

introduced imputations of conditional mean values for each paern of missing

observations in a multivariate normal vector of variables.

Broad, formal recognition of imputation as a statistical technique for dealing with