ACS Data Users Group

Hi all, I am working with the 2009-2013 ACS 5-year estimates detailed tables, and have a general question about comparing ACS data at different geographic levels. I have downloaded ACS data at both the census tract and the ZCTA levels. Does anyone know whether the same raw data is used for all of ACS's geographic levels (i.e. whether I should expect the ZCTA and census tract ACS data to be "equivalent"), or if the data from these two geographic levels draw from different samples? I am primarily asking because I am planning to crosswalk both of these datasets to the 5-digit ZIP code level, and would like to know what to expect when comparing the resulting numbers to each other.

Yes. the tract level estimates and ZCTA level estimates are derived from the same underlying sample.

Thank you, Matthew!

I have a related question... we've noticed that total estimated population is about 20K lower for ZCTAs than it is for counties and tracts in the 2020 ACS 5-year. What could explain this "gap"?

Is this a relic of sampling methods? Perhaps respondents didn't enter in a ZIP that has a counterpart in the ZCTA file?

I don't believe ZCTAs cover the entire country.
It's definitely not based on the respondents' ZIP; ZCTAs are explicitly different than mailing zip codes, based on coordinate location, not address.

[quote userid="3025" url="~/discussion-forum/f/forum/1269/comparing-acs-data-from-different-geographic-levels-zcta-vs-census-tracts/3405#3405"]I don't believe ZCTAs cover the entire country.[/quote]

That's right, they don't.

Margins of error might also account for some of the difference.

For 2017-2021, I'm counting a total tract (and county) population of 333,036,755, and a total ZCTA population of 333,032,934, or a difference of 3,281.

For 2016-2020, the difference is 21,905.

for 2015-2019, the difference is 17,490.

The ACS "raw" data, i.e. the survey form answers, is based on something called the Master Address File, which contains geocoded data for each structure in the US. You can look up an address in the MAF using the census geocoder, https://geocoding.geo.census.gov/geocoder/geographies/onelineaddress?form. Every physical "address" has an assigned value for every level of geography, block group, tract, ZCTA (zip code tabulation area), county state PUMA etc.etc. Note ZCTA's are not the same as postal zip codes which change all the time. ACS population totals and some "marginals" (hispanic and race for example) are controlled at the county level. If you take the ACS values for total population B01003 and add them up across all the tracts in a county you will get the total population for the county. Tracts in a zcta do not necessarily add up to the total population of the ZCTA. Counties add up to states. Some other characteristics are "controlled." For example the number of females in all the tracts in a county add up to the number of females in the county (B01001) -- I think. In general "things" don't "add up." The ACS is a survey and the numbers on data.census.gov are estimates with sampling error.

www.census.gov/.../2021-01.html

Yes, it does appear that the margins of error are on average higher for ZCTAs than for county or tract.

My math for 2016-2020:

	SUM E_TOTPOP	MAX E_TOTPOP	AVG E_TOTPOP	STDEV E_TOTPOP	MAX M_TOTPOP*	AVG M_TOTPOP*	STDEV M_TOTPOP*
County	326,569,308	10,040,682	103,903	332,097	1,564	7	51.5
Tract	326,569,308	39,373	3,882	1,657	5,676	555	295.95
ZCTA	326,549,615	126,310	9,898	14,762	6,567	598	660.2

*nulls excluded

Thanks, that does make sense. And as I re-read the documentation, I see it was says ZIPs are coming from MAF/TIGER.

Something that is confusing me, though, is that is says "all inhabited areas have 2020 ZCTA coverage". Shouldn't this imply that all population is accounted for? Maybe they're referring to inhabited areas according to the decennial count?

Hmm, I did a quick check and the differences between ZCTA totals and county totals by state don't seem to be correlated with whether a state has lots of ZCTA "holes".

state	Stab	ZCTA_Pop	County_Pop	difference
1	AL	4,892,929	4,893,186	-257
2	AK	736,341	736,990	-649
4	AZ	7,176,644	7,174,064	2580
5	AR	3,012,005	3,011,873	132
6	CA	39338103	39346023	-7920
8	CO	5,685,154	5,684,926	228
9	CT	3,570,549	3,570,549	0
10	DE	967,679	967,679	0
11	DC	701,974	701,974	0
12	FL	21214455	21216924	-2469
13	GA	10516211	10516579	-368
15	HI	1,420,074	1,420,074	0
16	ID	1,754,012	1,754,367	-355
17	IL	12716106	12716164	-58
18	IN	6,696,688	6,696,893	-205
19	IA	3,150,430	3,150,011	419
20	KS	2,912,557	2,912,619	-62
21	KY	4,468,786	4,461,952	6834
22	LA	4,663,920	4,664,616	-696
23	ME	1,340,763	1,340,825	-62
24	MD	6,037,624	6,037,624	0
25	MA	6,873,004	6,873,003	1
26	MI	9,973,857	9,973,907	-50
27	MN	5,600,601	5,600,166	435
28	MS	2,981,835	2,981,835	0
29	MO	6,124,441	6,124,160	281
30	MT	1,062,386	1,061,705	681
31	NE	1,924,371	1,923,826	545
32	NV	3,030,453	3,030,281	172
33	NH	1,355,284	1,355,244	40
34	NJ	8,885,418	8,885,418	0
35	NM	2,094,668	2,097,021	-2353
36	NY	19514719	19514849	-130
37	NC	10386119	10386227	-108
38	ND	759,391	760,394	-1003
39	OH	11675275	11675275	0
40	OK	3,949,760	3,949,342	418
41	OR	4,176,208	4,176,346	-138
42	PA	12794885	12794885	0
44	RI	1,057,798	1,057,798	0
45	SC	5,091,517	5,091,517	0
46	SD	878,569	879,336	-767
47	TN	6,765,264	6,772,268	-7004
48	TX	28633873	28635442	-1569
49	UT	3,144,764	3,151,239	-6475
50	VT	624,308	624,340	-32
51	VA	8,509,607	8,509,358	249
53	WA	7,512,546	7,512,465	81
54	WV	1,807,137	1,807,426	-289
55	WI	5,807,020	5,806,975	45
56	WY	581,533	581,348	185

Note that ZCTAs are not assigned to states by the Bureau. (We do that ourselves at MCDC.)

The thing that's most beguiling about this isn't the differences, but that some states have exactly matching numbers. If margins of error were the culprit, I wouldn't really expect any states to have exactly matching numbers. Note that there are a very small number of zip codes that cross state boundaries, but I don't think that's what's going on here.

Right?! It's baffling.

Differential privacy? Surely not at the level of counties and ZCTAs?

somewhere the is a detailed methods whitepaper about how disclosure avoidance methods are applied to the ACS. There is quite a lot of data massaging so there is more than just sampling error in the data.census.gov. I can't seem to find the documentation