Hello,
If I wanted to link children to their parents in the MTUS sample, how can we know for sure that a child is the child of both the reference person (using relrefp) and the partner or spouse of the reference person? I assume that’s true in most cases, but one obvious exception would be when the reference person is in a second marriage and the child is from a previous partnership.
Thank you so much!
Unfortunately I am not aware of any variables in IPUMS MTUS samples that would allow you to determine the parentage of household children with more certainty. In some samples, there is more information available in the original data that would allow you to explicitly identify parent-child relationships and other relationships between household members. If this information is critical for your work, I would suggest looking through the original survey documentation and reaching out to the Centre for Time Use Research to request access to additional variables. I am not sure whether they would be able to provide this, but since they are the source of the data we provide via IPUMS MTUS, that is likely your best bet.
That makes sense, thank you!
Hello,
I hope you’re doing well. I had a couple of follow-up questions with the MTUS harmonized dataset, specifically surrounding income.
While reviewing both the codebook and the underlying data, I noticed some inconsistencies:
France 1998–99
The codebook indicates that this sample should have grouped income categories. However, in the actual dataset, the income measure appears to be continuous. I just want to make sure I’m not missing anything here.
France 2009
The dataset reports income categories coded 1–5, but the IPUMS codebook does not provide any category definitions or income bounds for these groups. Do you know is such a codebook exists for this sample?
Thank you for all your help so far!
I see that we do not provide value labels for the codes for INCORIG in the France 2009/2010 sample.
I also see that the codes provided in the INCORIG metadata for the France 1998/1999 sample do not align with the codes in the actual data. Here are the codes and frequencies in the data:
And here are the codes listed in the INCORIG codes section:
These codes and labels come from the CTUR’s harmonization documentation, but evidently they do not match the data.
I have inquired with our IPUMS Time Use team about these two issues. I will follow up with you when I have more information to share. Thank you for your patience.
Thanks for reaching out!
I’m writing to let you know we are still waiting for more information on these variable codes, and I will post here when we have something to share with you.
Hello Isabel,
Thanks so much! I have just been getting back into this project, so I appreciate your diligence in this matter!
I had two related questions that I was hoping you could help clarify.
First, regarding the variable WORKHRS: I’m finding that WORKHRS is always missing in the Italy 2002 sample. I was a bit unsure why this would be the case, since the IPUMS codebook notes under the Italy section that “In 2002, for those who reported that they were working but it is impossible to assign the exact number of hours, we assign the number of hours worked as ‘missing’ (-8). For those individuals who were not working when they answered the diary questionnaire, we assign the value ‘-7’ (not applicable).” Based on this description, I would have expected WORKHRS not to be missing across the entire sample.
Second, I had a question about the construction of the NCHILD variable. Is there a reason why IPUMS provides NCHILD but not separate variables for the number of sons and daughters? I wasn’t sure whether NCHILD is derived from an underlying household roster file, or if it comes from a direct survey question asking respondents how many children they have. Any clarification on how this variable is constructed would be very helpful!
Hi Hadin. This is good timing as we have just received some information from the Centre for Time Use Research about the INCORIG variable. I will also answer your other questions here.
1. INCORIG in France 1998 and 2009 samples
France 1998 sample
The codes and their corresponding labels are:
-8 “Doesn’t know”
1750 “Less than 3,500”
5250 “3,500 to 7,000”
8500 “7,000 to 10,000”
12000 “10,000 to 14,000”
15750 “14,000 to 17,500”
19250 “17,500 to 21,000”
28000 “21,000 to 35,0000”
42500 “35,000 to 50,000”
60000 “More than 50,000”
We will update our documentation to include the correct codes.
France 2009/2010 sample
Unfortunately our partners at the Centre for Time Use Research were not able to find any additional information about what these codes correspond to, in the original survey documentation or their own documentation. I was not able to find anything useful in the original survey codebook, which does not include a table or description of a household income table divided into five categories—only into ten categories (see table MRD2 on page 42). The original data that IPUMS received does not include any labels associated with these codes. I think it’s reasonable to assume that the codes -8 and -9 are special, non-income codes (e.g., “doesn’t know”) and that the codes 1-5 correspond to midpoints of income ranges, but we unfortunately do not have any information on what the income ranges are.
2. WORKHRS in Italy 2002 sample
I see the comparability note about the Italy 2002 sample you are referring to, but WORKHRS is not available for this sample. I am not sure why it is unavailable. The comparability note is confusing given that it is not available. I will inquire with the IPUMS MTUS team about whether we can provide this variable for the Italy 2002 sample, and in the meantime I will request that the comparability note is removed to avoid confusion.
3. NCHILD question
NCHILD counts the number of individuals under age 18 in the household. NCHILD is not specific to a given parent or adult in the household, and the children in the household may or may not have a parent in the same household. This variable is created by our partners at the Centre for Time Use Research by counting the number of person records in each household with an age under 18. In some samples, there is already a variable present in the original data that reports the number of children under 18 in the household. In these cases, the CTUR uses that information for NCHILD. You could certainly create variables counting the number of female children and the number of male children in each household.
Hi Isabel,
Thank you so much, this is all very helpful! I also thank you for updating the corresponding documentation on IPUMS as well!