Weights for Non Respondents

Are weights available for non respondents in ATUS? I need to calculate average characteristics of non respondents (children) and merge those to respondents (mothers). So I would like to calculate these averages by using weights.

The variable BWT provides a sampling weight for both respondents and non-respondents. Please note that ATUS employs a complex sample design and offers replicate weights for variance estimation purposes; the 160 replicate base weights can be added to your data file by including the RBWT variable.

A previous version of this response incorrectly stated that BWT was available for all persons; BWT values are 0 for household members.

1 Like

@KariWilliams I am using ATUS for years 2003-2015 and would like to get weighted characteristics of respondents and non respondents. So I weight by BWT. However around 90% of my sample have a BWT of zero. Why is BWT zero for so many individuals?

My previous answer overstated who is included in BWT. BWT provides a weight for respondents (eligible persons who were selected and completed the ATUS) and non-respondents (eligible persons who were selected but did NOT complete the ATUS); it does not cover household members (ineligible persons or eligible persons who were not selected to complete the ATUS). Based on your original question, you are interested in looking at weighted characteristics of household members–not non-respondents. I am sorry for missing this key detail and incorrectly directing you towards BWT; I will also modify the previous post to more clearly differentiate between household members and non-respondents.

To my knowledge, there is no weight for household members. While I don’t have a clear picture of your broader analytical plan and providing feedback on analytical choices is beyond the role of the IPUMS User Support team, it is not clear to me why weighting for household members would be necessary as you describe your original approach. I think of the children of respondents (and their characteristics) as attributes of those respondents; therefore it seems appropriate to me to summarize the traits of other household members on the respondent record and weight using WT06.

1 Like

That makes sense! Sorry if my question was not clear! I can see your point about using WT06 for children of the respondents.

1 Like

You asked a great and important question–I should have read more closely and thought about the differentiation of non-respondents versus household members. Thanks for following up so I could provide correct information!

1 Like

Thanks for all the information. It was really helpful. Especially knowing the difference between eligible respondents vs eligible non-respondent vs non-eligible non-respondents.