Cubit's Blog

Photo by Justin Eisner on Unsplash

Occasionally, I pull school district population data for clients as a custom data request. Here are 2 tools & 1 technique for pulling demographic data for school districts.

Background

While the US Census Bureau does help collect school district population data, I can’t find school district data on the Census Bureau’s website. For example while I can select “all school districts in New York” in the American Fact Finder 2, I can’t actually pull any demographic or population data for the school districts. The Census Bureau’s website says that they have a project called the Education Demographic and Geographic Estimates project which “produces a variety of geodemographic data for the National Center for Education Statistics.” Ah! Now we are getting somewhere.

Let’s head over to the National Center for Education Statistics website. They have a cool tool called the School District Demographics Systems (SDDS).

Tool 1: NCES’ School District Demographic Systems

• Good for basic demographics & heat maps
• Bad for detailed demographics & pulling the most current demographic data available

Positive: SDDS is Easy to Use

Make your selections in the green box to right, and the map will change accordingly. For example, here’s a map of the Under 18 Hispanic persons for all Texas Unified School Districts.

If you want the actual data that’s displayed on the map, click on the Data Table button. Then click the Save to File button. Pretty nifty, huh?

You can get the following data points:

• Total Students
• Total Population
• Total Occupied Housing Units
• Owner Occupied Housing Units
• Renter Occupied Housing Units
• Average Household Size
• Average Family Size
• Percentage – Householder/Male
• Percentage – Householder/Female
• White Alone
• Black or African American alone
• Asian Alone
• Hispanic or Latino
• American Indian and Alaska Native
• Native Hawaiian and Other Pacific Islander
• OtherMulti-Race
• Total Population Under 18
• Sex by Age Breakdown
• Total Teachers
• Total Librarians / Media Specialists
• Total Revenue

Positive: SDDS Displays Individual Schools & School Boundaries

To see individual schools, select schools in the greenbox to the right. Zoom in to get more detailed data about individual schools. In the map below, you can see schools by type for schools in Harris County, Texas. In addition to school district boundaries, you can also get see individual school boundaries by clicking on Map Layers & changing the selections there.

Positive: Export Spatial Data for School Districts with the SDDS

If you want to build custom maps or do spatial analyses, you can also export the spatial data. Click on the Map Tools button, and then click on Get Spatial data. Select the data which best meets your needs – Census, ACS or SAIPE. If you don’t already know what dataset to use, you’ll probably need a little help from a demographic data nerd or spend a couple of hours digging on the Census Bureau’s website. The differences between these datasets are way more complicated than can be summarized in this blog post. Or you could just download all of the datasets & then pick between then later on.

Negative: Very Limited Data Points are available with the SDDS

So we’ve seen that the SDDS is an awesome and easy to use tool for pulling school district population data–kudos to the builders of this tool–but there’s a potential problem. What if you need data points for school districts that aren’t included in the list above? Here are just a few examples of data points that folks working with school district data want & can’t get from the SDDS (even though these data points below are published by the Census Bureau):

• median household income
• median home value
• household type & family type
• poverty data
• unemployment rates
• industry & occupation data
• transportation statistics
• educational attainment
• languages spoken at home

Enter a 2nd NCES data tool.

Tool 2: NCES’  Demographic Profile Webmap

• Good for more detailed demographic data & heat maps
• Bad for getting the most current demographic data available

The Demographic Profile Webmap is another awesome tool that works like the SDDS tool with the added benefit of providing us with even more demographic data. Oh boy! The only problem is that as of today the Demographic Profile Webmap provides us the ACS 2006 -2010 data. The most current ACS 5 year dataset that is available today is 2007-2011. Hopefully, this app will be updated in the near future. But as of now, you can’t get the most current data from this tool.

Technique 3: GIS Analysis

Let’s say you want the most current data available for the number of people who speak Spanish in a school district. Here’s a very simplified outline of how to do this analysis (warning: if you don’t have a basic understanding of GIS, the following outline might be gibberish).

How to Estimate Data Points for School Districts that are not included in SDDS

1. Download the school spatial boundaries. If you want school district boundaries, you can download them either from the SDDS (see above) or the Census Bureau’s website. If you want individual school boundaries, try the SDDS, but for the most current boundary data, you’ll probably need to contact either the school district or the state education agency.
2. Download the Census boundaries for your area of interest (i.e. state, county, etc.). You can download these boundaries from the Census Bureau’s website. If you aren’t sure what geographic level to use, Census block groups are the smallest level of Census geography with almost all data points. But in the case of language data, we can get much more detailed language data for Census tracts than for block groups. So you may have to do some guess & check to determine the best Census geography to use.
3. Use the school spatial boundaries as “cookie cutters” (aka calculate intersections) to cut out the Census boundaries that are contained within the school boundaries. Let’s pretend that Census Tract A, B & C are all contained within your school district boundary.
4. Then pull the Census language data for Census Tracts A, B & C from the Census Bureau’s website. Now if you are a GIS guru, you could do this step along with step 2 & combine the demographic data with the spatial data. But if you are like me and are more comfortable working with databases than with GIS (GISs? I’m not sure how to pluralize GIS. Or is it like sheep – and GIS is the pural?), you can pull the demographic data at this point & write a quick query to grab only the data points that you need.
5. Finally, sum the counts of Spanish speakers for Census Tract A, B, & C to produce your estimate of Spanish speakers in a school district. So if Tract A has 10 Spanish speakers, Tract B has 5 Spanish speakers & Tract C has 0 Spanish speakers, we can estimate that the school district has 15 Spanish speakers. Summation works well for basic count data, but if you need to estimate “median” values, you’ll probably need to use something like a weighted average calculation.

Hopefully, this post has armed you with enough tools & techniques that you can now pull the school district population data that you need. But if you’d rather not learn the difference between Decennial Census vs ACS data or when it’s better use use Census blocks vs block groups vs tracts, you can always hire me to pull this data for you.

Photo by Blake Wheeler on Unsplash

Occasionally we get asked for ZIP Code-based Census data. We have to say that we don’t have ZIP Code data and neither does the Census. What!? That’s right! But don’t worry, we got your back – the Census publishes ZIP Code Tabulation Areas (ZCTA, pronounced ‘ziktah’).

To quote the Census:

“ZCTAs are generalized area representations of U.S. Postal Service (USPS) ZIP Code service areas. They represent the most frequently occurring five-digit ZIP Code found in a given area. Simply put, each ZCTA is built by aggregating 2010 Census blocks, whose addresses use a given ZIP Code. Each resulting ZCTA is then assigned the most frequently occurring ZIP Code as its ZCTA code.”

In a nutshell: ZCTA 78704 is a representative snapshot of ZIP Code 78704.

ZCTAs Rock! Here’s why:

ZCTAs map to US Census data; ZIP Codes don’t.

About ZIP Codes. We don’t have ZIP code boundaries, because the U.S. Postal Service doesn’t define or publish them. Also, some of ZIP Codes are just routes (a series of connected lines, not polygons).

ZIP Codes change constantly to help the USPS deliver mail more efficiently. That’s what ZIP Codes are designed for: delivery. If you’re planning delivery logistics, use ZIP codes. Some companies offer zip code boundary maps. They estimate (using a variety of methods) polygons based on groups of points, physical landmarks (like bodies of water), and other methods. Most companies have their own private/favorite method of estimating ZIP code boundaries. But because ZIP Codes change so often, boundary maps fall out of sync with US Census demographic data.

US Census Demographic Data + ZCTAs == AWESOME

ZCTAs are polygons fixed to match the census data. By using ZCTAs you get a consistent and accurate representation of a geographic area.

More info about ZCTAs from Census.gov

• http://www.census.gov/geo/ZCTA/zcta.html
• http://www.census.gov/geo/ZCTA/zctafaq.html

Photo by Beatriz Pérez Moya on Unsplash

So this blog post probably won’t be among the most captivating posts I’ve ever written. But it might help someone who is googling around for “how to cite Census data.”

UPDATE: this post was revised on May 8, 2013.

The citations below are in Modern Language Association format — or MLA. And the format of the citation that you’re going to use will depend on HOW you pull the Census data. For example if you use the Census Bureau’s American Fact Finder tool, you’ll use 1 format whereas if you use pull the data from the Census Bureau’s FTP site, you’ll use a different format.

How to Cite American Community Survey Data

1. Cite American Community Survey data pulled via the American Fact Finder (most popular)

United States Census Bureau / American FactFinder. “B11001 : Household Type (Including Living Alone).” 2007 – 2011 American Community Survey. U.S. Census Bureau’s American Community Survey Office, 2011. Web. 1 January 2013 <http://factfinder2.census.gov>.

Note: the date is the date that you pulled the data from the American Fact Finder.

2. Cite American Community Survey data pulled via the Census’ electronic files on the ftp site

United States Census Bureau.“Summary File.”2007 – 2011 American Community Survey. U.S. Census Bureau’s American Community Survey Office, 2013. Web. 1 January 2013 <http://ftp2.census.gov/>.

Note: the date is the date that you download the data from the ftp site.

How to Cite Decennial 2010 Census data

 1. Cite Census 2010 data pulled via the American Fact Finder (most popular)

United States Census Bureau / American FactFinder. “P12 : Sex by Age.” 2010 Census.U.S. Census Bureau, 2010.Web. 1 January 2013 <http://factfinder2.census.gov>.

Note: the date is the date that you pulled the data from the American Fact Finder.

 2. Cite Census 2010 data pulled via the Census’ electronic files on the ftp site

United States Census Bureau. 2010 Census.U.S. Census Bureau. 2010. Web. 1 January 2013 <http://www.census.gov/2010census/data/>.

Note: the date is the date that you download the data from the ftp site.

How to Cite Decennial 2000 Census data

1. Cite Census 2000 data pulled via the American Fact Finder

United States Census Bureau / American FactFinder.“Profile of General Demographic Characteristics: 2000 Census 2000 Summary File 1 (SF 1) 100% Data.” U.S. Census Bureau, 2000.Web. 1 January 2013 <http://factfinder.census.gov>.

Note: the date is the date that you pulled the data from the American Fact Finder.

Double Note: you can’t actually pull Census 2000 data from the the American Fact Finder 2. You used to be able to (or “used to could” for any Texans reading this post) get this data from the old American Fact Finder 1. So I’ll just leave this example here as a hypothetical example.  

 2. Cite Census 2000 data pulled via the Census’ electronic files on the ftp site

United States Census Bureau. 2000 Census.U.S. Census Bureau, 2000.Web. 1 January 2013 <http://www.census.gov/prod/cen2000/>.

Note: the date is the date that you download the data from the ftp site.

Photo by Alex Bertha on Unsplash

I recently got a question from a Cubit user about where to get parcel data. I’ve shared our conversation in hopes that it might answer other peoples’ questions about where to find parcel and real estate data.

The Question: Where Do I Find Parcel Data?

“I’m working on my final project for my masters program & am excited to use Cubit for all the demographic data I’ll need. In the meantime, I need to gather real estate data on recent sales of residential & commercial property in Los Angeles county. The data also needs to include characteristics such as parcel size, structure size, & use. I’ve heard of the MetroScan database, but don’t know how to access it without paying thousands of dollars to CoreLogic. Do you have any ideas on other ways of accessing data like this?”

My Reply — Including the 3 Best Sources for Free Parcel Data

Funny you should ask–we’re actually working on developing a Cubit-like tool for parcel data. But it will include Travis County, Texas parcels only at first. If folks like it, hopefully, we can expand to the rest of the US!

Back to your question, here’s a quick brain dump of free or cheap parcel data tools/sources.

• LoopNet:the standard for commercial property data
• Zillow’s Tool Recent Home Sales
• “parcel size, structure size, and use” That’s where things get tricky. You may know this already, but whenever I need parcel data, I start by contacting 3 government agencies:
  1. the Appraisal District or Tax Assessor,
  2. the Metropolitan Planning Organization or Council of Government and
  3. the City GIS office.

  I’d then use address info provided by Zillow and LoopNet to pull the parcel data from the government agency database.

• Also, in Texas we have a real estate data center as part of Texas A&M that rocks! I wonder if California has a similar resource.

A quick google search turned up that you can buy parcel data from here: http://assessor.lacounty.gov/extranet/Outsidesales/saleintr.aspx
Similar data cost me $85 for all of Travis County, Texas as a reference point. Maybe they’ll give you a student discount? Or offer to share the results of your research with them?
That’s all I can think of. Let me know how it goes!

Are there other good sources of free/cheap parcel data that aren’t included in this blog post? If so, please share in the comments.

[UPDATE: WE NO LONGER OFFER FREE TRIALS FOR THE WEB APP]

You have 2 options for figuring out which Census tracts are in your city. Important Note: What you and me and other normal people would call cities or towns, the Census Bureau calls places.

Option 1. Census Tracts by City for Complex Cities (i.e. New York City, Los Angeles, Chicago)

• • Go to the Census Bureau’s website here: http://www.census.gov/cgi-bin/geo/shapefiles2010/main
  • Download the 2010 Places File for your State & unzip the file.
  • Download the 2010 Census Tract File for the county that your city is in & unzip the file.
  • Open the files in your favorite GIS. My current favorite GIS is the free, light-weight MapWindow.
  • Use the GIS to calculate which Census Tracts intersect your city. I wrote down step-by-step instructions for how to do this via Map Window. You can download these instructions here.

These instructions are a great example of what really happens when you use open source GIS to solve a problem, because I ran into an error when I was using MapWindow. So I included the error & 2 different ways to solve the problem in the instructions.

Also, you can follow the instructions above to figure out block groups by city as well as Census blocks by city.

Option 2. Census Tracts by City for Simple Cities (i.e. Worchestor, MA, Shoshoni, WY)

• Sign up for a 7 day free trial of Cubit.
• In Cubit, type the name of the city into the Find Box to zoom to your city (see the Orange Circle).
• Zoom out slightly until you can see the entire City in the interactive map. The city outline will be a slightly different color than the rest of the map (see the Orange Arrow-the darker shade is the city limits).
• Click on the Polygon tool and draw a polygon slightly inside of the city limits (see the Orange box). Select Tract as the Geography Level if you want Census tract data (see the Purple box). Click Save.

On the next page, you’ll get a labeled map that shows the Census tracts in your city as well as demographic data reports.

Note: if you save your project and Cubit returns a Census tract that you can see is outside of your city, don’t panic. All that means is that some part of your line touched an area that is outside of the city. Click on the Edit Project button. Then, click on the Polygon button. Now, click on the points on your original line to adjust your line inside of the city so that you’ll only get tracts inside of your city. If nothing happens when you click on your original line, try clicking on different points along the line.

Again, you can follow the instructions above to figure out block groups by city as well as Census blocks by city.

The Really Nerdy Background

This blog post came about when I received a request to pull all Census tracts for a city. “No problem,” I thought, “I’ll just go to the Census 2010 data and look up the Places code for Census tracts.”

So I pulled up the Census 2010 data in MySQL and displayed all of the Census Tracts for my state of interest, Massachusetts. Great! There 1478 Census Tracts in Massachusetts. Then, I selected all Census Tracts in Massachusetts where the Place FIPS was not empty. Oh no! That returned 0 rows! That means there are no relationships in the Census 2010 data between Census tracts & Place.

Just to make sure I wasn’t missing something, I double checked the Census 2010 Redistricting PDF Summary Level Sequence Chart, and sure enough, there were no SumLevels with relationships from Cities to Census Tracts (or to BlockGroups or to Blocks for that matter). But don’t take my word for it–click on the link above in this paragraph and go to Summary Level Sequence Chart on page 4-1 to see for yourself.

So you’ll have to use either of the 2 methods above to figure out which Census tracts by City instead using relationships in the raw Census data. Was this post about Census tracts by City helpful or confusing? Any questions here?