Status:Closed    Asked:Aug 15, 2017 - 07:27 AM

Mismatch between IPUMS and Summary File for ACS?


I just downloaded the 2014 ACS and wanted to make sure I had done it correctly (the 1 percent file). I looked at summary statistics for household income and found a mean of 85K and a median of 63k. This is substantially higher than the summary file report for 2014, which is closer to 51k. I know the summary file draws on a much bigger sample than the PUMS, but my question is whether this sizeable difference is normal, or if I have something wrong.

I am using Stata and to get my summary values I just summarized hhincome using the household weight and eliminating the 9999999 responses. Thanks in advance for any help!

Do you have the same question? Follow this Question

Staff Answer


Jeff Bloem


This is a commonly omitted detail. When performing household-level analysis with IPUMS data you need to make sure that your dataset only has one observation per household. By default, with data in rectangularized format, IPUMS USA will provide observations for each individual grouped by households. So, before summarizing your data at the household level, you need to limit your dataset to only include one individual per household. The easiest way to do this in STATA is to keep if PERNUM==1. Once you do this, you should calculate statistics that are much closer to officially published statistics.


Aug 15, 2017 - 10:49 AM

Report it


THANK YOU! That worked like a charm.


Aug 15, 2017 - 11:08 AM

Report it


Is the 1990 Census Summary 1 File available in Excel?
Generate family weights for summary statistics for families (not people or ho...
Summary Health Insurance variables in the 2014 ASEC
Totals constructed from swap values, average-above-topcode-threshold values, ...
Login   |   Register

Recently Active Members

View More »

Share |