Weights and finding variables
My Masters project is looking at the possible association between commuting times and CRP levels. I have a few points that are not clear for me from the user guides. Firstly, I wish to do a cross-sectional analysis of Wave 2 data only e.g. main interview, nurse visit and blood sample data. However the cross sectional weight provided in the guide is for a combination of Wave 2 and Wave 3. Could you provide me with the weight for Wave 2 only?
My 2nd main area of confusion was that in the main interview questions for wave 2 there doesn't seem to be a clear way of distinguishing if someone takes anti-inflammatories, statins (not if they are new to taking statins, but if they are currently taking statins, prescribed or not) or contraceptive/hrt drugs, however this is important for crp. Is there any way to access this information without having a special licence?
#1 Updated by Redmine Admin almost 5 years ago
- Category set to Weights
- Target version set to X Ns
- % Done changed from 0 to 50
For your 2nd question, please see this and related variables: https://www.understandingsociety.ac.uk/documentation/health-assessment/dataset-documentation/wave/xwave/datafile/xindresp_ns/variable/bnf7_conhrt
Your 1st question has been referred to the methodology group. Please be aware that it may take us a little longer than usual to respond to these queries due to vacations and conference attendance.
On behalf of the team,
#2 Updated by Tracy Turc-Milloy almost 5 years ago
For your 2nd question, please see this and related variables: https://www.understandingsociety.ac.uk/documentation/health-assessment/dataset-documentation/wave/xwave/datafile/xindresp_ns/variable/bnf7_conhrt - I was aware that these variables exist but my problem lies with the fact that I am trying to use Wave 2 data only - as these variables are from data from wave 2 & 3 combined, how can I be sure the information from that variable was true in Wave 2 and not just Wave 3?
#5 Updated by Tracy Turc-Milloy almost 5 years ago
Tracy Turc-Milloy wrote:
Please could you tell me if the committee has determined yet a weighting that could apply to W2 members who had a blood sample taken?
I apologize, I have not stated the urgency of my request. I am currently in the middle of a week's leave from work to work on my Masters project - the first draft of which must be submitted in 2 weeks. I am unable to do the linear regression with the current weighting svyset b_psu [pweight=indbdub_xw], strata(b_strata)as I get missing values for F and P values. Please help me with a correct weighting,
#6 Updated by Peter Lynn almost 5 years ago
I suggest that you use the weight b_indinus_xw for your analysis. It is not quite optimal as it does not correct for the final stage of non-response (to the nurse visit, conditional on completing the main individual interview), but it should be good enough.
(And apologies for the slow reply; I was away last week.)