Proximity between homes and schools
April 25, 2011 10:50 AM   Subscribe

X% of people live within Y miles of a public school. Where can I find datapoints for X and Y?

One of my employer's projects is to provide a public service by placing the access to that service at schools, on the premise that most people live close to a school. I'd like to quantify that.

I'm primarily interested in California but would happily take data for other states, or the US as a whole.
posted by bac to Grab Bag (12 answers total) 1 user marked this as a favorite
This is the type of data that would be difficult to find without doing your own analysis.

The approach that immediately comes to mind is a GIS solution. You can get census data for census blocks (or tracts if it's a fairly dense area so you can work with less data), make a population density map and then sum the population that falls within your specified area. You could then find percentage based of the county. The location of the schools would probably be information you could get from the counties or states of interest, it'd be the most difficult to obtain.

If this is something critical hiring out wouldn't be a bad idea.
posted by EsotericAlgorithm at 11:10 AM on April 25, 2011 [2 favorites]

I do this type of analysis quite a bit using GIS. Frequently school data is published for states (I'm currently using ESRI Business Analyst Data), and US Census population can help you estimate how many people are in different areas. The last census was 2000, the most recent was 2010, but the results are not published yet (afaik).
posted by a womble is an active kind of sloth at 11:16 AM on April 25, 2011

I agree with EsotericAlgorithm - you're going to need GIS to do this.

You can download a table with names and addresses of all of the public schools in a county here. Based on a cursory search of the site, I didn't see any ability to download that information for an entire state, but I'd give that a look yourself.
posted by desjardins at 11:21 AM on April 25, 2011

If you can't find this sort of thing online, I'd find a local library near you with a GIS/Geoinformatics section. A bunch of larger public libraries offer this now, as do many of the larger research universities. They'd be able to help you run the queries you need (possibly including offering an intro class to learn the software) and you could get specific data for the counties or neighborhoods you're interested in working in.

If you really just want a quick general stat to throw out, I bet the census bureau has some data on this, but I'm not sure where to find it offhand. Try searching for information on the average distance students (sometimes termed pupils in various reports) travel to school. That's not quite the same thing as the distances between homes and schools (due to siblings, childless families, etc...), but it would be something for you.
posted by zachlipton at 11:30 AM on April 25, 2011

Response by poster: Thanks. Yeah, I could clearly do the analysis from scratch; I was hoping, though, that somebody might have already done it and published it. Maybe for real estate purposes, or allocating funding for school buses, or the like.
posted by bac at 11:31 AM on April 25, 2011

If you're ok with data from one or a couple of school districts instead of state or nationwide, I'd work the phone and call around to the transportation departments of various school districts. I bet a lot have this kind of data on hand as part of their planning processes.
posted by zachlipton at 11:36 AM on April 25, 2011

Well, since you're into a rough approximation, you might as well assume straight-line distance, draw some circles of radius y around schools on a map, and use known population density from the Wikipedia.
posted by klausman at 11:39 AM on April 25, 2011

You might be able to find this data through the National Center for Education Statistics. I've spent very little time digging in the data they have, but the School District Mapping and Demographics tool may have what you need. Or, of course, you can call/email a staff member and ask! They clearly have some GIS folks in-house, though, so I bet they have the data available in some way.
posted by Kpele at 12:12 PM on April 25, 2011 [1 favorite]

I'll stop messing with this now, but I found some data that gets toward what you're looking at. Note that this is the distance students travel to their school, which is a different question from the distance between homes and the closest school, but it will give you an upper bound for families with school-age children. I bet we could get a lot more detailed data of this sort from the National Household Travel Survey, but their data site seems to be down right now.
posted by zachlipton at 12:16 PM on April 25, 2011 [1 favorite]

I routinely work with county and ZCTA level population data. I know nothing about school district data, other than districts are drawn without regard to county and ZCTA boundaries, generally following bus routes and not census workers and/or mail trucks.

What this means is - the district map winds up being different from the population map, and to intersect the two is one of those processes that generally leads me to pull out my hair out. (same hair, different set of map files)

Working with data below the district-level - actually working with population density of the district and coordinates of all the schools within that district to calculate frequency range buckets - that would sort-of be a fun exercise.
posted by Nanukthedog at 12:35 PM on April 25, 2011

You could probably come up with a statistic using data from the The 2005/ 2006 American Housing Survey (link is to PDF). For instance, on page 130 it lists the number of sampled households with any children from ages 0-13. It also lists the number of sampled households with kids < 13 who live less than a mile from a public elementary school.
posted by oceano at 1:21 PM on April 25, 2011

If you do end up doing this analysis yourself, then it is a classic application of something like the Maximal Covering Model of the group of Location-Allocation Models. ArcGIS implements these models for both network and raster data. It used to be a very processor-intensive kind of analysis but I imagine those demands or costs have lessened a lot.
posted by Rumple at 8:20 PM on April 25, 2011

« Older Tracking Devices: How do they bleepin' work?   |   Baby shower play list Newer »
This thread is closed to new comments.