Monday, December 27, 2010

12/27/2010 new TIGER files, 2010 Census, part 1

The Census Bureau is in the middle of rolling-releasing the 2010 TIGER files, containing the Census Tract boundary files to be used when mapping Census 2010 data.

With only 21 States currently available, it looks like there is a 15% increase in the number of Census Tracts (from 2000/2004 to Census 2010). Generally, “inner city” census tracts, and rural census tracts, remain unchanged, while the additional tracts come from spliting suburban census tracts (almost all population changes from 2000-to-2010 are “increases in suburban areas”).

Having said that, the City of Elizabeth, NJ shows both combining and spliting of Census Tracts (Census 2000 census tracts have yellow boundaries and labels with Times New Roman font; Census 2010 census tracts have thin black boundaries and labels with Arial font):



Elizabeth, NJ had 24 Census Tract in 2000, and 26 Census Tracts in 2010. This is only a “net increase” of 2 Census Tracts, but 4 tracts were split into 8, 4 tracts were combined into 2, and 16 tracts were not changed.

The map below shows an interesting transition. In Iredell County, North Carolina (just north of Charlotte, NC), tract 616 split into 2 tracts, tract 615 split into 3 tracts, and tract 614 split into 8 tracts:


On my website, I have sliders illustrating 1990-to-2000 census tract transitions:
http://dixonspatialconsulting.com/Flex3projects/Fade_1990_to_2000/bin-release/Fade_1990_to_2000.html#
I look forward to posting some interesting areas for the 2000-to-2010 transitions.

Monday, December 20, 2010

12/20/2010 New Releases from the U.S. Census Bureau

These are very exciting times for anyone dealing with Census Data – the Bureau is releasing 1) the first set of 5-year ACS Census Tract-level data, 2) new TIGER files for the 2010 Census, and 3) the first release of Census 2010 data, specifically to be used for House of Representative Redistricting.

On Tuesday, December 14, 2010, the US Census Bureau announced the release of the first set of 5-Year American Community Survey Estimates:

http://www.census.gov/newsroom/releases/archives/american_community_survey_acs/cb10-cn90.html

This “5-Year data set” is important because it is the first set with data at the Census Tract-level, allowing the first set of “official Census data” since of the release of the Census 2000 data. At this time, it is not known which Federal Regulators will be using this data. Regardless of any “mandatory use”, it is essential to become familiar with these data and their method-of-distribution, because this is how the US Census Bureau will disseminate the Census 2010 data (at least that data that had formerly been associated with Summary File 3 [income and housing variables, for example]).

Wednesday, Dec. 15 saw a number of newspaper articles and maps about the data, generally focusing on the growth of Hispanic population in the past few years.

The New York Times
http://www.nytimes.com/2010/12/15/us/15census.html

The Washington Post
http://www.washingtonpost.com/wp-dyn/content/article/2010/12/14/AR2010121404031.html

The Chicago Sun-Times
http://www.suntimes.com/news/2850885-418/data-according-percent-census-integration.html

The Boston Globe
http://www.boston.com/news/local/massachusetts/articles/2010/12/15/a_revealing_look_at_the_population_in_mass/

The Seattle Times
http://seattletimes.nwsource.com/html/localnews/2013680231_census15m.html

This map shows Census Tracts in the Fort Lauderdale, FL area. Comparing the just-released Median Family Income (in 2009 inflation-adjusted dollars) numbers for Census Tracts with the historical 1999 Median Family Income numbers from Census 2000, it can be seen that although most Florida Census Tracts had modest increases, some had relatively-large increases (more than a 40% increase), and some even had declining Median Family Income numbers (shaded red on the map):

Monday, December 13, 2010

12/13/2010 Income as a % of MSA Median

In my 11/22/2010 blog, I discussed the 42 intersections in the US where low/moderate/middle/upper income census tracts meet. This week I will take this opportunity to discuss why those income ranges exist.

To combat red-lining practices in the 1970's (specifically in the insurance industry, but also active in the banking industry), Congress passed the HMDA and CRA legislations. Since then, Federal Regulators have inspected bank data to see if banks are soliciting business (home mortgages) in all the Census Tracts where they are chartered to do business. They decided to focus primarily on two variables: relative income, and minority percentage (historically, areas of relatively-low income, and areas of relatively-high minority percentage were avoided by financial institutions).

Regulators decided to use the variable "Median Family Income". In every Census Tract that has "Families>0", the Median Family Income value is calculated so that 1/2 the families have income less than the Median, and 1/2 the families have income greater than the Median. Note: "the Median" is not "the Average", which is the "sum of the Incomes" divided by the "total number of Families".

To account for regional differences, the Regulators use the MSA Median Family Income to adjust each Census Tract's Income level. The Median Family Income for Cambridge, MA Metropolitan Division is $74,121; contrast that with the Median Family Income for Miami, FL Metropolitan Division - $40,266. A Census Tract with Median Family Income of $37,000 would be Low Income in Cambridge, MA, but would be Middle Income in Miami, FL.

This map shows the distribution of Census Tract Income Levels in the area of Lynn, MA:


Not surprisingly, the Census Tracts associated with downtown Lynn are all Low-Income or Moderate-Income. Next week we will see how a fictional bank has activity in this area.

Monday, December 6, 2010

12/6/2010 Moving Census Tract Centroids

This weekly blog will deal with things GIS, including Computer Cartography, Thematic Cartography, Census Data, and other areas of interest.

An important aspect of any set of Mapping Data Polygons is the location of their centroids. The centroid is where the label for the geography gets placed, and also where each pie chart (or other "info graphic") gets placed.

Historically, the position for the centroid is half-way between the maximum/minimum latitude, and half-way between the maximum/minimum longitude. When I started in this business, that was the only specification for the centroid - this resulted the label for Florida to be out in the Gulf of Mexico! The only improvement since then has been the restriction that the centroid be within the physical boundary of the geographic entity. There is still no algorithm accounting for the shape of the polygon. This results in the situation illustrated below:



(clicking on a map will take you to a larger version)

Although the upper map illustrates the confusion of mis-placed text in Census Tracts in Washington, DC, the lower map shows you specifically which Census-Tract-Centroids have been moved (the original Centroid locations are red, and the moved Centroids are yellow). The improvement in readability and data-comprehension will make your maps look more professional. Similar to Editing in a newspaper article, it is noticed most when the effort is absent - your maps will look unprofessional and unsophisticated. (When the label falls on a boundary line, the result will be confusion, and a lack of confidence in your product)

Unfortunately, this attention-to-detail does not come for free; it takes approximately 36 hours to adjust the centroids of the 66,450 Census Tracts in the United States.

Dennis C. Dixon
Geo-Information Specialist
Dixon Spatial Consulting
dennis.dixon@dixonspatialconsulting.com

Monday, November 29, 2010

11/29/2010 Why Create/Install New Mapping Data Files?

This weekly blog will deal with things GIS, including Computer Cartography, Thematic Cartography, Census Data, and other areas of interest.

The process for creating new Mapping Data is very time consuming. A valid question is – Why create/install new Mapping Data files?

The primary answer is that there has been changes in Census Tract boundaries, either wholesale (Census 2000, Census 2010), or because specific Census Tracts have been modified (beginning in 2004, the new Broomfield County, CO was required – this new County was created by splitting 14 Census Tracts).

The secondary answer is that there have been changes to the “support” cartography files: new water boundaries have been incorporated into the new data, or, as in the case below, new roads have been constructed. This example shows, in the East Hartford, CT area, the continuation/finalization of Interstate 291. The top map is from old TIGER files (pre-2007), and the bottom map is from the 2009 TIGER files:





It is, of course, a business decision on how often to upgrade your Mapping Data (annually? Every 3-or-4 years?). My experience is that users are very familiar with their own local geographies, and are quick to point out the absence of Major Highway Construction in their Mapping Software. A reliable source of relatively-current Mapping Data is essential to satisfy, and retain, clients.

Dennis C. Dixon
Geo-Information Specialist
Dixon Spatial Consulting
dennis.dixon@dixonspatialconsulting.com

Monday, November 22, 2010

11/22/2010 GeoDemographic Analysis

This weekly blog will deal with things GIS, including Computer Cartography, Thematic Cartography, Census Data, and other areas of interest.

This past week has been busy - after downloading and installing the FFIEC 2010 Census Data Application (http://www.ffiec.gov/cra/censusproducts.htm#censusdata), a Report showing Tract Median Family Income and MSA-level Median Family Income is run for all the Census Tracts, but only one State at a time. The data is saved as a txt file, and opened in MapInfo. A master table is made by appending all the txt files. After appropriate database-field-manipulations, a unique FIPS code is generated for each record, which corresponds to the unique FIPS code in the previously-created Nationwide Census Tracts file. After calculating "Income as a % of MSA Median" for each Census Tract, each Tract is either
Low-Income (0.01% < 50%)
Moderate-Income (50% < 80%)
Middle-Income (80% < 120%)
Upper-Income ( >= 120%)
NA (zero families in a Tract means no Tract Median Family Income)

For my first project, I am interested in where 4 Census Tracts meet, and each of the 4 is in a different Income Level:



This map of Midland, TX shows 4 Tracts intersecting at N Big Spring St and W/E Scharbauer Dr, and each Tract is in a different Income category. A bank that wants to locate a branch in this area would make a big mistake to locate it in Census Tract 0002.00 (the Upper Income Census Tract); they would get "extra credit" in any Examination if they were to locate their branch in a Low Income Census Tract - in this case, Tract 0016.00.

This intersection is one of 42 that I have identified. Please contact me if you would like information about the other 41 locations. I enjoyed this project - it combines the need for sophisticated (and automated) data-processing with the need for just some raw "grunt work"; but the results are worth the effort.

Dennis C. Dixon
Geo-Information Specialist
Dixon Spatial Consulting
dennis.dixon@dixonspatialconsulting.com

Monday, November 15, 2010

11/15/2010 First Thoughts

This weekly blog will deal with things GIS, including Computer Cartography, Thematic Cartography, Census Data, and other areas of interest.

The first issue is to identify your client, and their requirements. A map of Housing Index data means both county-level and MSA-level data; an appropriate visualization would be to extrapolate current counties/MSAs back in time, according to the data. For a bank Compliance Officer, a 2010 map of Census Tracts shaded by "Income as a % of MSA Median", however, would use Census Tracts (boundaries defined by the US Census Bureau), but only those FIPS Codes mandated by the Federal Financial Institutions Examination Council (the FFIEC).

Each year, the FFIEC issues "A Guide To HMDA Reporting: Getting It Right!". Among other changes, the 2010 edition was updated to reflect the US Census Bureau making a couple changes to county codes in Alaska, separating Hoonah-Angoon from Skagway-Hoonah-Angoon (232) and assigning Hoonah-Angoon (105) as a separate small county; and giving a new county code 230 to Skagway.

The FFIEC has chosen to wait until 2010 to apply a County-change in Alaska that the US Census Bureau officially noted was effective June 20, 2007. Additionally, the FFIEC has not yet implemented 2 additional Alaska County-changes, one of which was was officially noted effective 5/19/2008, and the other effective June 1, 2008.
http://www.census.gov/geo/www/ansi/changenotes.html

The point is, for 2010 Census Tract Mapping for Bank Compliance Officers, yes - remove Alaska County 232 and include Alaska Counties 105 and 232, but no - do not change Alaska Counties 130, 201, 280.

Note for HMDA Geocoding users: 904 MAIN ST, SKAGWAY, AK 99840 should geocode to County 230, Census Tract 0001.00 for 2010, but to County 232, Census Tract 0001.00 for prior years.

Dennis C. Dixon
Geo-Information Specialist
Dixon Spatial Consulting
dennis.dixon@dixonspatialconsulting.com