Analyzing microdata

At a high level, whenever you are analyzing microdata, you are summing weights to calculate population estimates for various subgroups.

When using microdata, comparing margins of error or CVs is more difficult because you will need to calculate the margins of error and the CVs. Microdata is more “raw” survey data and doesn’t come with the MOEs like the pre-calculated tables.

While you can do this in Excel, or even within the Census Bureau microdata tool, we highly recommend doing this programmatically using a scripting software like R, Python, Stata, SPSS, etc. Analyzing microdata with Excel is possible, but not recommended because of the lack of documentation and the inflexibility (if any change is needed, you often have to re-do the whole process).

This documentation focuses on R. We recommend using the tidycensus package. When analyzing microdata with R, you’ll use the tidycensus package with the srvyr package. Reference these resources on how to calculate MOEs and analyze microdata in R:

• Introductory documentation on Working with Census microdata

• Documentation on Margins of Error

• Guidance on calculating standard error, margins of error, and CVs using the srvyr package

• More in-depth guidance in this complete e-book on tidycensus by its creator, Kyle Walker