Project

General

Profile

Support #1308

Cross-sectional weights 0 values

Added by Pascale Bourquin almost 3 years ago. Updated about 1 year ago.

Status:
Resolved
Priority:
Urgent
Assignee:
-
Category:
Weights
Start date:
01/29/2020
% Done:

100%


Description

Hello,
We are doing crossectional analysis combining all waves from USoc and the BHPS and include all household members (and all regions of the UK whenever become available). We are basically taking information from different waves, for example, wealth from the waves with wealth modules and labour market outcomes/earnings for when people are in their mid 20s. Currently we are using the crossectional weights from the last wave we observe people so we have constant weights even though we use information from various waves for each unique observation (so depending on which wave this are (a-f)xewght (waves 1-6 BHPS), (g-h)xewghte (waves 7-8 BHPS), (i-j)xewtsw1 (waves 9-10 BHPS), (k-..)xewtuk1 (waves 11 onwarfdd BHPS), a_psnenus_xw (wave 1 Usoc), (b-h)_psnenub_xw (wave 2 onwards of USoc)). We are then adjusting them to account for selection of our sample based on observable characteristics.
I have two specific questions:
1) Are the weights we are using as our base weight (that is, we are using these weights to create our own adjusted weights) the right ones for this kind of analysis? and
2) Why is it the case that some of these weights are 0? I understand why longitudinal weights would be 0, however cannot quite understand why any of the cross-sectional ones would be. It is especially confusing that specific individuals have non-zero cross-sectional weights for some waves and not for others.
Any help answering the above 2 questions would be much appreciated.
Many thanks and best,
Pascale

Also available in: Atom PDF