Using replicate weights for 2020 with rep weights for other years

TIME USE
1

Continuing the discussion from Using person and replication weights in a pooled 2003-2020 analysis:

Hi, this paper doesn’t address the question of replicate weights posed by the OP. What should we do about those, e.g. is it ok to use new variables equal to RWT06 for all years other than 2020, and the RWT20 weights 2020? Like this:

newRWT_01 = RWT06_01 if year <> 2020

newRWT_01 = RWT20_01 if year = 2020

<repeat for newRWT_02,…,newRWT_168>

Thanks much, SG

2

Neither the BLS nor the Census Bureau have provided any official guidance on using replicate weights in pooled analyses that include the 2020 ATUS data. Your suggested approach seems appropriate, but there is not yet (to our knowledge) any research published testing one method or another. While I can’t provide sound analytical guidance in this case given the lack of evidence, this is how I would create a pooled replicate weight, and I am not aware of another sensible option for weighting these data.