Missing values is one of the most pervasive problems in data analysis. Missing data are widespread in social science surveys because respondents are unable or unwilling to answer some questionnaire items. Most researchers who use survey data must grapple with the problem of how best to handle missing data. The seriousness of the problem depends on the pattern of missing data, the distribution of missingness, how much is missing, and why it is missing. We will explain and clear up misconceptions about fundamental concepts related to the distribution of missing values, such as missing at randomn (MAR), missing completely at random (MCAR) and missing not at random (MNAR). We will illustrate that the pattern of missing data is more important than the amount missing. The decision about how to handle missing data is very important as it affects the reliability and accuracy of inferences about the population of interest. This decision is more often than not among several bad alternatives. We will review and evaluate traditional missing data procedures of listwise deletion, pairwise deletion and mean substitution. We will show evidence against them and, with few exceptions, discourage their use. Recent advances in statistics have led to new missing data handling techniques with sound statistical bases. We summarise and evaluate the new Maximum Likelihood (ML) and Bayesian multiple imputation (MI) techniques. We hope to encourage researchers to apply the more recent MI and ML procedures to deal with the missing data problem in their work. 
Please RSVP by 8 November to: HSRC Cape Town: 12th Floor, Plein Park Building (Opposite Revenue Office)
Plein Street, Cape Town
Contact Ngxubaza Vuyo, on +27 (0)21 466 8099 HSRC Durban: 1st floor boardroom
750 Francois Road, Ntuthuko Junction, PODS 5 and 6
Cato Manor, Durban
Contact Johannes Khoele on +27 (0)31 242 5400 HSRC Pretoria: Video Conference Room
1st floor, HSRC Library
HSRC Building, 134 Pretorius Street Pretoria
Contact Arlene Grossberg on +27 (0)12 302 2801
or Baby Twala on +27 (0)12 302 2368 |
