Wrong MSA: Messing up the population counts

[Dirt Roads]

As I read (and contribute) to various threads about city population, it is amazing how our Census Bureau still uses old boundary concepts as the basis for determining the population of a Metropolitan Statistical Areas (MSAs) and Micropolitan Statistical Areas (μSAs).  When these are adjacent, the reported populations don't always match up with reality.  Recently mentioned in another thread:

• Portions of Raleigh are located in Durham County, the main part of the Durham-Chapel Hill MSA.
• A small portion of Morrisville (primarily Wake County) is also located in Durham County.
• A small portion of Durham is located in Wake County, the main part of the Raleigh-Cary MSA.

Additionally, a good sized portion of Apex (primarily Wake County) extends into Chatham County, part of the Durham-Chapel Hill MSA.  It is also fair to say that Cary has got their eyes on an adjacent portion in Chatham County.  For the record, most (if not all) of Chatham County east of Jordan Lake and downstream Haw River/Cape Fear River should belong to the Raleigh-Cary MSA.  Along with keeping Raleigh residents in the Raleigh-Cary MSA, and Durham residents in the Durham-Chapel Hill MSA, this would be a fairly easy correction for the Census Bureau.

When adjacent MSAs are as close as these two, such problems with population counts are likely to occur.  Perhaps this is the worst one.  Others?

[ZLoth]

It's because the population density of the eastern half of the continental United State is higher than the western half combined with the average land area of the United States. Having said that, I prefer to reference the areas by their MSAs rather than the city areas because of the population living in the cheaper suburbs and commuting to work, thus providing a truer image of what represents an area.

[Dirt Roads]

It's because the population density of the eastern half of the continental United State is higher than the western half combined with the average land area of the United States. Having said that, I prefer to reference the areas by their MSAs rather than the city areas because of the population living in the cheaper suburbs and commuting to work, thus providing a truer image of what represents an area.

Completely agree.  I should mention that I generally support of how the United States Census Bureau puts together the population of an MSA.  For a small city of say 250,000 residents, including several outlying counties that have populations around 10,000 each only adds 10% or less to the overall population.  For a city of 1,000,000 residents, including say four surrounding suburban counties and again several outlying counties, this same methodology only adds about 2.5% to the overall population.  That's well in the ballpark for comparison purposes.

The issue that I am referencing is where the MSAs overlap to the point where part of the logical population gets removed from one MSA and counted towards the wrong one (or simply missing for some other reason).

A prime example is Midland and Odessa.  The Midland MSA is entirely Midland County, and the Odessa MSA is entirely Ector County.  But a good chunk of Odessa crosses into Midland County, and small chunk of Midland crosses into Martin County.  It looks like the Census Bureau [secretly] adjusted the Midland MSA to include more than all of Midland County, but Odessa still got stuck with the exact count of Ector County.  I'm sure that there are other examples out West as well.

[kalvado]

It's because the population density of the eastern half of the continental United State is higher than the western half combined with the average land area of the United States. Having said that, I prefer to reference the areas by their MSAs rather than the city areas because of the population living in the cheaper suburbs and commuting to work, thus providing a truer image of what represents an area.

Completely agree.  I should mention that I generally support of how the United States Census Bureau puts together the population of an MSA.  For a small city of say 250,000 residents, including several outlying counties that have populations around 10,000 each only adds 10% or less to the overall population.  For a city of 1,000,000 residents, including say four surrounding suburban counties and again several outlying counties, this same methodology only adds about 2.5% to the overall population.  That's well in the ballpark for comparison purposes.

The issue that I am referencing is where the MSAs overlap to the point where part of the logical population gets removed from one MSA and counted towards the wrong one (or simply missing for some other reason).

A prime example is Midland and Odessa.  The Midland MSA is entirely Midland County, and the Odessa MSA is entirely Ector County.  But a good chunk of Odessa crosses into Midland County, and small chunk of Midland crosses into Martin County.  It looks like the Census Bureau [secretly] adjusted the Midland MSA to include more than all of Midland County, but Odessa still got stuck with the exact count of Ector County.  I'm sure that there are other examples out West as well.

And when things are that close, how do you reliably separate them?
Blue banana concept didn't appear out of the blue

[NWI_Irish96]

It's because the population density of the eastern half of the continental United State is higher than the western half combined with the average land area of the United States. Having said that, I prefer to reference the areas by their MSAs rather than the city areas because of the population living in the cheaper suburbs and commuting to work, thus providing a truer image of what represents an area.

Completely agree.  I should mention that I generally support of how the United States Census Bureau puts together the population of an MSA.  For a small city of say 250,000 residents, including several outlying counties that have populations around 10,000 each only adds 10% or less to the overall population.  For a city of 1,000,000 residents, including say four surrounding suburban counties and again several outlying counties, this same methodology only adds about 2.5% to the overall population.  That's well in the ballpark for comparison purposes.

The issue that I am referencing is where the MSAs overlap to the point where part of the logical population gets removed from one MSA and counted towards the wrong one (or simply missing for some other reason).

A prime example is Midland and Odessa.  The Midland MSA is entirely Midland County, and the Odessa MSA is entirely Ector County.  But a good chunk of Odessa crosses into Midland County, and small chunk of Midland crosses into Martin County.  It looks like the Census Bureau [secretly] adjusted the Midland MSA to include more than all of Midland County, but Odessa still got stuck with the exact count of Ector County.  I'm sure that there are other examples out West as well.

To be clear, we don't secretly do anything.

Counties/parishes are the base units for MSA definitions because they are the lowest level of geography that's consistent across all 50 states. Trying to us any subdivisions of a county would lead to inconsistencies from state to state.

The percentage of a county's population that commutes to the main county in the MSA is the primary factor in determining whether or not a county is included in the MSA.

[CNGL-Leudimin]

My main complain about MSAs is another. Really Needles or Blythe need to be part of the Riverside/San Bernardino/Ontario CA MSA? Wish they could do like the NWS with its county warning areas, where the San Diego office extends its area of responsability only as far as the Victor and Coachella Valleys and Las Vegas and Phoenix take up the remainder of San Bernardino and Riverside counties respectively.

Other quirks of MSAs being defined by county is that part of Sikeston MO actually falls outside its μSA, and that the Holland MI μSA only includes a small part of Holland proper (in fact it used to be the Allegan MI μSA until the original Holland MI MSA was absorbed by Grand Rapids/Wyoming).

[Dirt Roads]

A prime example is Midland and Odessa.  The Midland MSA is entirely Midland County, and the Odessa MSA is entirely Ector County.  But a good chunk of Odessa crosses into Midland County, and small chunk of Midland crosses into Martin County.  It looks like the Census Bureau [secretly] adjusted the Midland MSA to include more than all of Midland County, but Odessa still got stuck with the exact count of Ector County.  I'm sure that there are other examples out West as well.

Correction.  According to 2020 Census Bureau map, the Midland MSA has been increased to include Martin County.  The Midland MSA population was 175,220; Midland County was 169,983 and Martin County was 5,237.  Other sources incorrectly state that Midland County is the only county within the MSA and show the population as such.

But that does nothing to address the fact that a good chunk of Odessa is within Midland County.  When the final census reports come out, that datum will be available in the Midland County breakdowns (and perhaps other datum related to population with Odessa zip codes in Midland County)

To be clear, we don't secretly do anything.

Sorry, didn't mean to trigger anything with the [transcribed substitute word].  Occasionally, we all pick a word that is not correct for the particular situation.  Does this mean that you perhaps work for/with the Census Bureau?

[bandit957]

Are they still coming out with new MSA definitions in June?

[NWI_Irish96]

To be clear, we don't secretly do anything.

Sorry, didn't mean to trigger anything with the [transcribed substitute word].  Occasionally, we all pick a word that is not correct for the particular situation.  Does this mean that you perhaps work for/with the Census Bureau?

I do, though I do not work in the population division, so I don't have any knowledge about this subject that isn't publicly available.

[chrisdiaz]

I feel like its weird that they are not able to split counties into multiple different MSAs - or at least two. For example. Brunswick County, NC is included in the Myrtle Beach MSA. I do agree, that some parts of Brunswick County are aligned with Myrtle Beach, such as Calabash, Sunset Beach, and Ocean Isle Beach. However, Leland is SO much more aligned with Wilmington. It just feels weird for it to be included in the Myrtle Beach MSA despite it literally being just across a river from Wilmington.

[NWI_Irish96]

I feel like its weird that they are not able to split counties into multiple different MSAs - or at least two. For example. Brunswick County, NC is included in the Myrtle Beach MSA. I do agree, that some parts of Brunswick County are aligned with Myrtle Beach, such as Calabash, Sunset Beach, and Ocean Isle Beach. However, Leland is SO much more aligned with Wilmington. It just feels weird for it to be included in the Myrtle Beach MSA despite it literally being just across a river from Wilmington.

As I noted upthread, once you get below the county level, subdivisions are not consistent across states. That's the largest problem when it comes to splitting counties.

[Dirt Roads]

I feel like its weird that they are not able to split counties into multiple different MSAs - or at least two. For example. Brunswick County, NC is included in the Myrtle Beach MSA. I do agree, that some parts of Brunswick County are aligned with Myrtle Beach, such as Calabash, Sunset Beach, and Ocean Isle Beach. However, Leland is SO much more aligned with Wilmington. It just feels weird for it to be included in the Myrtle Beach MSA despite it literally being just across a river from Wilmington.

The same thing is true up on the other end of the North Carolina coast.  Currituck County is part of the Virginia Beach-Norfolk-Newport News MSA, but Corolla certainly belongs with Duck and the rest of the Kill Devil Hills μSA.  Perhaps even worse, good-sized portions of Gates County and Camden County, plus a small portion of Currituck County should belong to the Elizabeth City μSA instead of Virginia Beach-Norfolk-Newport News MSA. 

[Flint1979]

I feel like its weird that they are not able to split counties into multiple different MSAs - or at least two. For example. Brunswick County, NC is included in the Myrtle Beach MSA. I do agree, that some parts of Brunswick County are aligned with Myrtle Beach, such as Calabash, Sunset Beach, and Ocean Isle Beach. However, Leland is SO much more aligned with Wilmington. It just feels weird for it to be included in the Myrtle Beach MSA despite it literally being just across a river from Wilmington.

As I noted upthread, once you get below the county level, subdivisions are not consistent across states. That's the largest problem when it comes to splitting counties.

Right like with my local area. Saginaw County consists of Metro Saginaw, Bay County consists of Metro Bay City, Midland County consists of Metro Midland but they are all a part of the Tri-Cities where it is called the Saginaw-Midland-Bay City CSA. Flint is another one, it's metro area consists of just Genesee County.

[skluth]

I was going to write a long diatribe but really it all boiled down to "Geography is not an exact science." I personally never noticed that Raleigh and Durham-Chapel Hill were two distinct metros when I visited it, (I'm actually surprised today to find out that they aren't.) much like LA/Inland Empire and Midland-Odessa. It's not surprising that all are grouped in CSAs. All feel more like Tidewater which is correctly classified as one big metro though locals often refer to the Peninsula as a separate entity from the South Side; I visited the Outer Banks as often as the Peninsula when I lived in Portsmouth.

Could the Census Bureau use something other than county boundaries? Yes, especially these days when it is relatively simple to parse the data compared to even fifty years ago. Should they? I don't know. But I guarantee if they changed it people would still bitch because it's not what they would have done.

[Dirt Roads]

How about when an entire city is split between MSAs.  Nitro, West Virginia (population 6,618) is split between Kanawha County and Putnam County.  Putnam County was reassigned to the Huntington-Ashland-Ironton MSA back before the 2010 Census.  I can't find a recent split, but looking back historically about 17% of Nitro's population was in Putnam County.  You could make a very, very, very strong argument that the entire population of Poca, Rock Branch and everything east of the Pocatalico River belongs to the Charleston MSA. 

I'm still having a hard time with my hometown of Scott Depot being placed in the Huntington-Ashland-Ironton MSA, but the truth is that my family was more representative of the opposing argument so I'd better keep hushed.

[bandit957]

Some of the far northern reaches of the Cincinnati MSA might be in the Dayton MSA if they could split counties, especially since the Dayton "urban area" goes into these places.

[SkyPesos]

I'm more surprised that Raleigh and Durham are separate MSAs, considering there's the RDU airport and Research Triangle area between the two, probably taking in commuters from both cities. I thought it would be a single one like the Twin Cities.

[Flint1979]

I'm more surprised that Raleigh and Durham are separate MSAs, considering there's the RDU airport and Research Triangle area between the two, probably taking in commuters from both cities. I thought it would be a single one like the Twin Cities.

That seems strange. The distance between the two downtowns is about the same from downtown Detroit to downtown Pontiac following Woodward.

[golden eagle]

I'm more surprised that Raleigh and Durham are separate MSAs, considering there's the RDU airport and Research Triangle area between the two, probably taking in commuters from both cities. I thought it would be a single one like the Twin Cities.

Don't Minneapolis and St. Paul border each other? That's probably why. I'm pretty sure Raleigh and Durham are getting close to touching each other (if they don't already) due to annexations, but their downtowns are still rather far apart. But then again, Dallas and Fort Worth do share the same metro.

[bandit957]

The government has very exact rules on this stuff.

[doorknob60]

It bothers me that Davis County, UT is part of the Ogden MSA. What that means is that the city of North Salt Lake, which literally borders Salt Lake City, is not in SLC's MSA. The problem is that the Clearfield/Layton area, which is also part of Davis County, does belong in Ogden's MSA. And I am not opposed to having Ogden and SLC be separate MSAs, it's something that does make sense to me (as does the separate Provo/Orem MSA). But in ideal world we could split the county into two different MSAs, with the cutoff around Farmington. I know it doesn't work like that, but maybe it should.

[bandit957]

I would guess the cutoff between Cincinnati and Dayton is around Middletown, but Middletown has its own "urban area." Thus, if MSA's were built on subcounty areas, Middletown might have its own MSA.

I guess the cutoff between Cincinnati and Lexington is around Corinth, but maybe Dry Ridge, since a lot of people in Dry Ridge have listened to Lexington radio.

[Rothman]

It bothers me that Davis County, UT is part of the Ogden MSA. What that means is that the city of North Salt Lake, which literally borders Salt Lake City, is not in SLC's MSA. The problem is that the Clearfield/Layton area, which is also part of Davis County, does belong in Ogden's MSA. And I am not opposed to having Ogden and SLC be separate MSAs, it's something that does make sense to me (as does the separate Provo/Orem MSA). But in ideal world we could split the county into two different MSAs, with the cutoff around Farmington. I know it doesn't work like that, but maybe it should.

North Salt Lake is dinky.  I don't see this mattering one way or another.

[tmthyvs]

Northern Colorado has always been interesting to me. North of the Denver MSA are 3 smaller MSAs: Boulder (Boulder County), Fort Collins (Larimer County), and Greeley (Weld County). At the CSA level, Boulder naturally gets folded into Denver. That makes a ton of sense. Greeley also gets folded into Denver. But that's not because there's so much traffic between Greeley and Denver; it's because the southern parts of Weld County (Firestone, Frederick, Dacono, Erie) are much more aligned to the Denver and Boulder areas than to Greeley. Move the southern border of Weld County north to where the southern edge of Larimer county is, and I'd wager you'd have the Denver-Boulder CSA and the Fort Collins-Greeley CSA.

[bandit957]

Would the "urban areas" be a good substitute for MSA's? These are broken down to the census block level. But they are designed to include only areas that are urban or suburban, not the rural parts of counties or county subdivisions.