The exit poll data do break it down by income and by race, but separately. So they just have "low income" vs "high income". And "white" vs "black".

I am looking for data that break it down by both income and race. So something like "low-income whites", "low-income blacks", "high-income whites", "high-income blacks".

Where can I find such data?

I am particularly interested in the 2016 Presidential Election, but I'm also looking for data for previous elections.


The American National Election Study is the definitive poll for American elections. You will have to register to download any of the data, but you will find all you want and considerably more. The 2016 data is not yet available.

In the time series data, annual income (as a number, not a category) can be found in the column INCPO_TOTINC ("Total Income Amount").

Race is self-reported in 1 of 5 categories. See the item DEM_RACECPS for those details.

Note that this is a data file you will have access to, not synthesized results. You will need to use Excel or the statistical software of your choice to actually analyze the data. If you want to generalize about the US population you should consult their helpful document "How to Analyze ANES Survey Data". The most important takeaway is that since they used a weighted sampling design, you will need to weight responses appropriately.

  • This is great, thanks. But just to point out two limitations: (1) Data for quite a number of variables are restricted (e.g. INCPO_TOTINC is restricted). Though fortunately there are some INCGROUP variables that are not restricted. (2) The sample size is also somewhat smaller (5,916) than the CNN exit poll (over 20,000) . – user2212 Dec 17 '16 at 1:59
  • I wouldn't consider that sample size to be a serious limitation, especially when contrasted to an exit poll. Exit polls rely on terrible methodologies which require very high sample sizes to get anywhere; ANES uses a much more sophisticated sample methodology and doesn't require the same sample size. – indigochild Dec 17 '16 at 2:10
  • I see. I wasn't aware that there was a difference in methodologies. But now that you mention it, I really should be skeptical of any research done by CNN (et al.). – user2212 Dec 17 '16 at 10:05

You must log in to answer this question.