Are weights necessary if we are just computing frequency percentages for the Race and Occupation variables ?

Are weights necessary if we are just computing frequency percentages for the Race and Occupation variables from 1980-2012?

While some of the IPUMS-USA datasets are flat samples, the years you are interested in require the use of weights due to the complex sample designs employed in the creation of the datasets. The Census Bureau recommends using the weights to generate representative statistics.

I hope this helps.