from the description of MIGCITY5 and MIGICITY1 it is stated that they report the city where the respondent resided 5 years ago or 1 year ago, if that city could be identified under the confidentiality requirements of a given sample.
For all those cases in which migcity5 (migcity1) takes value “N/A and not identifiable” (“N/A”), is there a way to assess how many individuals didn’t provide info on the city of migration and how many of them reported their city but it is not identifiable due to privacy requirements?
The variables MIGCITY5 and MIGCITY1 are actually created by IPUMS-USA based on MIGCOGRP (for MIGCITY5 in 1980), MIGPUMA (for MIGCITY5 in 1990 and 2000), and MIGPUMA1 (for MIGCITY1 from 2001-2011). The Census Bureau does not actually release the address gathered from the “Where did this person live 1 year ago?” question. Instead, the Census Bureau uses the response to place the respondent within a MIGPUMA. MIGPUMA boundaries (likePUMA boundaries) are drawn based on population first and foremost (requiring at least 100,000 people to reside within the boundaries). When the boundaries of MIGPUMAs do not line up with the boundaries of a city (which is especially common for smaller cities) that MIGCITY is considered “Not Identifiable” and all respondents from the MIGPUMAs that do not match MIGCITY boundaries are given a code of ‘0000’ for MIGCITY. MIGPUMA, and MIGPUMA1 both have N/A codes of ‘000’ which generally means the respondent didn’t provide geographic information about prior place of residence because they didn’t move. So, you could assign a unique code to MIGCITY1 and MIGCITY5 based on if MIGPUMA or MIGPUMA1 are coded as 0 (or 990 for MIGPUMA), this would identify persons who did not provide any geographic information for their previous location.
Thank you very much for your help! Could I ask one more question? I understand how a respondent is associated to a MIGPUMA but it is not clear to me how individuals who did not answer to the question “Where did this person live 1/5 year/s ago?” (leaving the answer blank) are treated by IPUMS-USA, if any . Are they simply associated to MIGPUMA with a missing value? Sorry if I insist but I need to know whether the group of individuals whose city of migration is not reported are just those living in a MIGPUMA which does not coincide with the MIGCITY boundaries or they also include individuals who deliberately decided not to report where they were living 1 or 5 years ago.
Hi, in relation to my previous question I can provide the following example: in my sample 2,243,848 out of 4,123,907 observations are associated to the variable migcogrp=0 (N/A). 36,395 of them have migcogrp=0 because they were living abroad while 299,705 have migcogrp=0 because they were not born in 1975. I am left with 1,944,143 for whom it is not possible to understand why migpuma=0: is it because they didn’t answer to the census questionnaire or were they living in a County group with less than 100,000 individuals?
Since you are using MIGCOGRP I assume you are looking at 1980, which was a unique sample for migration variables. As mentioned in the MIGCOGRP Universe Statement, this question was only asked of people aged 5+ (as you noted), and only of 50% of the population (see MIGSAMP). This should explain the rest of the zeros. For MIGPUMA and MIGPUMA1, the question was only asked if the respondent had lived in another house 1/5 year/s ago (you can use MIGRATE1 and MIGRATE5 to identify people who did or did not move). Just like with PUMA, the variables MIGPUMA/1 and MIGCOGRP cover the entire country and represent the smallest identifiable geographic indicator. If there was an area that did not have 100,000 people it would have been combined with an adjacent area and therefore become identifiable. Only variables like MIGCITY can become “Not Identifiable” because they depend on matching up with MIGPUMA boundaries.
I think I am left with one doubt and I would really appreciate if you could help me dissipating it.
Actually I am looking at 1980 (MIGCOGRP), 1990-2000 (MIGPUMA) and 2005-2010 (MIGPUMA1). In relation to this last variable, I was trying to understand why people report MIGPUMA1=0. As before, some of them were not born 1 year ago, some didn’t move and others were living abroad. However, 2000ish individuals are associated to a migpuma1=0 even if, looking at migrate1d variable, I can see that they “moved between non-contiguous states”. Using migplac1, I found out that they were all living in Puerto Rico 1 year before.
Is there a reason why these observations are not pull together in the category “living abroad 1 year ago”?
I will never be able to thank you enough for your help!
The Census Bureau actually made the decision to not include Puerto Rico in the “Abroad” Category. As you can see in the Questionnaire Text for the migration questions, the space that was given for a respondent to provide a foreign country uses the prompt “outside the United States and Puerto Rico”. But, since the USA files do not include PUMA codes for Puerto Rico, MIGPUMA1 is set to zero.