ACS Data Users Group

Hi team: long-time viewer, first time poster. I've been scratching my head about this one all day - I'm simply looking to find a table (or some method) that would allow me to calculate the median household income for a given geography (preferably counties) by vehicle ownership. I've spent at least an hour running through different ACS tables, with no luck. Is this something I'll have to use PUMS for instead? Any guidance would be greatly appreciated.

You need to use a PUMS file The vehicle question (household section of the ACS survey form) is #12 How many automobiles, vans, and trucks of one-ton capacity or less are kept at home for use by members of this household? Split the PUMS data by using None v one or more. Then use the household income variable to compute median income. If you want to leverage the PUMS results use the household income distribution split by vehicle "ownership" add in covariates that might affect vehicle ownership. Use the resulting multi-way cross tabulation with small area estimation to get other geographies.

David - pardon the delay, but thank you for clarifying. Was finally able to find the time to get the cross tabulation going.

Mo--

If you don't mind using old ACS data, from 5+ years ago?, then try the CTPP database.
Yes, CTPP is a thing that still exists. http://data5.ctpp.transportation.org/ctpp1216/Browse/browsetables.aspx And it has many crosstabs that include number of vehicles possessed by household.

Possibly the crosstabs you want are

A112214 - Vehicles available (6) by Household income in the past 12 months (2016 $) (9) (Households) - Large Geos Only

or

A112306 - Household size (5) by Vehicles available (5) by Household income in the past 12 months (2016 $) (9) (Households)

Good luck.

Apologies. Hit send too fast.

more CTPP tables:

B112205 - Mean Household income in the past 12 months (2016$) (1) by Vehicles available (6) (Households) - Large Geos Only or

B112207 - Median Household income in the past 12 months (2016$) (1) by Vehicles available (6) (Households) - Large Geos Only

That's probably what you want.

Todd - CTPP did come to mind (glad I wasn't off in considering it). However, my prime issue with it is how old the dataset is. Was looking for something more recent. Nonetheless, thank you for the guidance.

Mo:

The answer is to use the PUMS. The best way to do this (IMHO) is by using the IPUMS USA (https://www.ipums.org/) at the University of Minnesota. It's free to register. This is the fastest/best/amazing way to access PUMS data (IPUMS is pronounced eh-pums).

I did a tabulation on median household income by vehicle availability level for the NATION using the 2019 ACS. It took a few seconds to run. Honestly.

Go to www.ipums.org

Go to the IPUMS USA sub-site https://usa.ipums.org/usa/ register for free, of course!!

Click the link to "Online Tool for Analysis" (This is the Survey Data Analysis package developed at UC Berkeley)

Click on "Use data from a single sample" - I chose the "2019 ACS"

Click on the "Means" tab on the right side of the SDA analysis page.

Fill in the forms:

DEPENDENT: HHINCOME (The variable that I want means and medians!)

row variable: vehicles(r:9 "No Vehicles"; 1 "One Vehicle"; 2 "Two Vehicles"; 3-8 "Three of More Vehicles") - this is my recode for the VEHICLES variable.

selection filter(s): pernum(1) (this restricts the analysis to the 1st person in each household, the householder)

weight: hhwgt - household weight

Click ON the checkboxes for the "median/percentile" options

Run it. Wait a few seconds for a new browser tab to appear!

To limit your analysis to your state, region, county, I would either use counties in the column variable, or state/counties in the "selection filter" box. Here's an example of filtering by counties for the SF Bay Area.

for column variable: countyfip(r: 1 "Alameda"; 13 "Contra Costa"; 41 "Marin"; 55 "Napa"; 75 "San Francisco"; 81 "San Mateo"; 85 "Santa Clara"; 95 "Solano"; 97 "Sonoma")

and restrict it to California households: selection filter(s): pernum(1),statefip(6)

Wish I could post screenshots!

Results for the USA:

0 vehicle households: $22,000 median HH income; $46,213.81 mean household income, 10,575,965 total households.

1 vehicle households: $40,100 median; $58,690.40 mean; 39,803,758 households

2 vehicle households: $80,000 median; $106,236.20 mean; 45,268,831 households

3+ vehicle households: $104,000 median; $132,347.06 mean; 27,154,350 households

TOTAL households: $62,930 median HH income; $91,429.77 mean household income; 122,802,904 total households.

Hope I didn't screw it up.

Chuck

Chuck - I can assure you that you didn't screw this up. I appreciate the detailed response.

I did play around with the PUMS, though I ended up downloading the files first. It had been a while since I touched the PUMS, so I was feeling hesitant. My feelings aside, your response saved me an enormous chunk of time. When I downloaded the files, I was unsure how to deal with the weights, while using the online cross tabulation made that more seamless.

Thank you.

Just as an aside there are several Census webinars about using the PUMS files. They go over how to compute with weights. I think that this might be the webinar that I watched https://www.youtube.com/watch?v=Mtymad91pbE It is pretty good.

Yes, the replicate weights are a chore to deal with, though Kyle Walker's tidycensus program is the most straightforward approach, if you've downloaded your state's PUMS. That's why I strongly prefer the IPUMS USA survey data analysis system. The standard errors are automatically produced!

IPUMS USA is amazing, but doesn't have every solution to every PUMS problem. I had a discussion with IPUMS staff, and they agreed that what I wanted couldn't be done in the SDA system, namely:

households by workers in the household (0, 1, 2, 3+) by vehicles available in the household (0, 1, 2, 3+). The "households by workers in household" is essentially a "tally" variable, which can't be computed in the SDA. (It can be done in R, SAS, etc.)