Hi,
I used UC Berkely SDA online software to tabulate some data from ACS using the user guide PDF pages online. The tabulation shows that the weighted N is in billions of people. Why is that? Even the user guide PDF pages show ACS tabulation by race and labor force status and the last number in each cell in the table is in billions. I would appreciate help understanding this.
Thank you so much.
Basel Saleh
I’m guessing you’re tabulating variables for multiple samples pooled together and applying weights. This will result in frequency estimates that are much larger than are correct. You can divide the frequency estimate by the number of samples you’ve pooled together to approximate the average frequency (number of observations fitting specified criteria, like race) over the entire time period.
For example, I tabulated RACE for all ACS samples from 2001-2024 and selected PERWT as the weight. Here is the table output:
I’ve pooled together 24 samples (2001-2024). Dividing 7,492,230,308 (the column total frequency) by 24 yields 312,176,262, which you can interpret as the average U.S. population between 2001 and 2024. You do not need to make any adjustments to the percentages estimated or non-frequency estimates. The percentages here represent the average percent of the U.S. population that was each race between 2001 and 2024.
Thank You, Isabel. This is really helpful.
Best,
Basel