Statistical Matching for Combining Time-Use Surveys with Consumer Expenditure Surveys: An Evaluation on Real Data

Abstract : Performing a statistical match to combine two surveys made over the same population by traditional methods is shown to give biased estimates and variance of the imputed values. A method proposed by Rubin (1986) allows imputing an unobserved variable using observations in another dataset by taking into account the partial correlation between the variables that are jointly unobserved for any unit. We use a dataset where households report their expenditures and time-uses to show that fusioning expenditure and time-use surveys by Rubin's procedure allows to recover the true distribution of the missing variables and to yield minimally biased estimates.
Complete list of metadatas

Cited literature [8 references]  Display  Hide  Download

https://halshs.archives-ouvertes.fr/halshs-01529699
Contributor : Lucie Label <>
Submitted on : Wednesday, May 31, 2017 - 11:45:51 AM
Last modification on : Tuesday, January 30, 2018 - 5:50:04 PM
Long-term archiving on : Wednesday, September 6, 2017 - 3:02:25 PM

File

17024.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : halshs-01529699, version 1

Collections

Citation

Anil Alpman, François Gardes, Noel Thiombiano. Statistical Matching for Combining Time-Use Surveys with Consumer Expenditure Surveys: An Evaluation on Real Data. 2017. ⟨halshs-01529699⟩

Share

Metrics

Record views

167

Files downloads

414