There are many Federal sources of income data, and they do not all measure the same thing, or even the same workers. The most widely used BLS data includes all forms of employee compensation for employees covered by unemployment insurance (about 80-85% of the workforce). This data tallies both wages/salaries and benefits, such as health insurance, retirement contributions, unemployment insurance, etc. This data set does not include proprietors (owners of businesses), people earning their income from commissions (e.g., real estate agents, car sales people, etc.), student interns, or elected officials.
BEA data includes both employees and proprietors and their income, by summing the BLS data for employees, with IRS data for proprietors. People earning their income only from commissions are lumped in with proprietors. Student interns and elected officials are classed as employees rather than proprietors. The BEA data will have different numbers for total employment for the same geographic area, compared to BLS.
ACS asks the respondent how much cash income they earned in the last twelve months, so it does not include the value of benefits (which most respondents would likely not know and report very reliably anyway). Up until the 2000 Census, this question was asked for the previous calendar year, and since the survey was done mostly in April, most respondents probably had just filed their income tax return, and had a pretty good idea of their cash income for the previous year. Starting with the 2005 ACS, the surveys have been conducted every month. The question has been what was your income for the previous 12 month period, which for most included parts of two tax years, the more recent of which was still on-going. For this reason, many users of the ACS income data are suspicious of comparing the ACS data with the old decennial Census data (to look for trends) because of these differences in methodology.
Bottom line, comparing income data from the various sources is comparing apples and oranges. Since ACS asks households only about cash income, while BLS asks employers about wages plus benefits, you can count on there being a substantial difference, with no underreporting implied. Make comparisons across geographies using the same dataset, and you should have reasonably useful results.
Pete Swensson
1201 Terrace Lane SW
Olympia, WA 98502
(360) 352-7909 home
(360) 628-6621 cell
pete@swensson.us
From: ctpp-news [mailto:ctpp-news-bounces@chrispy.net] On Behalf Of Sam Granato
Sent: Monday, December 04, 2017 3:39 AM
To: ctpp-news@chrispy.net
Subject: Re: [CTPP] The Washington Post: Scientists can now figure out detailed, accurate neighborhood demographics using Google Street View photos
Cars parked on the street don't always reflect who lives there, and the same can be said for what gets reported to the Census. (Take a look at what the BLS publishes to get a sense of how much, in aggregate, the populace underreports income in the ACS.). Maybe the best lesson from this study is that aggregating as high as zip code level can cancel out a lot of errors.
Sent from my iPhone
On Dec 1, 2017, at 8:26 PM, Pete Swensson <pete@swensson.us> wrote:
Just an anecdotal observation: Vehicle make and model preferences vary a lot by geography for reasons other than income or demographics. Lots more pick-ups in some parts of the country than others. Subarus ALL have 4-wheel drive, and they are consequently a very popular make in my county. Japanese makes of cars (e.g., Honda) have long been more popular on the West Coast (regardless of ethnicity) than in the Midwest, where there is stronger brand loyalty to American makes manufactured in the Midwest (in theory – many are now made in Mexico).
Even if one accepts the underlying premise of correlating vehicle types to demographics and voting patterns, I think Ed’s suggestion of processing vehicle registration files instead of Google photos is right on target. Much easier. And it wouldn’t surprise me if users of Big Data for political analysis are already doing this.
Pete Swensson
1201 Terrace Lane SW
Olympia, WA 98502
(360) 352-7909 home
(360) 628-6621 cell
From: ctpp-news [mailto:ctpp-news-bounces@chrispy.net] On Behalf Of Steve Wilson
Sent: Friday, December 01, 2017 9:04 AM
To: ctpp-news@chrispy.net
Subject: Re: [CTPP] The Washington Post: Scientists can now figure out detailed, accurate neighborhood demographics using Google Street View photos
Streetview + ACS can also be used to develop a model estimating likelihood of encountering man on porch yelling “hey you kids – get off my lawn”. :)
From: Polzin, Steven [mailto:polzin@cutr.usf.edu]
Sent: Friday, December 01, 2017 10:58 AM
To: ctpp-news@chrispy.net
Subject: Re: [CTPP] The Washington Post: Scientists can now figure out detailed, accurate neighborhood demographics using Google Street View photos
Be careful who you let parked in front of your house. It might change your income.
From: ctpp-news [mailto:ctpp-news-bounces@chrispy.net] On Behalf Of Scott Ramming
Sent: Friday, December 01, 2017 11:55 AM
To: ctpp-news@chrispy.net
Subject: Re: [CTPP] The Washington Post: Scientists can now figure out detailed, accurate neighborhood demographics using Google Street View photos
As I understand it, their models aren’t based on auto ownership levels, but the type and age of the autos parked in view.
There’s a big difference in neighborhoods where you find concentrations of say 2018 Range Rovers versus 1964 Ford Mavericks.
Plus one to Thomas’s point about high density areas where people either don’t own vehicles or park the vehicles they do in garages. The same lack of sample would also apply to where HOAs dictate that all vehicles must be parked in personal garages with the door lowered. Perhaps lack of variation in the paint color scheme could be used as an explanatory variable for such neighborhoods.
Scott Ramming, PhD PE | Senior Travel Modeler | Transportation Planning & Operations
Direct: 303-480-6711 | Fax: 303-480-6790 | Email: sramming@drcog.org
1290 Broadway • Suite 100 • Denver, Colorado 80203-5606
main: 303-455-1000 • email: drcog@drcog.org • web: www.drcog.org
<image003.png> <image004.png> <image005.png>
From: ctpp-news [mailto:ctpp-news-bounces@chrispy.net] On Behalf Of Alan E. Pisarski
Sent: Friday, December 01, 2017 8:59 AM
To: ctpp-news@chrispy.net; 'Ed Christopher' <edc@berwyned.com>
Subject: Re: [CTPP] The Washington Post: Scientists can now figure out detailed, accurate neighborhood demographics using Google Street View photos
Re auto ownership: it was at one time a pretty good proxie for wealth etc. but vehicles are so ubiquitous now that it is far less reliable. Not far from here in a neighborhood that is largely low/mid income housing the streets are lined at night with white vans that the immigrant painters, home construction guys use. Great area to steal a ladder. Alan
Alan E. Pisarski
703-941-4257 landline
703 650-8925 cell
From: ctpp-news [mailto:ctpp-news-bounces@chrispy.net] On Behalf Of Krishnan Viswanathan
Sent: Friday, December 01, 2017 10:15 AM
To: Ed Christopher <edc@berwyned.com>
Cc: ctpp-news@chrispy.net
Subject: Re: [CTPP] The Washington Post: Scientists can now figure out detailed, accurate neighborhood demographics using Google Street View photos
Totally agree with you Ed. I think the Washington Post headline is a dose of hyperbole. I have a question about spurious correlations and as far as I know the ACS does not ask what type of vehicle is there in the household so this paragraph in their paper gave me pause:
Using ACS and presidential election voting data for regions in our training set, we train a logistic regression model to estimate race and education levels and a ridge regression model to estimate income and voter preferences on the basis of the collection of vehicles seen in a region. This simple linear model is sufficient to identify positive and negative associations between the presence of specific vehicles (such as Hondas) and particular demographics (i.e., the percentage of Asians) or voter preferences (i.e., Democrat).
On Fri, Dec 1, 2017 at 9:38 AM, Ed Christopher <edc@berwyned.com> wrote:
Interesting stuff Krishnan--
If their basic assumption were true, that vehicle ownership somehow translates into demographics (ie voter behavior), then why not just cut Google out and process vehicle registration files. It seems to me that would be a lot easier and cheaper. Then again you have to buy the basic assumption. Also, if you look at precinct by precinct voter behavior two things will surprise you. First, precincts are not all one color (red or blue) in most places and the number of people who do vote are very small when considering the total population. While I found this work interesting I would not be out their trying to oversell what its capabilities are without a whole lot more work and research. As I see it we have a very long way to go before we have something that is a kin to the ACS and all its by-products.
On 11/30/2017 8:39 PM, Krishnan Viswanathan wrote:
This will interest people in this group and also foster discussion about the methods used. The article itself has a link to the paper.
Scientists can now figure out detailed, accurate neighborhood demographics using Google Street View photos
http://wapo.st/2AnuP9LKrishnan Viswanathan
5628 Burnside Circle
Tallahassee FL 32312
_______________________________________________ctpp-news mailing listctpp-news@chrispy.nethttps://www.chrispy.net/mailman/listinfo/ctpp-news
--Ed ChristopherTransportation Planning Consultant708-269-5237
--
Krishnan Viswanathan
5628 Burnside Circle
Tallahassee FL 32312
_______________________________________________
ctpp-news mailing list
ctpp-news@chrispy.net
https://www.chrispy.net/mailman/listinfo/ctpp-news