Replicating health insurance estimates from ACS


#1

Trying to understand why I am not replicating some results in Table HI05, found here: http://www.census.gov/hhes/www/hlthins/data/incpovhlth/2013/acs-tables.html

I downloaded 2013 IPUMS ACS data, and am using Stata.

For example, I typed this command and got these results:

. tab hcovpriv if age < 65 & year ==2013 [w=perwt]

(frequency weights assumed)

Private health insurance coverage | Freq. Percent Cum.

----------------------------------------±----------------------------------

Without private health insurance covera | 94,770,598 34.91 34.91

With private health insurance coverage |176,694,873 65.09 100.00

----------------------------------------±----------------------------------

Total |271,465,471 100.00

Table HI05 gives the population number for this group as 267,804,470 (i.e., somewhat lower), and the number with private insurance as 175,253,980 (also lower, but giving a somewhat higher percentage of 65.44). Do I have the wrong sample, or am I using the wrong weights, or is it something else? Thanks.


#2

In order to match the official ACS table, you need to restrict your sample to civilian, non-institutionalized respondents. Respondents with EMPSTATD = 14 or 15 are members of the Armed Forces and should not be included in your sample. Respondents with GQTYPE = 1 are institutionalized and should not be included in your sample. After excluding these respondents, the results using the IPUMS data fall within the margin of error for the official ACS table.

Hope this helps.