Pop density by census tract

14 replies to this topic

#1 benbakelaar Posted 25 August 2006 - 01:07 PM

benbakelaar

Ultimate Contributor

• 658 posts
• Gender:Male
• Location:North Brunswick, NJ
• Interests:maps, information, technology, scripting, computers
• United States

Can someone describe to me how I would go about calculating population density per census tract? I know I will need land area per census tract, which doesn't seem to be available in the Census data (they have it for county level, I know). So, would I be able to use some tool/script in ArcMap to calculate the land area contained in each census tract? And then use that number, divide with the total population, to get population density for each census tract?

Does this even make sense? What I want is a population density map of New York City at a more detailed level than county.

#2 Matthew Hampton Posted 25 August 2006 - 01:30 PM

Matthew Hampton

Hall of Fame

• Moderator
• 1,281 posts
• Gender:Male
• Location:Portland, Oregon
• Interests:Playing in the mountains and rivers.
• United States

You got it right. Just calc the area for all the census tracts and devide with the population. You could use x-tools or the following procedure:

Click Editor on the Editor toolbar and click Start Editing.

You can make calculations without being in an editing session; however, in that case, there is no way to undo the results.

Right-click the shapefile layer you want to edit and click Open Attribute Table.

Right-click the field heading for area and click Calculate Values.

If there is no field for area values, you can add a new field for area by clicking the Options button and selecting Add Field. However, to add a new field, you need to exit the editing session.

Type the following VBA statement in the first text box:

Dim dblArea as double
Dim pArea as IArea
Set pArea = [shape]
dblArea = pArea.area

Type the variable dblArea in the text box directly under the area field name.

Click OK.

Tip
The property area returns a field type of double. For best results, your area field should also be a double field type.

Oregon Metro - Portland, OR
www.oregonmetro.gov

#3 Dennis McClendon Posted 25 August 2006 - 01:31 PM

Dennis McClendon

Hall of Fame

• Validated Member
• 1,052 posts
• Gender:Male
• Location:Chicago
• Interests:map design, large-scale maps of cities
• United States

Land_area and water_area are available for census tracts, at least when you query census.gov

Can I suggest, though, that what you really want to do is a dot density map by census tracts. Not only is this much more understandable than a quantile choropleth map, but you don't need to have the land area for tracts.
Dennis McClendon, Chicago CartoGraphics
chicagocarto.com

#4 benbakelaar Posted 25 August 2006 - 01:45 PM

benbakelaar

Ultimate Contributor

• 658 posts
• Gender:Male
• Location:North Brunswick, NJ
• Interests:maps, information, technology, scripting, computers
• United States

cartomap, thanks for mapping out the "how to", appreciated!

dennis, I agree with you from a design/cartographer perspective (pretty much all of that perspective has come from hanging around here! ), however we are using the chloropleth's as underlays/base maps, which we are displaying other point locations over. It has to do with the client, and the needs of their audiences.

#5 Rob Posted 25 August 2006 - 02:09 PM

Rob

Legendary Contributor

• Moderator
• 418 posts
• Gender:Male
• Location:Kailua, Hawaii
• Interests:anything outside.
• United States

check out this site

on data normalization

will give a good background on how to approach these data/map issues, and why it is important to do so.

#6 benbakelaar Posted 25 August 2006 - 02:37 PM

benbakelaar

Ultimate Contributor

• 658 posts
• Gender:Male
• Location:North Brunswick, NJ
• Interests:maps, information, technology, scripting, computers
• United States

thanks rob, that's a great resource as well! i will read through all of it over the weekend. despite having received little in the way of cartographic training in my geography undergrad classes, i am conceptually aware of the weighting/normalizing issue, but not necessarily how it applies to an arbitrary request. in the case of population density, it is obvious, but i'm sure there are much more nuanced examples where the person making the map must be able to make an informed decision.

#7 Martin Gamache Posted 26 August 2006 - 12:31 PM

Martin Gamache

Ultimate Contributor

• 980 posts
• Gender:Male
• Location:Washington DC
• Interests:History of Topographic Cartography
Topographic Mapping
History of Relief Depiction
Thematic Cartography
Demographic Cartography
Cartographic techniques, methods, and tools
Orienteering
Panoramic &amp; Kite Photography
• United States

Ben,

You can also normalise on the fly by a field containing the pplygon area in the choropleth map dialogue in arcmap. There is an option to normalise with any field in your dataset.

#8 benbakelaar Posted 26 August 2006 - 04:09 PM

benbakelaar

Ultimate Contributor

• 658 posts
• Gender:Male
• Location:North Brunswick, NJ
• Interests:maps, information, technology, scripting, computers
• United States

Type the following VBA statement in the first text box:

Dim dblArea as double
Dim pArea as IArea
Set pArea = [shape]
dblArea = pArea.area

Type the variable dblArea in the text box directly under the area field name.

I tried this but ran into some errors. Also found this thread http://forums.esri.c...t=102008#286912 but couldn't make that code work either.

I ended up using XTools -> Table Operations -> Calculate Area, and setting the units to miles. I am using martin's tip to set the "normalization" field in Symbology. However, I'm wondering if the results are accurate.

Without normalization, population per census tract - classed:
0-2309
2310-4453
4454-6926
6927-12780
12781-24523

With normalization by XTools calculated area field, in miles: (square?)
0-18772
18772-43384
43384-74549
74549-111886
111886-188991

Can pop density really be 100,000+ per square mile, even in NYC?

#9 Rob Posted 26 August 2006 - 04:42 PM

Rob

Legendary Contributor

• Moderator
• 418 posts
• Gender:Male
• Location:Kailua, Hawaii
• Interests:anything outside.
• United States

the field you are running the area calc script provided by cartomat needs to be formatted as a double precision number field, else this script won't work. and it will return a value using whatever unit the projection uses, such as meters or feet squared.

your numbers do seem a little out of whack. I'd double check the areas and make sure they seem realistic first by doing a quick XY measure on screen.

#10 benbakelaar Posted 28 August 2006 - 09:30 AM

benbakelaar

Ultimate Contributor

• 658 posts
• Gender:Male
• Location:North Brunswick, NJ
• Interests:maps, information, technology, scripting, computers
• United States

the field you are running the area calc script provided by cartomat needs to be formatted as a double precision number field, else this script won't work.  and it will return a value using whatever unit the projection uses, such as meters or feet squared.

your numbers do seem a little out of whack.  I'd double check the areas and make sure they seem realistic first by doing a quick XY measure on screen.

<{POST_SNAPBACK}>

I did format as a double precision, still ran into some errors, but I ended up using XTools which did the trick. I double-checked my legend numbers, they actually are correct. Imagine a census tract less than a mile square, with 25000+ people in it... the numbers quickly jump to 100,000+ per square mile, even though that's not exactly accurate. However, in terms of density, it is, and since the purpose is for a visual comparison of density, not explaining the actual population, I think this is ok then?

#11 danielle Posted 28 August 2006 - 10:19 AM

danielle

Key Contributor

• Validated Member
• 52 posts
• Location:NYC
• United States

In NYC when you go from census tracts to census blocks, the numbers get even crazier. There are a few with people per square mile more than 1 million.

These are tiny polygons that are just the corners of blocks or slivers between streets. Eventually I will do some research on this, but right now my guess is that it is some sort of statistical adjustment.

-Danielle

#12 Martin Gamache Posted 28 August 2006 - 10:35 AM

Martin Gamache

Ultimate Contributor

• 980 posts
• Gender:Male
• Location:Washington DC
• Interests:History of Topographic Cartography
Topographic Mapping
History of Relief Depiction
Thematic Cartography
Demographic Cartography
Cartographic techniques, methods, and tools
Orienteering
Panoramic &amp; Kite Photography
• United States

Here is some work I did on this problem in Boston.

Population density is a tricky subject and is entirely relative to what you consider the living area. If you use the census definition you get one number, if you use the actual area where people live ( square footage of residential property) you get another. But how do you define the area where people live, if you think about it it includes everything around us that gives our living space it's feel, whether it is how wide the streets are to how high the buildings around are and how many parks and open spaces or how much of a view of the sky we have. So what do you include , gross living space square footage of building, assessing parcel area, block area, do you substract the sidewalks and streets and parks??

IMO there is not really a correct answer but it makes sense to make clear what you use and to apply it consistently.

Dasymetric

#13 danielle Posted 28 August 2006 - 11:00 AM

danielle

Key Contributor

• Validated Member
• 52 posts
• Location:NYC
• United States

Martin, that sounds like something out of my thesis proposal. I'm going to be looking at various kinds of "density" and see how they all add up. People, buildings, traffic, etc. How much "stuff" is there in space, and how does it feel. Can I cite your paper, or have you published anything else on the subject?

-Danielle

#14 benbakelaar Posted 28 August 2006 - 08:00 PM

benbakelaar

Ultimate Contributor

• 658 posts
• Gender:Male
• Location:North Brunswick, NJ
• Interests:maps, information, technology, scripting, computers
• United States

Sorry to take this thread away from the original post, or even the last two....

Talking about the "accuracy" of population density really brings up the larger issue, the "accuracy" of any numeric data or specifically in our case, geographic data. The way my client is using this population density chloropleth is as a reference, to get a general "sense" of the land, so in that sense its accuracy is less of a concern than if it was being used by policy makers to determine distribution of federal funding, for instance. But then again, you never know who will use your map or how it will be used in the end

Realistically, societies have always struggled with maps and mapping agendas, and now that maps are beginning to proliferate due to technology, it will become a higher priority (we hope?) for the global community to discuss these issues.

So, mandatory spatial education for all?

#15 Martin Gamache Posted 28 August 2006 - 08:30 PM

Martin Gamache

Ultimate Contributor

• 980 posts
• Gender:Male
• Location:Washington DC
• Interests:History of Topographic Cartography
Topographic Mapping
History of Relief Depiction
Thematic Cartography
Demographic Cartography
Cartographic techniques, methods, and tools
Orienteering
Panoramic &amp; Kite Photography
• United States

Ben,

Of course if the Census data and enumeration units work for you or the task you have been asked to do, no need to concern yourself with anything more. But if the numbers and results don't "sound" right there may be a reason for that.

mg

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->