Computer and Internet Use Variable Universe

Hi all, and thank you in advance.

IPUMS CPS contains only 9 harmonized variables and no unharmonized variables (that I can find) from the November 2021 “Computer and Internet Use Supplement.” This is notably less then the universe of questions and responses in the public use microdata file. I understand there is likely a reason for this decision, but have been unable to find any documentation/forum discussion on the matter. I wish to clarify if there is a technical or data quality reason. Is there documentation describing why the C&I Use variable choice was made, and potentially whether weights are appropriate for exercises on the expanded universe of C&I Use variables available through the Census microdata?

For background, I am studying the role of geographical disparities in physical internet infrastructure and access on a set of labor and economic outcomes. I am planning to use the “Computer and Internet Use” Census raw release to decompose internet use effects into separate use case channels for years from 2014 to 2023.

IPUMS CPS provides data from a number of CPS supplements, including the 2021 Computer and Internet Use Supplement. For some supplements, we provide a subset of the variables in the original supplement, rather than the full set of variables. While it is not possible at this time for us to provide all variables from all supplements, we are continuously working to increase the amount of data we provide and stay up to date with providing the new CPS data that are released monthly. I am not aware of current work in progress to harmonize, integrate, and release additional variables from this particular supplement; however, I will pass your interest in these data on to our IPUMS CPS team. You can merge IPUMS CPS data with the original CPS data from the Census Bureau/BLS using the variables HRHHID, HRHHID2, and PULINENO (LINENO in IPUMS) as linking keys. This would allow you to use the harmonized IPUMS data along with unharmonized variables from the original dataset that we do not yet provide.

Regarding weights, yes, it would be appropriate to use the IPUMS weights CIPWT and/or CISUPPWT and/or HWTFINL with unharmonized CPS data from the Census Bureau/BLS. These weights are present in the original data. See the description of each weight for guidance on which weight is appropriate for which types of analysis.