search

Open-NC / Are-We-Represented

Main repository for the Are We Represented? project
4 stars 1 forks source link

Prepare data on voting-age demographics by county #5

Open ejaxon opened 6 years ago

ejaxon commented 6 years ago

Obtain census data on demographics of the voting-age population of NC by county. Deliverables for this task include the data, documentation on how it was obtained, ideally in the form of a Python or R script that does it.

ChrisTheDBA commented 6 years ago

https://github.com/ajdamico

wandernat commented 6 years ago

What kind of demographics are desired? The American FactFinder provides data on eligible voters aged 18 and over but only breaks it down by male/female. They also provide separate data on race, also by county, but those demos are not broken down by voter eligibility. (https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_16_5YR_CP05&prodType=table)

cassiodorus commented 6 years ago

From my POV at this time what would be of value is some way to estimate the number of non-citizens* by county. This does not need to be precise but of course defensible and not distortive. What I would like to do is create a kind of funnel that starts with Census Bureau total population and works its way down to NCSBE voters. Along the way, it seems to me, we need to pare down population counts in accordance with the portion who can vote. Not so easy since we will be tracking race (and perhaps gender), so the non-electors would have to be spread against the total and its categories. The whole issue of Hispanic/Latino categorization is a conundrum since race is mixed in with that ... somehow.

Also, just for interest, https://www.brookings.edu/blog/the-avenue/2017/05/18/census-shows-pervasive-decline-in-2016-minority-voter-turnout/

*Also, those disqualified due to felonies etc. but this might be small???

David

From: Nat notifications@github.com Sent: Saturday, April 21, 2018 8:10:32 PM To: Open-NC/Are-We-Represented Cc: Subscribed Subject: Re: [Open-NC/Are-We-Represented] Prepare data on voting-age demographics by county (#5)

What kind of demographics are desired? The American FactFinder provides data on eligible voters aged 18 and over but only breaks it down by male/female. They also provide separate data on race and age by county, but those demos are not broken down by voter eligibility. https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_16_5YR_CP05&prodType=table

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/Open-NC/Are-We-Represented/issues/5#issuecomment-383342850, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ACI2ObKVGFEV0bxxgEG_XJDL9hqAS483ks5tq8p4gaJpZM4SL0uk.

ejaxon commented 6 years ago

@wandernat take a look at what @cassiodorus has prepared here.

A few random thoughts on this:

  1. My guess is that non-citizen and felon counts will be relatively small overall (the last especially), but may play a more significant role in particular places and for particular races.
  2. Re: hispanic/latino, what about breaking out hispanic-white and hispanic-other as separate "races". There may be others, but basic idea is to look at ethnicity as a subcategory under race.
  3. I like the idea of a funnel, but some of those numbers are going to be relatively hard numbers and some are going to involve some assumptions and modeling. Besides being careful about our assumptions and modeling, we should probably think about presentation issues around this.
ChrisTheDBA commented 6 years ago

Race and ethnicity are separate fields. I took a small county's data and cross-tabbed it...

It shows that Hispanic/Latino spans all racial categories. Wouldn't it would require an extensive geo-microsimulation to statistically model this for all counties?

Chris Mathews

On Sun, Apr 22, 2018 at 12:11 PM, Eric Jackson notifications@github.com wrote:

@wandernat https://github.com/wandernat take a look at what @cassiodorus https://github.com/cassiodorus has prepared here https://github.com/Open-NC/Are-We-Represented/tree/master/AWR_data_munging .

A few random thoughts on this:

  1. My guess is that non-citizen and felon counts will be relatively small overall (the last especially), but may play a more significant role in particular places and for particular races.
  2. Re: hispanic/latino, what about breaking out hispanic-white and hispanic-other as separate "races". There may be others, but basic idea is to look at ethnicity as a subcategory under race.
  3. I like the idea of a funnel, but some of those numbers are going to be relatively hard numbers and some are going to involve some assumptions and modeling. Besides being careful about our assumptions and modeling, we should probably think about presentation issues around this.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/Open-NC/Are-We-Represented/issues/5#issuecomment-383392607, or mute the thread https://github.com/notifications/unsubscribe-auth/AAOGtUEhU3hb6DSj6m77vLVUwrNCoBVjks5trKu6gaJpZM4SL0uk .

cassiodorus commented 6 years ago

Using Eric's random thoughts ...

[Eric:] My guess is that non-citizen and felon counts will be relatively small overall (the last especially), but may play a more significant role in particular places and for particular races. [David]: Probably, but we should quantify this upfront. I suggest something the order of 1% is negligible, but 10% is significant. We are not demographers (although it would be nice if we could recruit one as a consultant) and as far as I know such data is not readily available, if it exists at all. I think the overall NC non-citizens may be between 5% and 10%, but are certainly concentrated and non-uniform by county. I would feel a lot better about our work if we had at least some defensible estimates by county. I have no idea at all how to obtain data about non-elector felons. Some overall numbers would be of help.

[Eric:] Re: hispanic/latino, what about breaking out hispanic-white and hispanic-other as separate "races". There may be others, but basic idea is to look at ethnicity as a subcategory under race. [David:] My feeling is that we cannot do that and maintain consistency since it would constitute ambiguous double-counting. Census Bureau recognizes this as demonstrated in their data files and methodology papers (I can provide files and links if anyone is interested).

[Eric:] I like the idea of a funnel, but some of those numbers are going to be relatively hard numbers and some are going to involve some assumptions and modeling. Besides being careful about our assumptions and modeling, we should probably think about presentation issues around this. [David:] Yes. I am preparing some Rmarkdown scripts that address some of these issues. Hopefully available for discussion this coming week, but I am a slow worker. As for presentation, I think there are some real opportunities for a UI person to be creative - as soon as possible.

David Hopp

From: Eric Jackson notifications@github.com Sent: Sunday, April 22, 2018 12:11:38 PM To: Open-NC/Are-We-Represented Cc: David Hopp; Mention Subject: Re: [Open-NC/Are-We-Represented] Prepare data on voting-age demographics by county (#5)

@wandernathttps://github.com/wandernat take a look at what @cassiodorushttps://github.com/cassiodorus has prepared herehttps://github.com/Open-NC/Are-We-Represented/tree/master/AWR_data_munging.

A few random thoughts on this:

  1. My guess is that non-citizen and felon counts will be relatively small overall (the last especially), but may play a more significant role in particular places and for particular races.
  2. Re: hispanic/latino, what about breaking out hispanic-white and hispanic-other as separate "races". There may be others, but basic idea is to look at ethnicity as a subcategory under race.
  3. I like the idea of a funnel, but some of those numbers are going to be relatively hard numbers and some are going to involve some assumptions and modeling. Besides being careful about our assumptions and modeling, we should probably think about presentation issues around this.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/Open-NC/Are-We-Represented/issues/5#issuecomment-383392607, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ACI2Oeb68wKAx1nSIy9iMlispmE7Z9whks5trKu6gaJpZM4SL0uk.

ejaxon commented 6 years ago

So does anyone have a suggestion about how to handle hispanic representation? Whether they fit into the funnel model or not, I do think it's critical to address the question in some way. It was actually the precipitating question for the entire project.

ejaxon commented 6 years ago

One other issue here, of course, is how to make whatever we do repeatable. If we do some deep investigation to answer a question (whether felons or non-citizens or whatever), but there's not a good way to update the information for a future election, we have a problem.

LauraMB12 commented 6 years ago

I think there will have to be a section comparing race and a separate section comparing ethnicity.

Laura M. Biediger Community Engagement Coordinator Neighborhood Improvement Services, City of Durham 807 E. Main Street Durham, NC 27701

P 919-560-1647 X34259 F 919-560-1790

Laura.Biediger@Durhamnc.govmailto:Laura.Biediger@DurhamNC.gov www.DurhamNC.govhttp://www.durhamnc.gov/

Visit our team’s new website at DurhamCommunityEngagement.org.

From: Eric Jackson [mailto:notifications@github.com] Sent: Sunday, April 22, 2018 3:35 PM To: Open-NC/Are-We-Represented Cc: Subscribed Subject: Re: [Open-NC/Are-We-Represented] Prepare data on voting-age demographics by county (#5)

So does anyone have a suggestion about how to handle hispanic representation? Whether they fit into the funnel model or not, I do think it's critical to address the question in some way. It was actually the precipitating question for the entire project.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/Open-NC/Are-We-Represented/issues/5#issuecomment-383406530, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AQg-mlHFRyVKZObyDj_NnODHlls41p29ks5trNtSgaJpZM4SL0uk.

cassiodorus commented 6 years ago

I am working on something re this issue. Getting near ready for display. Hope to talk w/Laura Tuesday at CfD.

David Hopp

From: LauraMB12 notifications@github.com Sent: Tuesday, April 24, 2018 12:11:33 PM To: Open-NC/Are-We-Represented Cc: David Hopp; Mention Subject: Re: [Open-NC/Are-We-Represented] Prepare data on voting-age demographics by county (#5)

I think there will have to be a section comparing race and a separate section comparing ethnicity.

Laura M. Biediger Community Engagement Coordinator Neighborhood Improvement Services, City of Durham 807 E. Main Street Durham, NC 27701

P 919-560-1647 X34259 F 919-560-1790

Laura.Biediger@Durhamnc.govmailto:Laura.Biediger@DurhamNC.gov www.DurhamNC.govhttp://www.durhamnc.gov/

Visit our team’s new website at DurhamCommunityEngagement.org.

From: Eric Jackson [mailto:notifications@github.com] Sent: Sunday, April 22, 2018 3:35 PM To: Open-NC/Are-We-Represented Cc: Subscribed Subject: Re: [Open-NC/Are-We-Represented] Prepare data on voting-age demographics by county (#5)

So does anyone have a suggestion about how to handle hispanic representation? Whether they fit into the funnel model or not, I do think it's critical to address the question in some way. It was actually the precipitating question for the entire project.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/Open-NC/Are-We-Represented/issues/5#issuecomment-383406530, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AQg-mlHFRyVKZObyDj_NnODHlls41p29ks5trNtSgaJpZM4SL0uk.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/Open-NC/Are-We-Represented/issues/5#issuecomment-383990535, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ACI2OQWNsumlS6bP7_e2-Jc2iAZM9afJks5tr061gaJpZM4SL0uk.

cassiodorus commented 6 years ago

Well, ethnicity, it seems, is a problem. Census Bureau data presents reasonably reliable data but NCSBE voting records do not. Looking at the 2016 general election, there were about 4,575,000 voters, of whom some 3,643,000 identified themselves as non-Hispanic, about 92,000 as Hispanic, and (here's the rub) 887,000 chose not to respond to the question. That's nine times as many "UN" as "HL". Not a good basis for any kind of analysis.

It seems to me that proceeding with any analysis that depends on ethnicity would best be done with a demographer or quantitative sociologist involved, preferably from a university or research institute.

If you have any thoughts or questions about the data, let me know. If you want to see the data, I can provide it via our S3 instance, but the files are very large - too large for spreadsheet programs. There may be some way to get data digest from NCSBE to check my claim.

David Hopp

From: David Hopp cassiodorus@hotmail.com Sent: Wednesday, April 25, 2018 3:03:51 PM To: Open-NC/Are-We-Represented Subject: Re: [Open-NC/Are-We-Represented] Prepare data on voting-age demographics by county (#5)

I am working on something re this issue. Getting near ready for display. Hope to talk w/Laura Tuesday at CfD.

David Hopp

From: LauraMB12 notifications@github.com Sent: Tuesday, April 24, 2018 12:11:33 PM To: Open-NC/Are-We-Represented Cc: David Hopp; Mention Subject: Re: [Open-NC/Are-We-Represented] Prepare data on voting-age demographics by county (#5)

I think there will have to be a section comparing race and a separate section comparing ethnicity.

Laura M. Biediger Community Engagement Coordinator Neighborhood Improvement Services, City of Durham 807 E. Main Street Durham, NC 27701

P 919-560-1647 X34259 F 919-560-1790

Laura.Biediger@Durhamnc.govmailto:Laura.Biediger@DurhamNC.gov www.DurhamNC.govhttp://www.durhamnc.gov/

Visit our team’s new website at DurhamCommunityEngagement.org.

From: Eric Jackson [mailto:notifications@github.com] Sent: Sunday, April 22, 2018 3:35 PM To: Open-NC/Are-We-Represented Cc: Subscribed Subject: Re: [Open-NC/Are-We-Represented] Prepare data on voting-age demographics by county (#5)

So does anyone have a suggestion about how to handle hispanic representation? Whether they fit into the funnel model or not, I do think it's critical to address the question in some way. It was actually the precipitating question for the entire project.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/Open-NC/Are-We-Represented/issues/5#issuecomment-383406530, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AQg-mlHFRyVKZObyDj_NnODHlls41p29ks5trNtSgaJpZM4SL0uk.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/Open-NC/Are-We-Represented/issues/5#issuecomment-383990535, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ACI2OQWNsumlS6bP7_e2-Jc2iAZM9afJks5tr061gaJpZM4SL0uk.

ejaxon commented 6 years ago

Agree that it doesn't serve the project to do something half-baked. I think we have to address it, but we can perhaps have a page/section on ethnicity where we give general population stats from census and then outline the issues with trying to carry that through the voter registration or elected body portions.