Project

General

Profile

Support #1347

weighting

Added by Laura Silva 4 months ago. Updated about 1 month ago.

Status:
Feedback
Priority:
Normal
Category:
Weights
Target version:
Start date:
05/14/2020
Due date:
% Done:

50%

Estimated time:

Description

1)I am using the waves a-f using info from the youth and the adults questionnaire in each wave. Which weight is the most adequate to use?

History

#1 Updated by Olena Kaminska 4 months ago

Laura,

The best weight for you would be a longitudinal adult response weight from the last wave of your analysis (f_indinus_lw or f_indpxus_lw depending on your analysis). This is a sub-optimal weight but will be close to the optimal one for your analysis.

Best wishes,
Olena

#2 Updated by Laura Silva 4 months ago

Hi Olena,

many thanks for your reply, I'll look into that.

Best,
Laura

#3 Updated by Gundi Knies 4 months ago

  • Private changed from Yes to No
  • % Done changed from 0 to 100
  • Target version set to X M
  • Assignee changed from Olena Kaminska to Laura Silva
  • Status changed from New to Resolved

You are welcome!

#4 Updated by Laura Silva about 2 months ago

Hi,

Apologies for the follow up.

I have a further question. Since I have individuals coming from only the youth and child samples, I do not find in these two datasets the f_indinus_lw or f_indpxus_lw weights you suggested. However since I link them to their parents, I have the weights from their parents. Which is the best way to weight? Use parental weight (maternal or paternal if only one of the two is available and the mean of the two if both are present)? Or there is a specific weight for these two samples?

Is is possible/correct not to weight? How much relevant is the use of weights in UKHLS?

Best regards,

Laura

#5 Updated by Alita Nandi about 2 months ago

  • % Done changed from 100 to 50
  • Assignee changed from Laura Silva to Olena Kaminska
  • Status changed from Resolved to In Progress

#6 Updated by Olena Kaminska about 2 months ago

Laura,

My previous suggestion of the weight is incorrect as I misunderstood the set up of your dataset. But to reply I need more information. Could you let me know who the unit of your analysis is: children (which age) or parents or something else (like child-parent combination)? Also, which instruments or datasets do you take your information from (if you have at least one variable from a dataset let me know), and which waves? I will then be able to suggest the best weight for you.

You can use unweighted analysis only if you are interested in the participants of our panel. If you want to generalise to a population you need to use weights.

Thank you,
Olena

#7 Updated by Alita Nandi about 1 month ago

  • Status changed from In Progress to Feedback

#8 Updated by Laura Silva about 1 month ago

Hi Olena,

apologies for the late reply.

The unit of my analysis is the child. I use both the child and the youth datasets, thus children from 0 to 15. I then use info from the adult sample (indresp) to retrieve parental controls. In addition I use the hhresp data to link other info such as tenure.

I hope I have clarified my query.

Best regards,

Laura

#9 Updated by Olena Kaminska about 1 month ago

Laura,

Thank you. I understand that you study 0-15 year olds and give them some attributes from their parents. We do not have specific weight for this, but a suboptimal best weight is f_psnenus_lw weight (I understand that you study children over time over waves a-f).

Hope this helps,
Olena

Also available in: Atom PDF