1

I am using Amelia to generate five new hypothetical data sets for each variable with missing data. Now, I want to create new variables after imputation.

The reproducible dataset is in the Zelig package.

require(Zelig)
require(Amelia)
data(freetrade)
       year    country   tariff   polity  pop        gdp.pc    ...
    1  1981    SriLanka  NA       6       14988000   461.0236 
    2  1982    SriLanka  NA       5       15189000   473.7634 
    3  1983    SriLanka  41.3     5       15417000   489.2266

creating 5 imputed datasets with Amelia II 

a.out <- amelia(freetrade, m=5, ts="year", cs="year")

transforming imputed variables

I learnt how to create transformations of the imputed variables for use in further analysis, see 4.8. in the manual. For instance:

a.out<-transform(a.out, pol_gdp = polity * gdp-pc)  
a.out<-transform(a.out, pol.gdp = polity / gdp)  
head(a.out$imputations[[1]])

  year  country  tariff   polity pop      gdp.pc  ... pol_gdp  pol.gdp 
1 1981  SriLanka 39.55176 6      14988000 461.0236    2766.142 0.013014518  
2 1982  SriLanka 55.70500 5      15189000 473.7634    2368.817 0.010553791 
3 1983  SriLanka 41.30000 5      15417000 489.2266    2446.133 0.010220214
     ...

PROBLEM

I do not know how to create a new variable depending on the value in another variable. That is, usually I would use the if-else argument, which does however not work in Amelia.

a.out$imputations$new.var<-ifelse(a.out$imputations$polity==5,1,0)

Any ideas would be highly appreciated!

Many many thanks.

TiF
  • 615
  • 2
  • 12
  • 24

1 Answers1

0

Many thanks to all those who started to think about how to answer this question.

I think I have found the solution myself:

a.out <- transform(a.out, new.var=ifelse(polity==5,1,0))
TiF
  • 615
  • 2
  • 12
  • 24