Thursday, March 24, 2011

Overestimates or Undercounts? Does this mean Detroit didn't lose quite so many people?

When the 2010 census count for New York City came out today, it struck me as a little light.  So I decided to compare the 2009 population estimates for the ten largest cities in the country againt that 2010 counts.  In all cases expect San Diego, the census count was lower than the 2009 estimate.  The average difference was four percent, which is four years of population growth at the national growth rate.  Here are the numbers: the first column of numbers is the 2009 estimate; the second is the 2010 count.  What is going on here?

New York City8,391,8818,175,133
Los Angeles 3,831,868 3,792,621
Chicago 2,851,268 2,695,598
Houston 2,257,9262,099,451
Phoenix 1,593,6591,445,632
Philadelphia 1,547,2971,526,006
San Antonio 1,373,6681,327,407
San Diego 1,306,3001,307,402
Dallas 1,299,5421,197,816
San Jose 964,695945,942


Matthew said...

It means Detroit didn't lose so many people in the last two years, but it still lost them.

I don't know why the 2009 estimates were so high, but no doubt you are aware the ACS numbers are not really intended to estimate population, but composition of population.

Don Coffin said...


2009 estimate: 95.707

2010 Census: 80,294

Or, the Census count was 16% below the 2009 estimate.

Since the inter-census population estimates (in Indiana, at any rate) are based on reported briths, reported deaths, and an estimate (based on the recent past) of net migration, this suggests that the estimated net migration was off by a huge amount.

The 1999 population estimate, prepared using the same methods, was 110,270; the 2000 Census count was 102,746, less than 8% below the prior year's estimate.

Interestingly, the data of which I am aware shows no comparable decline in housing units or in occupied housing units for Gary...

So I have no idea what's going on. (And, as I work at a university in Gary, I have been asked...)

Don Coffin said...

Here are the 9 largest cities in Indiana (OK, so I'm obsessive; I live and work in Indiana). The 2010 estimate is based on the 2000 to 2009 estimates and assumes the same annual rate of growth between 2009 and 2010 as occurred between 2000 and 2009 in the estimates. The variance is the difference between the 2010 estimate and the 2010 Census:

2010 Estimate: 810,449
2010 Census: 829,718
Variance +2.4%

Fort Wayne:
2010 Estimate: 256,111
2010 Census: 253,691
Variance -0.9%

2010 Estimate: 121,582
2010 Census: 116,073
Variance: +4.7%

South Bend
2010 Estimate: 103,764
2010 Census: 101,168
Variance: -2.5%

2010 Estimate: 94,982
2010 Census: 80,294
Variance: -15.5%

2010 Estimate: 75,877
2010 Census: 80,830
Variance: +6.5%

2020 Estimate: 72,037
2010 Census: 80,405
Variance: +11.6%

2010 Estimate: 71,595
2010 Census: 79,191
Variance: +10.6%

2010 Estimate: 67,459
2010 Census: 70,085
Variance: 3.9%

Make of that what you will.