Project

General

Profile

Support #1221

Using UKHLS at the quarterly level

Added by Tom Waters about 3 years ago. Updated about 3 years ago.

Status:
Resolved
Priority:
Normal
Assignee:
Category:
Weights
Start date:
07/31/2019
% Done:

100%


Description

Hi,

I am trying to look at various outcomes by quarter in a dataset which pools multiple UKHLS waves together with another dataset (the FRS).

However I understand that there is non-random sampling within wave. For example, in wave C Northern Ireland is only sampled in the first year (2011). I was sent a document from someone at Essex which discusses pooling data from different waves in UKHLS. It says, "The weights provided currently are not designed for pooling as they are scaled to a mean value of 1.0 within each wave, and therefore produce different weighted sample sizes in each wave. As a result, cases from later waves will be under-represented."

I have three questions:
1. Why is it that later waves would be under-represented? I should've thought that if they had the same average weight, then cases would be equally represented when you weight?
2. What steps should I take to ensure that quarterly level outcomes are representative?
3. Are there any additional steps I should take to take account of the fact that we are pooling with the FRS? Currently we are making sure that, within a financial year, the FRS observations and UKHLS observations have the same average weight. Is that sufficient?

Many thanks for your help.

Tom

Also available in: Atom PDF