« Back to all events

Missing Data in Quantitative Analysis: the theory and practice of what to do about it

Dates:
  • Thu 14 Nov 2019 11.30 - 15.00
  Add to Calendar 2019-11-14 11:30 2019-11-14 15:00 Europe/Paris Missing Data in Quantitative Analysis: the theory and practice of what to do about it

What do we do when we have holes in our datasets, when we have, so called, item missing data? One thing's for sure, we do something. And software defaults, most often, do not offer the optimal solution to the problems missing data pose. In fact, missing data requires theoretical thinking about our data, why it is missing. Analytical procedures are available to alleviate the associated bias. This workshop covers various theories of missing data that should be considered when devising solutions to the missing data problem. We briefly review full information maximum likelihood estimation as one of the (better) solutions and practice with the other: multiple imputation. The class will also demonstrate common solutions seen in software and the literature that are probably better left alone never to be used. In discussion we may also briefly cover unit nonresponse strategies and even causal inference as a missing data problem.

Workshop uses R. Basic functional knowledge in R is assumed. We also assume a basic understanding of inferential statistics, knowing what standard errors are and where they come from, and familiarity with basic tests of differences (e.g. t-test) and relationships (e.g. multiple regression).

Seminar Room 2, Badia Fiesolana DD/MM/YYYY
  Seminar Room 2, Badia Fiesolana

What do we do when we have holes in our datasets, when we have, so called, item missing data? One thing's for sure, we do something. And software defaults, most often, do not offer the optimal solution to the problems missing data pose. In fact, missing data requires theoretical thinking about our data, why it is missing. Analytical procedures are available to alleviate the associated bias. This workshop covers various theories of missing data that should be considered when devising solutions to the missing data problem. We briefly review full information maximum likelihood estimation as one of the (better) solutions and practice with the other: multiple imputation. The class will also demonstrate common solutions seen in software and the literature that are probably better left alone never to be used. In discussion we may also briefly cover unit nonresponse strategies and even causal inference as a missing data problem.

Workshop uses R. Basic functional knowledge in R is assumed. We also assume a basic understanding of inferential statistics, knowing what standard errors are and where they come from, and familiarity with basic tests of differences (e.g. t-test) and relationships (e.g. multiple regression).


Location:
Seminar Room 2, Badia Fiesolana

Affiliation:
Department of Political and Social Sciences

Type:
Workshop

Contact:
Monika Rzemieniecka (EUI - Department of Political and Social Sciences) - Send a mail

Organiser:
Prof. Elias Dinas (EUI - Department of Political and Social Sciences)

Speaker:
Prof. Levente Littvay (Central European University)
 
 
 

Page last updated on 18 August 2017