Panel Status of Mexico Censuses

I am currently working with census data from Mexico for the years 1990 to 2000. I have a question about the nature of these data.

  • Can these data be considered panel data, or are they just repeated cross-sectional data?

If they can be considered panel data:

  1. Does the variable SERIAL identify the same household across different censuses? For instance, would the household with SERIAL number 10001 in the 2000 census be the same household with SERIAL number 10001 in the 1990 census?

  2. Additionally, does the variable PERNUM refer to the same individual within a household across different census years?

In other words, can I use these two variables together to reliably track the same person over time? If I can’t use these two variables, what variables can I use?

Thanks for your help.

The census of Mexico is a cross-sectional survey of the entire population. From the 1990 and 2000 censuses, IPUMS International provides samples (subsets) of the full datasets. You can see details of the sample characteristics here. Since the census of Mexico is a population census, meaning each citizen of Mexico is included in the full dataset, many individuals are observed in both censuses. However, these are not pre-linked data and identifying characteristics like names are removed to preserve confidentiality, so it is not possible to track individuals or households from one census to another. Also, since the data IPUMS provides are samples (and not the full datasets that include all citizens), there are likely many individuals who are in the 1990 sample but not the 2000 sample, and vice versa. The variable SERIAL uniquely identifies a household within a sample. The combination of SERIAL and PERNUM uniquely identifies an individual within a sample. I am not sure whether our international partner in Mexico, INEGI, provides restricted use versions of the data in which individuals or households are linked across time.

Thank you for your reply.
I understand that the variable SERIAL uniquely identifies a household within a sample. However, what is still unclear to me is whether this number is fixed for a household over time. For instance, is the household identified by the SERIAL number 10001 in 1990 the same household as the one identified by the SERIAL number 10001 in 2000?
Thank you for your assistance.

The specific variable SERIAL is not in the original data IPUMS receives from INEGI; SERIAL is created by IPUMS when processing the sample and simply assigns a 1 to the first household in the sample, 2 to the second household, all the way to the nth household. When we process a new sample, the SERIAL values for the new sample will also range from 1-n without any consideration for other samples. PERNUM is assigned the same way, but within households rather than within samples.

Thank you very much for the clarification.