Analyze Online: Confidence interval for Weighted N

USA
1

Hello,
Is there some way to use the Analyze Data Online tool to create an export with Weighted Ns and the the low and high end of Confidence Intervals showing for the Weighted Ns in a numerical values?

I tried using the options under “Output Options” > Cell Contents: Percentaging > Confidence Intervals, but the CI’s are in percentages and I’m not sure what I’m getting as the output.

Example:
Row: year (2012-2020)
Column: migrate1d(24, 31, 32, 40)
Filter: city(6290)

I exported once with cell contents and percentages for columns selected and once with rows.

The percentage for CI-Low 95 for the cell year 2012 and “24-Different States” in the column version x the column total for the sum of the years is: 11.13% x 286,417 (sum of the years column) = 31,893.

The row version of CI Low 95 for year 2012 and “24-Different states” = 48.11% x 67,285 (sum of the migration categories in the row for year 2012) = 32,370.

I would prefer an export with the Weighted N’s CI low and high calculated, but if not, is the row version the correct CI low and high range? It makes sense to add up the migration categories for the year’s total number of people for the year, but does not make sense to add up the column of a migration category for each year.
Thanks!

2

The online guide notes that confidence intervals using the SDA system are only available for percentages. As you state, adding percentages by migration category across years doesn’t make much sense since you’re analyzing net migration over time. For this reason, you’ll want to generate percentages across rows and calculate confidence intervals from there. Alternatively, you can calculate your own confidence intervals by downloading an extract with the above samples and variables directly from IPUMS USA.

1 Like