Trails Month: Trail Events, Trail Mileage, and Residential Access

June is National Trails Month, and in honor of that (and National Trails Day, coming up on Saturday, June 3), we’re taking a look at trails in the Champaign County area: what the mileage and residential access are, what the local trail plans and jurisdictions are, and what local events are coming up on National Trails Day.

What is the trail mileage in our area? What is the residential access to trails? What additional data would be useful?

We’ve assessed both trail mileage and trail access at three levels of geography: all of Champaign County, the five largest municipalities in Champaign County (Champaign, Urbana, Rantoul, Savoy, and Mahomet), and Champaign-Urbana alone.

First, we’ll look at trail mileage.

Trail Mileage in Champaign County

AreaLength (Miles)% of Total Trail Mileage
Champaign County85.22100.00%
Champaign, Urbana, Rantoul, Savoy, Mahomet84.7299.41%
Champaign-Urbana67.6279.35%

Source: CUUATS Data Portal; Champaign County GIS Consortium.

As one would expect, the greatest trail mileage is in the largest area: all of Champaign County. But the difference between the trail mileage of the whole county and the trail mileage of the five largest municipalities within the county is very small, only half a mile. And over three-quarters of the county’s total trail mileage is within Champaign-Urbana.

Next, we’ll check out residential access.

Trails Access in Champaign County Areas

AreaTotal Residential Parcels Residential Parcels within 1/2-Mile Buffer Residential Parcels within 1/4-Mile Buffer 
Parcels%Parcels%Parcels%
Champaign County52,810100.00%32,61261.75%20,75639.30%
Champaign, Urbana, Rantoul, Savoy, Mahomet35,776100.00%29,89483.56%19,29853.94%
Champaign-Urbana28,280100.00%24,14585.38%15,65855.37%

Source: CUUATS Data Portal; Champaign County GIS Consortium.

We calculated trail access in the same way we calculate our Accessibility indicators (Access to Bicycle Infrastructure, Access to Parks, and Access to Transit). Based on parcel land use, we generated a total number of residential parcels within our three areas of interest. We then performed a geographic analysis to determine how many of the total parcels in each area fall within a half mile and a quarter mile of a trail, and presented that as a percentage.

This method does have a few shortfalls. For the sake of simplicity, the distance of any given parcel from a trail is measured as linear distance, rather than the walking distance that it would take for residents of that parcel to reach the trail. This method greatly increases the ease of analysis, but may not be an accurate representation of actual walking distance from some parcels.

Methodological disadvantages aside, the residential access data tells a similar story to the trail mileage data. Residential access to trails is greatest in Champaign-Urbana for both half-mile and quarter-mile distances. The difference between residential access in Champaign-Urbana and residential access in the five largest municipalities is relatively small, only a couple of percentage points. But residential access in all of Champaign County, which includes all of the strong residential access in the five largest municipalities, trails (pun absolutely intended) residential access of the other two areas by over 20 percentage points for the half-mile distance and almost 15 percentage points for the quarter-mile distance.

These percentages may change when the first phase of the Kickapoo Rail Trail opens, which is anticipated to happen late this summer. The almost seven miles of trail between Urbana and St. Joseph will add to the trail mileage and to the residential access statistics.

But there’s more to trails data than mileage and access. Data about trail usage, like bicycle and pedestrian counts, can be helpful in applying for grant funding and planning additional outreach. A local average cost per mile for trail construction would be another useful figure for trail planning. However, there are challenges associated with both of these data needs. Traffic counts, whether bicycle, pedestrian, or vehicle, are time-consuming when collected manually, and construction costs can differ based on circumstantial and site-based factors. So while these pieces of data would be useful to have, they are not immediately available.

What are the existing trails jurisdictions? What existing trails plans are there in our area?

Trails can be held by a variety of jurisdictions. Many are held by the Champaign Park District, Urbana Park District, and Champaign County Forest Preserve District. Others are held by municipalities: the City of Champaign, City of Urbana, Village of Mahomet, Village of Rantoul, and Village of Savoy. Several, all physically located within the Village of Mahomet, are held by IDOT. Some are institutionally held, by the University of Illinois or the Urbana School District, while others are private.

There are a number of existing trails plans in the Champaign County area: the Champaign Park District Trails Master Plan, the Urbana Park District Trails Master Plan, the Champaign County Greenways & Trails Plan, the Champaign Trails Plan, and the Savoy Bike & Pedestrian Plan. Most of these plans include not only the current state of trails in that jurisdiction, but also a vision for future expansions and improvements.

What are the different ways the public can interact with trails? What trails events are coming up in the Champaign-Urbana area?

The Urbana Park District will hold a National Trails Day event in Busey Woods. This Saturday, June 3rd, starting at 9 a.m., volunteers will be able to both enjoy and improve Busey Woods. Tasks and activities will include hiking, tree planting, and trash and invasive species removal. The National Hiking Society website also lists more information for this event.

Looking for a trail near you so you can get out and about for National Trails Month? Trail locations and information are available on the CUUATS Data Portal. So whether you’re trying to find an off-street bike route or planning a recreational outing, you can find the route that best suits your needs.

Bike Month: Celebrating and Educating

May is here, bringing warm weather, the end of the 2016-17 academic year, and National Bike Month. To celebrate Bike Month, in this post, we’ll be talking about the prevalence of bicycling in Champaign County, the pros and cons of the data available, and the BikeMoves Illinois app. Finally, we’ll present some upcoming events happening during Bike Month here in Champaign-Urbana.

To assess how prevalent bicycling is in Champaign County and selected Champaign County municipalities, a good place to start is commuter mode share. (“Mode share” is generally defined as the percentage of travelers that opt for a given mode of transportation.)

 Bicycle Mode Share (%), Workers 16 and Over 
PlaceEstimateMOE
Champaign County2.60.3
City of Champaign2.70.5
City of Urbana6.21.1
Village of Mahomet0.00.7
Village of Rantoul1.31.0
Village of St. Joseph0.50.8
Village of Savoy0.40.6

Source: U.S. Census Bureau; American Community Survey, 2011-2015 American Community Survey 5-Year Estimates, Table S0801; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (27 March 2017).

The data in the table above is from the U.S. Census Bureau’s 2011-2015 American Community Survey (ACS) 5-Year Estimates. You can find data from the ACS 1-Year Estimates datasets at the county level discussed as part of our Commuter Mode Share indicator, and at the municipal and township level under the Transportation section of our Regional Dashboard. For additional data, the League of American Bicyclists website has more resources on commuting by bike.

We can see that a relatively small percentage of workers commute by bicycle in the Champaign County area (under 10% in all areas, and under 5% in most), but that there is also some variance from municipality to municipality. The Village of Mahomet is estimated to have no bicycle commuters, while the City of Urbana’s estimated bicycle commuter mode share, the highest in the county, is 6.2%.

The primary drawback of commuter mode share data is just that: that it measures commuter mode share. While it’s sound data for what it measures, it does not offer a complete picture of transportation behaviors in a given area. Commuter mode share data does not capture trips that are not work-related. It provides no insight on how residents of a given area run errands or make recreational trips, or how students get to school. Although it is helpful for the topic it covers, it is by no means comprehensive.

CCRPC/CUUATS is working on other resources to provide a more complete data set, including the BikeMoves Illinois app.

What’s the BikeMoves Illinois app, and where can I find it?

Released in August 2016, BikeMoves Illinois is a mobile app for Android and iOS that allows cyclists to record and share their routes with local planners. It also provides cyclists with information such as the locations of bicycle facilities and bike parking. BikeMoves Illinois was developed by the Champaign County Regional Planning Commission in partnership with the Illinois Department of Transportation and has been used to record and share more than 500 bicycle trips. Download the app from the App Store and Google Play Store, and look for a new version with improved mapping and automatic recording this summer!

What Bike Month events are coming up in the Champaign-Urbana area?

There is a full calendar of events going on in our area throughout the month of May to celebrate, and educate about, bicycling. A run-down of some events is below; even more events, additional details, and registration information can be found on the C-U Bike Month 2017 website, event calendar, and Facebook page.

  • Tuesday, May 2: Bike to Work Day
    Registration is still open for Champaign County Bike to Work Day, and welcome stations will be located around the community to offer free breakfast to participating cyclists.
  • Wednesday, May 10: C-U Bike to School Day
    Registration is open through this form on the C-U Bike Month 2017 website.
  • Wednesday, May 10, 6 p.m.-dusk: Community Full Moon Night Ride
    The Community Full Moon Night Ride, to be held at Meadowbrook Park in the Windsor Road parking lot, includes activities for all ages, from live music, bike check-ups, and a raffle to guided bike rides and the optional night ride to Sidney Dairy Barn.
  • Saturday, May 20, 9 a.m.-12 p.m.: Playing It Safe Bike Rodeo
    This family safety fair, with opportunities for kids to learn and practice biking skills, will be held at Leonhard Recreation Center (2307 Sangamon Drive, Champaign).
  • Saturday, May 20, 2 p.m.-dusk: C-U Food Cruise
    Participants can join in any of seven guided tours through the Champaign, Urbana, and Savoy area, with stops at a variety of local eateries. All tours begin at Hessel Park and end at Jarling’s Custard Cup (309 W. Kirby Avenue, Champaign). Registration is required.
  • Monday, May 22, 5-7 p.m.: Spring Fling Bike Rodeo
    This community resource fair, another chance for kids to learn and practice biking skills, will be held at Garden Hills Park in Champaign.

So whether bicycling is your commuting mode of choice, a healthy way to run errands, or just a way to have fun, keep an eye out for opportunities to celebrate Bike Month here in Champaign County.

 

Data Literacy: Testing for Statistically Significant Difference Between Estimates

It’s time for another data literacy blog post, and this month we’ll be discussing how to work with the concept of statistical significance. But to be clear from the outset, what we’ll be talking about is understanding statistical significance as a data user, rather than as a statistician: interpreting published figures, not calculating or publishing your own figures.

For starters, let’s establish what we mean by “statistical significance” for the purposes of this discussion. What we’ll be talking about is determining whether the difference between two estimates is statistically significant. This means, basically, whether the two estimates can be viewed as really different, or if the difference between them may just be due to some chance or statistical reason[1].

The U.S. Census Bureau offers free training presentations on a variety of topics, and one of them, “Using ACS Estimates and Margins of Error,” introduces the following formula as a basic way to test whether two estimates are statistically different from each other[2]:

|Est1 – Est2|
(√(MOEest12+MOEest22))

It’s much easier than it looks. Don’t worry – we’ll break it down and provide examples. But to make a long story short, if the number produced by this formula is greater than one, the difference between the two estimates is significant. If it’s less than one, it’s not[3].

To demonstrate how this formula works, we’re going to walk through two related examples: the estimated percentage of the population over 25 years of age with a bachelor’s degree in the years 2014 and 2015, and the estimated percentage of the population over 25 years of age with a graduate or professional degree in the years 2014 and 2015. (In both cases, these estimated percentages are from Champaign County. All the estimates we’re about to discuss are from the U.S. Census Bureau’s American Community Survey 1-Year Estimates datasets. Something else to keep in mind is that all ACS estimates are at a 90% confidence interval – meaning that it is 90% certain that the actual value lies within the range denoted by the margin of error.[4])

Example 1

 % of Population Aged 25 Years or Older with a Bachelor's Degree Statistical Significance Test Result
YearEstimateMOE
201421.8%1.9
201522.0%2.10.07062

|Est1 – Est2|
(√(MOEest12+MOEest22))

|21.8 – 22.0|
(√(1.92+2.12))

|-0.2|
(√(3.61+4.41))

_0.2_
(√(8.02))

_0.2_
2.83196

= 0.07062

As we see here, the resulting figure is less than one, so the two estimates are not statistically different from one another. What this means is that, even though the two years’ estimates appear different, the difference is small enough and the margins of error are large enough that we can’t be sure that this apparent increase reflects an actual increase.

This really becomes relevant when we’re trying to describe trends: we can’t describe the years 2014-2015 as part of an increasing trend in the percentage of Champaign County residents aged 25 and over with bachelor’s degrees. All we can say is that, between 2014 and 2015, the percentage of the given population with a bachelor’s degree remained roughly the same.

Let’s take a look at a related statistic: the percentage of Champaign County residents aged 25 and over with a graduate or professional degree.

Example 2

 % of Population Aged 25 Years and Over with a Graduate or Professional Degree Statistical Significance Test Result
YearEstimateMOE
201420.5%2.0
201523.9%2.01.20216

|Est1 – Est2|
(√(MOEest12+MOEest22))

|20.5 – 23.9|
(√(2.02+2.02))

|-3.4|
(√(4.0+4.0))

_3.4_
(√(8.0))

_3.4_
2.82843

= 1.20216

Unlike the result from the first example, 1.20216 is greater than one: this means that the difference between the two years’ estimates is statistically different. In other words, the increase from 20.5% to 23.9% isn’t due to statistical error or chance; it reflects an actual, albeit small, increase in the real percentage of the given population with a graduate or professional degree in Champaign County. If the differences between these and other previous or subsequent years are also significant, we can start discussing concrete trends.

With what types of data is this formula useful?

As we said earlier, this formula should only be used to test the statistical significance of the difference between estimates. Exact counts are exact counts – they don’t have margins of error, and you don’t need to test whether the pile of seven hats you counted today is really greater than the pile of five hats you counted yesterday. The question just isn’t relevant. (For more information about the difference between estimates and counts, check out our August data literacy post, about margins of error, and our December data literacy post, about the U.S. Census Bureau’s American Community Survey.)

Obviously, you’ll need margins of error to work with in order to carry out these calculations. Ideally, these will be available with the estimates in question.

What about the shortcut method?

You may also be aware of a bit of a shortcut: using the given margin of error for each estimate to calculate a possible range for that estimate, then comparing the ranges of two estimates to see if they overlap, with the assumption that estimates with overlapping ranges are not statistically different from each other, and estimates with non-overlapping ranges are.

We’re sorry to have to tell you this, but it doesn’t work like that. Let’s go through those two examples again with this method.

If the 2014 percentage of the population aged 25 and older with a bachelor’s degree in Champaign County is 21.8 +/- 1.9, we have a range of 19.9-23.7. Similarly, the same estimate from 2015, 22.0 +/- 2.1, has a range of 19.9-24.1. Clearly, these ranges overlap, and the difference between the estimates is not statistically significant.

If the 2014 percentage of the population aged 25 and older with a graduate or professional degree in Champaign County is 20.5 +/- 2.0, we have a range of 18.5-22.5. The same estimate from 2015, 23.9 +/- 2.0, has a range of 21.9-25.9. These ranges also overlap – but as we demonstrated above, the difference between the estimates is statistically significant.

What we can learn from this is that the assumption that all estimates with overlapping ranges are not statistically different from each other is pretty flawed.

The shortcut method does have its uses: if you’re asked a question about this in person, or if you otherwise really do not have time to do the full calculation, it can allow you to provide a guess. If the ranges don’t overlap, you can say that the estimates are probably, but not definitely, statistically different. If the ranges do overlap, the best thing to say is that it’s possible that the estimates are not statistically different – but that it’s also possible that they are. Be upfront about the fact that all you can do is hazard a guess. Even better, offer to follow up with the person who asked the question later, and get back to them with a solid answer after you have time to run the calculation. It’s better to give a correct answer a day later than a wrong answer immediately.

What does all of this mean about the data visualizations elsewhere on the site?

If you’ve spent time looking at different pages around this site, especially in the Indicators section, you’ll notice that many of the pages include charts and graphs plotting the year-to-year data. Some of these year-to-year data points are not significantly different from each other. In those cases, graphs may look like they show an upward or downward trend, but with a quick test of the data, it’s clear that, like in Example 1 above, we can’t actually assign that trend to those years.

Does that mean that these graphs are meaningless? No – no more so than the tables of data associated with them. Each data point is the estimate for its period of time. That much is solid. The only thing we can’t get concrete about is exactly what’s going on between one data point and the next.

Every graph on this website has an associated table of data with it, and all estimates include their MOEs, so you can run the formula for yourself and check the statistical difference between any years you’re interested in. And where we can’t assign a year-to-year trend, we’ve usually noted this in the analysis.

 

Thanks for sticking with us through 1,500 words of statistics! We hope this post cleared things up a little, and gave you one more skill for being an informed, savvy data user.

 

 

[1] U.S. Census Bureau, Sirius Fuller. (April 6, 2016). “Using ACS Estimates and Margins of Error.” <https://www.census.gov/content/dam/Census/programs-surveys/acs/guidance/training-presentations/2016_MOE_Slides_01.pdf>. (Retrieved 14 March 2017).

[2] Ibid.

[3] Ibid.

[4] U.S. Census Bureau; Census Glossary; “Confidence Interval (American Community Survey”; <https://www.census.gov/glossary/#term_ConfidenceintervalAmericanCommunitySurvey>; (Retrieved 23 March 2017).

Table sources: U.S. Census Bureau; American Community Survey, 2015 American Community Survey 1-Year Estimates, Table S1501; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (19 September 2016). U.S. Census Bureau; American Community Survey; 2014 American Community Survey 1-Year Estimates, Table S1501; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (16 March 2016).

Library Patronage and Community Services: The Quantifiable and Non-Quantifiable Benefits of Public Libraries

March is National Reading Month, so we’re taking this opportunity to talk about public libraries, measuring library patronage, and the community, recreational, and informational benefits of public libraries in general.

Champaign County is home to 10 library districts: the Champaign Public Library, the Homer Community Library, the Mahomet Public Library District, the Ogden Rose Public Library, the Philo Public Library District, the Rantoul Public Library, the St. Joseph Township-Swearingen Memorial Library, the Sidney Community Library, the Tolono Public Library District, and the Urbana Free Library. For each one, we’ve calculated the library patronage percentage: the percent of residents in the library’s area that have an unexpired library card.

Library Patronage

  2015 2016 
LibraryPopulationNumber of Unexpired Resident Users Cards% of Population with Library CardsNumber of Unexpired Resident Users Cards% of Population with Library Cards
Champaign Public Library81,05531,55538.9%29,13035.9%
Homer Community Library1,19359549.9%53845.1%
Mahomet Public Library District12,6236,08948.2%7,00455.5%
Ogden Rose Public Library81017822.0%17021.0%
Philo Public Library District1,95463932.7%58029.7%
Rantoul Public Library12,9414,33833.5%3,75029.0%
St. Joseph Township-Swearingen Memorial Library5,8761,87631.9%1,56526.6%
Sidney Community Library1,23336029.2%35729.0%
Tolono Public Library District11,6753,79732.5%3,50530.0%
Urbana Free Library41,25012,59230.5%12,45430.2%

Source: Illinois State Library, Office of the Illinois Secretary of State

You can see that there is a wide range of resident populations among the library districts, but the patronage rates tend to be more consistent. The total range (roughly 20%-50%) is not small, but six of the 10 libraries have patronage rates clustered within a few percentage points of 30%. More analysis on this data can be found on our Library Patronage page – it’s one of the indicators in our Engagement category.

You may notice that there’s only one year of data in the table above, and that’s 2015. This is because CCRPC was not looking at library patronage before the revision of our indicators set in summer 2016 (and you can read more about that process in our April 2016 and July 2016 blog posts). We’re planning to update the Library Patronage page with newly-released 2016 data and additional analysis later this spring.

One item to note is that, for the libraries with outlying patronage rates, there does not seem to be any correlation between the number of residents in the area and the percentage with an unexpired library card. The library with the lowest participation rate, Ogden Rose Public Library (22.0%) also has the smallest number of residents in its area (810). However, the library with the highest participation rate, Homer Community Library (49.9%), has the second-smallest number of area residents (1,193). The Mahomet Public Library District comes in right behind the Homer Community Library with a patronage rate of 48.2%, but has a much larger population (12,623).

This data comes from the Illinois Public Library Annual Report (IPLAR), a resource that is compiled by the Illinois State Library and comprised of self-reported data from libraries throughout the state. Some, but not all, of this data is available online here, and the full data set is requestable.

Library patronage data doesn’t give us a complete picture of the role of libraries in our communities. To try to create a fully quantified measure of a library, we would need to account for their other services. The prospect of rolling library patronage together with things like event attendance, sessions on public computers, and resources lent (among others), while accounting for the size of the library, its funding, its staffing, and its other resources, is a daunting one. And even if that were feasible, it would still miss residents that come in to pick up a tax form, ask a question at the information desk, page through a book or a newspaper, or use a library computer as a guest – all services that may not require the resident to be a cardholder. But just because something can’t be accurately quantified does not mean it’s unimportant. (Obviously.) Libraries in Champaign County and elsewhere provide their communities with services that may not be available anywhere else in town. They act as community centers, by holding events and by providing public space for informal gathering. They offer free Internet and computer access that patrons may not have at home. And that’s in addition to their information and lending services, which offer enormous educational, informational, and recreational benefits.

What library patronage data can tell us is something about community engagement as a whole, which is why we use it as one of our Engagement indicators. As we’ve pointed out, libraries act as both formal and informal community centers. We posit that residents that are engaged at and invested in their local library may be more engaged and invested, socially, economically, and politically, in the community as a whole. This is not the result of a long study, but a shorthand assumption: involvement at the library implies involvement in the community.

If you’re part of the percentage of residents with an unexpired library card, celebrate Reading Month and keep enjoying your library! If you’re not, this could be the perfect time of year to go check out a library in your area – you’ll probably find something interesting.

Source: Illinois State Library, Office of the Illinois Secretary of State

The Statistics of Valentine’s Day: Measuring Relationships with Census Data

It’s February. With Valentine’s Day on the horizon, we’re matching up a sentimental holiday with statistical rationality to address the following question: can we determine, by numbers or percentages, an accurate, basic relationship status breakdown in Champaign County?

The short answer is yes and no. There is data on marital status and household type available from the U.S. Census Bureau’s American Community Survey, which gets us some of the information we’re looking for. However, the categories that this data is broken down into are highly detailed in some respects, but not at all detailed in others, making it difficult to arrive at a concrete answer.

Marital Status – Categories and Limitations

The American Community Survey includes questions on respondents’ marital status, and offers five response options: “Now married (except separated),” “Widowed,” “Divorced,” “Separated,” and “Never married.” For the purpose of our discussion, we’ll refer to the first four categories as the “marriage and post-marriage” categories.

 

Champaign CountyNow married (except separated) Widowed Divorced Separated Never married 
EstimateMOEEstimateMOEEstimateMOEEstimateMOEEstimateMOE
Population 15 years and over39.3%0.94.1%0.38.7%0.51.5%0.246.4%0.7

U.S. Census Bureau; American Community Survey, 2011-2015 American Community Survey 5-Year Estimates, Table S1201; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (7 February 2017).

 

This table offers a very detailed and useful breakdown for answering questions about populations who are married, or who have been married and are now in one of the three post-marriage categories (“Widowed,” “Divorced,” and “Separated”).

It does bear keeping in mind that the American Community Survey is a point-in-time study, with respondents selecting the option that best reflects their marital status at the time of their response. This means that we can’t interpret the four marriage and post-marriage categories as mutually exclusive – an individual who is “now married” may not be on his or her first marriage, meaning that he or she may also have been widowed, divorced, or separated at some point in the past. However, for anyone concerned with present marital status of individuals who are or have been married, the marriage and post-marriage categories provide sound and helpful data.

It’s the monolithic “Never married” category where this level of detail starts to break down – because we can’t interpret “never married” as “single.” Some of the 46.4% of the surveyed population who have never been married may be in a relationship, but based on these categories, we can draw no sound conclusions about what that percentage might be.

Households and Families – Categories and Limitations

Another table, Households and Families, can shed a little light on this question, but also does not provide an easy breakdown into categories of “single” and “not single.” It uses a few categories that cover family households, broken down into households made up of married couples and householders with no spouse present, as well as a category for non-family households (U.S. Census Bureau; American Community Survey, 2011-2015 American Community Survey, Table S1101). Households made up of married couples can be assumed to correspond pretty strongly to the “Now married (except separated)” category of the Marital Status table, while householders with no spouse present could fall into any of the Marital Status categories other than “Now married (except separated).” This table adds no detail, and in fact can confuse the conversation, when discussing the marriage and post-marriage categories of the first table.

Where this table does add some detail is to that “Never married” group identified in the Marital Status table, because it includes lines for “Unmarried-partner households.” While we can’t directly compare these two tables, because the Marital Status table surveys individuals while the Households and Families table surveys households, this at least provides a little information on the population who are in a relationship, but are not married.

But even this brings up caveats and limitations. Some householders of “Unmarried-partner households” may fall into the “Never married” marital status category, but others may be “Widowed,” “Divorced,” or “Separated,” per the categories of the Marital Status table, and cohabitating with a partner they are not married to. As we saw, the Marital Status table calls for the respondent’s present status, and “Dating” and “Cohabitating” aren’t options (and forestalling an even more complex analytical nightmare, neither is “It’s Complicated”). Furthermore, while the “Unmarried-partner households” category provides some help in measuring non-married relationships, it offers no information on relationships where the individuals are neither married nor cohabitating.

 

Given the available Census data on the topic, we can’t arrive at a comprehensive categorization of “single” or “not single” – the categories of the data available don’t line up in a way that’s conducive to that much simplification. So 750 words later, we’re forced to revise our short answer: “almost, but actually no.”

Quantifying Winter: Climate and Weather Data Resources

It may be a new year, but for better or for worse, we have over two months of winter left. Whether you love snow or hate it, and whether you can’t get enough cold weather or can’t wait to move somewhere warm, you can always find detailed data on what the weather’s been doing where you are (or where you wish you were).

What do we measure about winter weather?

Two of the major dimensions when talking about any weather or climate topics are temperature and precipitation. Other common factors include wind speed and wind chill. Items that do not usually make their way into neatly aggregated datasets, but that could provide additional detail on the on-the-ground, so to speak, impacts of winter weather include tonnage of road salt and other de-icing agents in a given season, or the number of winter weather advisories and warnings in a given month.

How are these measurements usually taken, and by whom? Who is the source for the weather data that RPC uses?

The highly specific weather factors (temperature, precipitation, wind speed, etc.) are measured by agencies like state surveys, and are often released on a monthly basis in datasets associated with different sites.

Locally, the Illinois State Water Survey’s Illinois Climate Network releases their Monthly Summary Data reports online. This is the data that we reference for our Regional Dashboard’s Natural Resources page, and each report includes much more information than the basics in the Natural Resources page’s Weather Summary table. If you’re looking for data from other states, we recommend starting with the websites of similar agencies, like state water and geological surveys, and extending your search from there if necessary.

Two more great resources are the National Oceanic and Atmospheric Administration (NOAA) and NOAA’s National Centers for Environmental Information (NCEI). These websites offer data, analysis, news items, and more, for the nation as a whole, categorized by topic, and broken down by region.

How does CCRPC use environmental data?

The plans, reports, and studies that CCRPC produces often have an environmental component, whether it’s an impact review, a sustainability chapter, or an appendix of data and analysis. Weather and climate resources like those discussed above help us create better documents based on sound data.

What about measuring other potentially winter-related occurrences, such as traffic incidents?

It’s easy to hypothesize about incidents and occurrences to measure and analyze that could be correlated to winter weather. A prime example of this is traffic. That a greater number of traffic incidents occur either in winter months or on days with inclement weather are both reasonable hypotheses.

There is data available on traffic incidents in our area: CUUATS produces the Selected Crash Intersection Location (SCIL) reports, which include facts about traffic incidents at various intersections in Champaign-Urbana. These reports cover overlapping five-year periods. The earliest report available covers 1993-1998, while the most recent report, published in 2015, covers 2009-2013. All reports dating back to the 2001-2005 edition include some analysis on weather and road surface conditions.

In addition to the SCIL reports, data on traffic crashes that resulted in severe injuries or fatalities between 1999 and 2014, including details on weather, road surface, and other conditions and associated spatial data are available on the CUUATS Data Portal.

What about information on winter preparedness and safety?

Of course, there’s more to winter weather information than knowing this month’s total precipitation or minimum temperature: winter safety is important in any temperate climate. You can usually find winter weather advisories wherever you find your forecast information, which will keep you up to date on major weather events coming up in the East Central Illinois area.

In addition to knowing what weather is on the horizon, it’s important to know how to prepare for winter storm events in general. The National Weather Service and National Oceanic and Atmospheric Administration have a Winter Safety webpage full of tips and advice for preparing for winter weather and staying safe through the season.

 

We talk a lot on this blog about the importance of good data – that’s true regardless of the topic. Whether we’re using this data as part of our Regional Dashboard or as an input in an environmental impact review, or whether we just want complain about the weather in a quantifiable way, these climate and weather data resources are important to be familiar with.

 

Data Literacy: Using the American Community Survey

It’s already December, and that means it’s time for another data literacy post: this time, in honor of the upcoming American Community Survey (ACS) 2011-2015 5-Year Estimates release on Thursday, December 8th, we’ll be talking about what the ACS is, what topics it covers, how it can be used by members of the public, and what we use it for here at CCRPC.

 

What is the ACS?

The ACS is a data product from the U.S. Census Bureau. It is a series of estimates that cover a comprehensive range of topics and geographic areas, from the general to the highly specific. ACS data is available for free in detailed tables, area summary reports, and maps.

 

What data topics are included in the ACS?

The ACS covers a wide range of topics. Two of the major categories are demographics (e.g., age, sex, race, ethnicity, etc.) and economics (e.g., income, poverty, employment). Beyond those areas, there is also data on education, transportation behaviors, native and foreign-born populations, household and family topics, veteran status, populations having a disability, housing topics, and more.

 

Who can access and use ACS data, and where can I find ACS data?

Everyone! Data from the ACS, like data from other Census products, is public and free to use. Obviously, participants’ full responses are not released, since those include personal information like names and addresses. But once estimates are compiled into tables and published, anyone can go to Census.gov and access the data, and anyone can use the data as a source for projects, papers, presentations, or just general curiosity.

The user interface that allows access to most Census data is American FactFinder: users define their geographies of interest, pick their years of interest, and then get down to searching for their topic. If you’re not sure how to get started, the Census Bureau offers a number of tutorials on how to use American FactFinder, covering tasks from the basic to the advanced.

It should go without saying that, if you’re going to use ACS or other Census data, you should make sure you cite it. There is a recommended citation format for tables from American FactFinder outlined in a U.S. Census FAQ (and also, handily, at the bottom of this page, where we footnote some income figures); that’s what we use here at CCRPC when we use Census data, and we strongly encourage other data users to take the time to cite it right.

 

How often is the ACS produced and released?

The ACS is conducted on a rolling basis. Unlike the Decennial Census, where every household in the nation is asked to respond, only some households receive materials to respond to the ACS each month. Data collected from these sample populations is then extrapolated into the estimates about the total population.

Both datasets are released every year. In general, the new 1-Year Estimates are released in September of every year, and the new 5-Year Estimates are released in December.

 

What’s the difference between 1-Year Estimates and 5-Year Estimates?

There are actually several. The primary difference is the timeframe that the data covers. The 1-Year Estimates are applicable to the year named in their title: for example, the 2015 1-Year Estimates that were released in September are applicable to 2015. The 5-Year Estimates are applicable to the five-year period named in their title: for example, the 2011-2015 5-Year Estimates that will be released on December 8th cover the entire five-year period. Keep in mind that figures from the 5-Year Estimates are not the same as the figures from single years within that five-year period, and cannot be presented as such. Put simply, the 5-Year Estimates are a five-year average, while the 1-Year Estimates cover their single title year.

To offer a concrete example, let’s say that you’re looking for data on income, and you specifically want to know what percentage of households in Champaign County made between $25,000 and $34,999 in 2014. In that case, you should use data from the 2014 1-Year Estimates, and the figure you would find is 10.3% (+/-1.6)[1]. If you use the 2010-2014 5-Year Estimates, you’ll find a figure of 10.5% (+/-0.7)[2]. The figure from the 5-Year Estimates is slightly different, and it cannot be used to accurately represent the percentage of households in that income bracket in 2014. Both of these things are because the 5-Year Estimates include not only data collected in 2014, but also data collected in 2013, 2012, 2011, and 2010.

Also for this reason, you shouldn’t compare data from 5-Year Estimates datasets with overlapping years: for example, when the 2011-2015 5-Year Estimates dataset is released next week, you can’t compare its data with data from the 2010-2014 5-Year Estimates, because both datasets include the data collected in 2011, 2012, 2013, and 2014. If you want to compare 5-Year Estimates datasets, it’s only really doable with datasets with non-overlapping periods (e.g., comparing the 2005-2009 5-Year Estimates and the 2010-2014 5-Year Estimates, or comparing the 2006-2010 5-Year Estimates and the 2011-2015 5-Year Estimates).

Another major difference between the 1- and 5-Year Estimates datasets is the geographies they’re available in. The 1-Year Estimates include data from geographies with populations of 65,000 or more, while the 5-Year Estimates include data from all geographies in the United States. The limited number of geographic areas means that, even if you’re looking for data from a single year, it may be necessary to use 5-Year Estimates because your geography has a population of below 65,000, and has no 1-Year Estimates available. In that case, it’s important to know how to work with the 5-Year Estimates, and how to interpret and talk about data from that resource.

One more thing to keep in mind is that the margins of error of the 5-Year Estimates datasets tend to be smaller than the margins of error of the 1-Year Estimates datasets, because the 5-Year Estimates include more responses. In general, the larger the input dataset, the smaller the margin of error. This may be a reason to opt for using a figure from the 5-Year Estimates even when data is available from your geography of interest in the 1-Year Estimates.

Finally, data from 1-Year Estimates can’t be compared to data from 5-Year Estimates, even if they’re the same figure drawn from the same table and covering the same geography – the difference in their respective time periods makes the datasets not comparable.

 

Why does ACS data include margins of error when Decennial Census data doesn’t?

As we noted above, and have mentioned in previous blog posts, ACS data are estimates, and estimates can always be off, either by a small amount or by a massive amount. The margins of error reflect that range of possibilities: that the estimate is X, but that the actual number could be as high as X+Y, or as low as X-Y. For more information on margins of error, check out our last data literacy post, which covers them in detail.

On the other hand, Decennial Census data are exact counts. Decennial Census surveys are now much shorter and have much less detail than the ACS, but they are point-in-time counts instead of estimates based on a population sample. Counts don’t have margins of error because they’re assumed to be absolutely correct (give or take some non-responding households) – they don’t need them.

 

How does CCRPC use ACS data?

The ACS is a great resource for public agencies as much as it is for individuals. Having the wide variety of topics available on a consistent basis is a big part of what allows us to develop and maintain this website, its indicators and analysis, and the regional dashboard. We use ACS data in our own reports, like the Long Range Transportation Plan. We also use it to create reports for other agencies and organizations, and to respond to data requests from private individuals. Not every request or report uses ACS or Census data, but many of them do.

ACS data is necessary for our work beyond any specific plans, reports, and data requests. CCRPC is one of many public, service-providing agencies in Champaign County and around the country; for organizations like ours, knowing about the population we serve and represent, both in general and in detail, is fundamental to our operations. The ACS may not be a perfect resource, and may not be suited to every data need or project, but in many cases, it is the best and only data that is reliable, released on a regular schedule, and free to use. It has enormous value, to CCRPC and as a public resource.

For even more information on the ACS, check out the ACS homepage on the Census Bureau website. In closing, we’d just like to say that we love our Census data, and that the ACS is a great resource for a wide range of users and topics. So check it out! You’ll probably find something interesting, too.

 

[1] U.S. Census Bureau; American Community Survey, 2014 American Community Survey 1-Year Estimates, Table S1901; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (28 November 2016).

[2] U.S. Census Bureau; American Community Survey, 2010-2014 American Community Survey 5-Year Estimates, Table S1901; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (28 November 2016).

Voter Participation Statistics: Methods and Sources

You may already be aware of this, but Election Day is right around the corner. And although there are a lot of polarizing issues under discussion this season, we can all at least agree that we love statistics. (Right? Right?!) So here’s the rundown on voter participation rate statistics, how they’re calculated, and where to find them.

 

Compared Voter Participation Methods

Voter participation statistics can be calculated in two different ways, but both center on comparing the number of ballots cast with a larger population.

For bonus points, using the same data needed for these calculations, you can also work out the voter registration rate by comparing the number of registered voters with the population that is eligible to vote.

Method 1: Compares the number of ballots cast with the total population that is eligible to vote

Voter Participation Rate (%) = (Number of Ballots Cast)/(Eligible Population) x 100

Method 2: Compares the number of ballots cast with the number of registered voters.

Voter Participation Rate (%) = (Number of Ballots Cast)/Registered Voters) x 100

Both methods are in use, and both are sometimes referred to as “voter participation rate,” so it’s good to be familiar with both. And neither method is categorically better than the other one. All this also makes it important, when you’re reading voter turnout analyses, to check how they’re doing their calculations. As an example of how the two different methods return different results, let’s take a look at some Champaign County data from past elections.

Table 1: Voter Participation Rate Methods[1]

Eligible Population Registered Voters Ballots Cast Participation Rate: Method 1 Participation Rate: Method 2
2006 137,209 113,905 53,869 39.3% 47.3%
2008 141,979 123,150 84,804 59.7% 68.9%
2010 146,955 122,441 54,818 37.3% 44.8%
2012 150,055 112,933 78,939 52.6% 69.9%
2014 153,959 113,122 55,434 36.0% 49.0%

 

As you can see in the tables above, comparing ballots cast with the eligible population will almost invariably return a lower participation rate than comparing ballots cast with the number of registered voters. The only exception is in cases of complete voter registration saturation, where every person that is eligible to vote is registered to vote: in that case, the participation rate figures would be equal. Either method is viable and neither is the preferable option across the board, but it’s important to know which one is being used for the analysis you’re reading: ask the right questions, and your interpretation of the data will be sounder than assuming one method or the other. (This is a good rule to go by in most data analysis situations: read the methodology, be a more informed data consumer.) Also, if you’re calculating or compiling voter participation rate statistics for a study or a project that you’re working on, it’s important to check the methodologies and make sure that the different datasets are actually comparable. If the rate in one state was calculated with Method 1, and the rate in another state was calculated using Method 2, these two numbers can’t give you an apples to apples comparison: you either need to select new regions to analyze or get the raw data and recalculate the rates yourself, using the same method for every region.


Historic Voter Participation Statistics

If you’re looking for historic voter participation statistics, you can find Champaign County’s on the Champaign County Clerk website. The results available go back to 1968, and many years include results from both primary and general elections, as applicable, so you can go forth and do historical analyses of the county’s voting record to your heart’s content. At the national level, the U.S. Census Bureau produces some voting and registration products based on the Current Population Survey. If you’re looking for data only for a particular state, we recommend starting with that state’s board or department of elections and going from there; for Illinois, that’s the Illinois State Board of Elections, which provides historic election results at the state and county level. And finally, if you’ll excuse a moment of shameless self-promotion, we’ve compiled some historic Champaign County voter turnout data for our Voter Participation indicator.

Champaign County’s voter participation statistics have two staggered release dates. Unofficial statistics will be available on the Champaign County Clerk website after polls close on November 8th. The full reports will be available on the Champaign County Clerk website two weeks after Election Day.


Voter Participation Statistics, Additional Analysis, and Why This Data is Important

If you’re looking for analysis in addition to just the raw numbers, the U.S. Census Bureau has a “Voting and Registration” topics page where you can find the voting and registration products mentioned above, as well as resources like state voting profiles; topical reports; national, state, and regional-level data; and voting data broken down by age, sex, race, ethnicity, marital status, and other demographic dimensions.

Being able to reliably measure and compare voter participation statistics is important for several reasons. Working with this data gives the average data user a thorough grounding in election statistics. The same datasets that provide the raw data for voter participation rate can also include candidate- or party-based breakdowns of the election results, which, depending on what level of geography you’re working with, can provide a county-level view of election results. Finally, comparing the participation and registration rates at the county level, both among counties and from year to year, can provide information on how much of the population in that area is – or isn’t – voting. This can provide insight into concepts that are harder to pin down with data, like civic engagement, and also suggest when and where there might be a need for greater outreach in order to encourage residents of that area to vote or otherwise get involved in the local political process.

This all brings us to our last note before we close out the entry on this topic: read on for the obligatory (but important) public service announcement.


Obligatory Public Service Announcement

To wrap up, we need to take a moment to be serious and encourage you to go out and vote this election season.  Early voting is open today, this weekend, and Monday at locations throughout Champaign County, and if you do vote early, you can vote at any open early voting location, instead of only at your assigned polling place. To find early voting locations, check the Early Voting Schedule on the Champaign County Clerk website. To get your personal voter information, including checking your voter registration status and assigned polling place, go to the Voter Information Lookup page on the Champaign County Clerk website. If you’re planning to vote on Election Day but don’t know how you’re going to get to your polling place, CUMTD has you covered: MTD is offering free bus rides on November 8th to help people get out and vote.

 

[1] Sources: http://www.champaigncountyclerk.com/elections/results/index.html, http://www.elections.il.gov/Publications.aspx, http://elections.gmu.edu/voter_turnout.htm, U.S. Census Bureau; American Community Survey, 2006 American Community Survey 1-Year Estimates, Table B05003; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (15 March 2016), U.S. Census Bureau; American Community Survey, 2008 American Community Survey 1-Year Estimates, Table B05003; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (15 March 2016), U.S. Census Bureau; American Community Survey, 2010 American Community Survey 1-Year Estimates, Table B05003; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (15 March 2016), U.S. Census Bureau; American Community Survey, 2012 American Community Survey 1-Year Estimates, Table B05003; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (15 March 2016), U.S. Census Bureau; American Community Survey, 2014 American Community Survey 1-Year Estimates, Table B05003; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (15 March 2016).

Measuring the Urban Forest: Benefits and Methods

With fall finally underway, our urban forest is heading into leaf season. And with increased interest in green infrastructure has come increased interest in measuring the benefits of the urban forest – defined as “all publicly and privately owned trees within an urban area” [1]  by the U.S. Department of Agriculture’s Forest Service. The urban forest provides numerous ecosystem services, ranging from decreased stormwater runoff, flooding, urban heat island conditions, and energy use to increased habitat and improved air quality[2].

So where do measurement and data come into this?

Part of it is fiscal. Many of the ecosystem benefits provided by the urban forest are among the goals of implementing green infrastructure – or “vegetation, soils, and other elements and practices [used] to restore some of the natural processes required to manage water and create healthier urban environments,” [3] as defined by the U.S. EPA. However, traditional stormwater infrastructure – the pipes, drains and culverts that make up much of how municipalities currently handle runoff – is a known quantity, and deviating from longstanding practice can be a hard, and risky-sounding, proposition to sell. Being able to measure the urban forest and quantify the monetary value of the ecosystem services it provides can be a strong argument in favor of green infrastructure techniques and practices.

In addition to the advantages of being able to communicate about the benefits of the urban forest in terms of monetary value, there are also operational advantages to surveying the urban forest. Municipal public works departments care for and maintain trees on public land and in public rights-of-way, and having detailed data on the age, size, and species of these trees can aid in planning and implementing maintenance tasks. Tree species can be an especially important data point with species-specific pests and problems, such as the emerald ash borer, a major concern.

So what kinds of tree data and other resources are available for the Champaign-Urbana area? On its Forestry Section webpage, the City of Champaign offers a percentage breakdown of tree species accounting for more than one percent of the total trees in the City rights-of-way. The City of Urbana, which has held the designation of Tree City USA since 1976[4], has a publicly accessible tree inventory that shows not only trees on public land in Urbana, but their calculated ecosystem benefits, both aggregated and categorized as greenhouse gas benefits, water benefits, energy benefits, air quality benefits, or property benefits. The University of Illinois was named a Tree Campus USA in 2015[5], and has published the plan 2015 Illinois Tree Campus: A Tree Care Plan for the University of Illinois at Urbana-Champaign. The University of Illinois Extension also offers numerous tree-related resources on its website.

Like any data resource, tree inventories don’t just grow on trees. (Sorry, not sorry.) As we noted in our July 1 post, all data collection and publication processes require staff time and (let’s all say it together, now) funding. So as useful as full tree inventories are (and we applaud the City of Urbana’s), there can be operational barriers to developing one. If you’re trying to quantify or assign a monetary value to ecosystem benefits, without a publicly accessible, comprehensive tree inventory to work from, we’ve got a couple suggestions on how to get started.

Use of a tree value calculator is the easiest way to go, and there are a few options available online that are free to use. One option is the i-Tree suite of applications, which includes a variety of tools appropriate for different scales, levels of detail, and topics. The i-Tree suite is developed and maintained by the U.S. Forest Service, Davey Tree Expert Co., the Arbor Day Foundation, the Society of Municipal Arborists, the International Society of Arboriculture, and Casey Trees. Another option is the Tree Benefit Calculator, also developed by Casey Trees and Davey Tree Expert Co., which is based on i-Tree data and provides data on the benefits of individual trees.

Once you choose a tool, the next step is to get your input data. The i-Tree tools allow the user to identify their area of interest on a map in the application, so minimal, if any, data collection on the user’s part is needed. The Tree Benefit Calculator requires the user’s zip code of interest, then the species and diameter of the tree of interest. If you’re only looking for the value of the trees in a small area that you have access to (the trees on your property, for example, or in the immediate area of your academic building), doing a manual survey of trees in the area by species and performing the basic diameter measurement for each one can be a viable option. If you’d only like to know the benefit you would get by planting one more tree in your yard, that’s easy – simply input one tree of the type or types you’re considering, and there you have it.

But if your area of study is larger (like your neighborhood or your entire zip code) and manual data collection is out of the question, it might be time to call your municipality’s public works department. They may be willing and able to share whatever data they have on trees on public land – as far as trees on private land go, you’re likely out of luck unless it’s private land that you happen to own.

The urban forest is a complex system, but methods of collecting data and learning about it are becoming more common and accessible. So enjoy using the available data resources, and then go out and enjoy the urban forest.

[1] David J. Nowak, Susan M. Stein, Paula B. Randler, Eric J. Greenfield, Sara J. Comas, Mary A. Carr, and Ralph J. Alig, Sustaining America’s Urban Trees and Forests (United States Department of Agriculture Forest Service: General Technical Report NRS-62), 2010. PDF. http://www.fs.fed.us/openspace/fote/reports/nrs-62_sustaining_americas_urban.pdf

[2] The Value of Green Infrastructure (Center for Neighborhood Technology, American Rivers), 2010. PDF. http://www.cnt.org/sites/default/files/publications/CNT_Value-of-Green-Infrastructure.pdf

[3] “What is Green Infrastructure?”, US EPA, last modified September 23, 2016, https://www.epa.gov/green-infrastructure/what-green-infrastructure.

[4] “Urbana’s Tree City USA Designation,” City of Urbana, created November 4, 2009, http://www.urbanaillinois.us/visitors/tree-city-usa.

[5] “Tree Campus USA,” University of Illinois at Urbana-Champaign Facilities & Services, accessed September 26, 2016, http://fs.illinois.edu/services/more-services/grounds/tree-campus-usa.

Impacts of University Populations on Communities’ Median Age, Median Income, and Poverty Rate

A view across the quad on the University of Illinois at Urbana-Champaign campus. Paths cross the quad and students are studying. A large brick building, with columns and an oxidized copper dome is at the far end of the quad.

It’s now September, and in Champaign-Urbana and other cities and towns with major universities, that means the population has risen with the start of the fall semester. But how does the university population affect Champaign-Urbana’s overall statistical outlook? And how are college students counted for U.S. Census Bureau programs?

To answer the second question first, a major goal of Census programs is to count everyone where they live for the majority of the time. For students enrolled in residential colleges, that usually means in the community where their school is located, not at their family’s home address. Even if the student is at their home address at the time a Census form is completed, they should not be counted as living at that address unless they are actually a current, full-time resident there. So all total population figures, both counts and estimates, released by the U.S. Census Bureau for a given area should include the student body of any residential universities and colleges with the boundaries of that area, with some allowance for respondent error.

A large university population can impact a community’s overall statistical makeup. To start, let’s look at the university enrollment and total population. The table below shows population, enrollment, and student percentage of total population in several communities with large universities, and one community, Decatur, IL, with a similarly sized total population but with a much smaller university, for the sake of comparison.

UniversityCommunityStudent Population Total City Population % of Total Population Accounted for by Students
EnrollmentYearEstimateMOE
University of Illinois at Urbana-ChampaignChampaign-Urbana, IL45,1402016124,6087636.2%
Champaign, IL82,84851
Urbana, IL41,76056
Indiana UniversityBloomington, IN46,416201681,96335056.6%
Purdue UniversityLafayette, IN38,770201669,9823755.4%
University of IowaIowa City, IA31,387201670,5975844.5%
University of MichiganAnn Arbor, MI43,6252016115,9855037.6%
Virginia TechBlacksburg, VA31,224201643,2042972.3%
Millikin UniversityDecatur, IL2,191201675,0884182.9%

Sources: U.S. Census Bureau; American Community Survey, 2010-2014 American Community Survey 5-Year Estimates, Table B01003; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (24 August 2016).; U.S. News & World Report. Higher Education. (2016). University of Illinois — Urbana-Champaign. <http://colleges.usnews.rankingsandreviews.com/best-colleges/uiuc-1775>. (24 August 2016).; U.S. News & World Report. Higher Education. (2016). Indiana University — Bloomington. <http://colleges.usnews.rankingsandreviews.com/best-colleges/indiana-university-1809>. (24 August 2016).; U.S. News & World Report. Higher Education. (2016). Purdue University — West Lafayette. <http://colleges.usnews.rankingsandreviews.com/best-colleges/purdue-1825>. (24 August 2016).; U.S. News & World Report. Higher Education. (2016). University of Iowa. <http://colleges.usnews.rankingsandreviews.com/best-colleges/university-of-iowa-1892>. (24 August 2016).; U.S. News & World Report. Higher Education. (2016). University of Michigan — Ann Arbor. <http://colleges.usnews.rankingsandreviews.com/best-colleges/university-of-michigan-9092>. (24 August 2016).; U.S. News & World Report. Higher Education. (2016). Millikin University. <http://colleges.usnews.rankingsandreviews.com/best-colleges/millikin-university-1724>. (31 August 2016).; U.S. News & World Report. Higher Education. (2016). Virginia Tech. <http://colleges.usnews.rankingsandreviews.com/best-colleges/virginia-tech-3754>. (31 August 2016).

Compared to the population percentages accounted for by the other major universities, Champaign-Urbana’s student percentage is actually relatively modest: at  36.2%, it’s a big portion of the total population, but for perspective, check out some of the other percentages. In Bloomington, IN and Lafayette, IN, students make up more than half the estimated populations of the cities. In Blacksburg, VA, Virginia Tech students account for nearly three-quarters of the estimated population. And on the other end of the spectrum, Decatur, IL, with its small university, has a student population percentage of only about 3%.

We hypothesize that the presence of these large universities will affect the statistical makeup of their communities: lower median ages and median incomes, higher poverty rates. We further hypothesize that larger student population percentages will have greater impacts. So let’s see, below, how these theories bear out.

Median Age

UniversityCommunity% of Total Population Accounted for by StudentsMedian Age 
EstimateMOE
N/AUnited StatesN/A37.40.1
Millikin UniversityDecatur, IL2.9%40.70.8
University of Illinois at Urbana-ChampaignChampaign-Urbana, IL36.2%
Champaign, IL27.10.6
Urbana, IL23.90.5
Indiana UniversityBloomington, IN56.6%23.50.2
Purdue UniversityLafayette, IN55.4%31.60.6
University of IowaIowa City, IA44.5%25.90.4
University of MichiganAnn Arbor, MI37.6%27.90.4
Virginia TechBlacksburg, VA72.3%21.80.2

Source: U.S. Census Bureau; American Community Survey, 2010-2014 American Community Survey 5-Year Estimates, Table S0101; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov); (24-25 August 2016).

Compared to the national figure, all five comparison communities with large universities have lower estimated median ages, and the highest student percentages correspond with the lower estimated median ages. Bloomington, IN has the second-highest student percentage at 56.6% and the second-lowest median age of 23.5 (+/-0.2), while Blacksburg, VA has both the highest student percentage (72.3%) and the lowest median age (21.8, +/-0.2). But this trend is not consistent throughout the large university comparison communities. Lafayette, IN, home of Purdue University, has the third-highest student percentage at 55.4%, but its median age of 31.6 (+/-0.6), while lower than the national average, is the highest of the large university comparison communities. In contrast, Decatur, IL, which we predicted would not see major effects from its small university population, has a higher estimated median age than the United States as a whole.

Median Income

UniversityCommunity% of Total Population Accounted for by StudentsMedian Household Income 
EstimateMOE ($)
N/AUnited StatesN/A$53,48295
Millikin UniversityDecatur, IL2.9%$39,5881,507
University of Illinois at Urbana-ChampaignChampaign-Urbana, IL36.2%
Champaign, IL$42,0771,833
Urbana, IL$30,8342,659
Indiana UniversityBloomington, IN56.6%$28,6602,147
Purdue UniversityLafayette, IN55.4%$39,3781,676
University of IowaIowa City, IA44.5%$42,1191,417
University of MichiganAnn Arbor, MI37.6%$56,8351,320
Virginia TechBlacksburg, VA72.3%$29,2714,162

Source: U.S. Census Bureau; American Community Survey, 2010-2014 American Community Survey 5-Year Estimates, Table S1903; generated by CCRPC staff; using American FactFinder; http://factfinder2.census.gov; (24-25 August 2016).

A low median income can be another indicator of a college or university town. Full-time students, both graduate and undergraduate, often have low household incomes, and a preponderance of student households can draw down the community-wide estimated median income.

The relationship between community-wide median income and student population percentage shows itself in much the same way as the median age relationship. Four of the university comparison communities have estimated median incomes below the national estimate. Bloomington, IN and Blacksburg, VA, with the two highest student percentages, have the two lowest estimated median incomes of the analyzed group. Lafayette, with the third-highest student percentage, has an estimated median income lower than the national figure, but not nearly as low as Bloomington, IN’s, despite their similar student percentages.  Ann Arbor, MI has an estimated median income slightly higher than the national estimate. This goes against our first hypothesis – that a large student percentage will lead to a lower estimated median income – but not necessarily against our second, that a smaller student percentage will has lesser impacts than a larger one, since Ann Arbor’s student percentage is the lowest of the large university comparison communities. Decatur, IL’s estimated median income is lower than the national estimate – and very similar to Lafayette, IN’s, despite their very different student percentages.

Poverty Rate

UniversityCommunity% of Total Population Accounted for by StudentsPoverty Rate 
EstimateMOE
N/AUnited StatesN/A15.6%0.1
Millikin UniversityDecatur, IL2.9%24.3%1.7
University of Illinois at Urbana-ChampaignChampaign-Urbana, IL36.2%
Champaign, IL27.3%1.6
Urbana, IL36.0%2.5
Indiana UniversityBloomington, IN56.6%39.0%1.5
Purdue UniversityLafayette, IN55.4%20.4%1.8
University of IowaIowa City, IA44.5%27.6%1.5
University of MichiganAnn Arbor, MI37.6%22.6%1.2
Virginia TechBlacksburg, VA72.3%46.3%2.7

Source: U.S. Census Bureau; American Community Survey, 2010-2014 American Community Survey 5-Year Estimates, Table S1701; generated by CCRPC staff; using American FactFinder; http://factfinder2.census.gov; (24-25 August 2016).

Poverty rate can be a complicated measure in any community, with or without a large college or university, because a community’s entire population is not necessarily measured when poverty is calculated. The Census Bureau’s list of “People Whose Poverty Status Cannot Be Determined” includes people in the following living situations: “institutional group quarters (such as prisons or nursing homes),” “college dormitories,” “military barracks,” “living situations without conventional housing (and who are not in shelters),” and “unrelated individuals under age 15.”[1]

For the purposes of this blog post, the relevant group on that list is residents of college dormitories. Regardless of their level of income, residents of college dorms are not counted toward the poverty rate. This means that, although a large student population can have an effect on a community’s poverty rate, assuming that a segment of that population lives off-campus, the effect is less than it would have been if poverty calculations included dorm residents.

The poverty rates of Champaign, Urbana, and all five comparison communities with large universities are above the national rate: some, like Ann Arbor, MI’s 22.6% (+/-1.2) and Lafayette, IN’s 20.4% (+/-1.8) are relatively close to the national rate, while others, like Urbana’s 36.0% (+/-2.5), Bloomington, IN’s 39.0% (+/-1.5), and Blacksburg, VA’s 46.3% (+/-2.7) are more than double the national rate of 15.6% (+/-0.1). Bloomington and Blacksburg continue to carry our second hypothesis: they’re communities with larger student percentages showing the predicted impacts to a greater degree. But Lafayette undermines it again: with a very similar student percentage to Bloomington’s, it shows the predicted impact much less. And Decatur complicates the comparisons further, with its low student percentage and an estimated poverty rate that is also above the national estimate.

The primary figure in the University of Illinois at Urbana-Champaign's Alma Mater statue. The figure is wearing robes and stands with open arms.At this point, our first hypothesis is holding up for the most part, but our second is looking unsteady – and that’s because correlation does not imply causation. We can reasonably say that large universities and their student populations have some impact on the overall statistical makeup of the communities in which they’re located. But they’re not the only factor. Yes, a large student population can draw down the median age by forming a disproportionately large population of young adults. But a community with a high birth rate and a disproportionately large number of young children could have a similarly low median age. So, possibly, could a community in which a large percentage of the late middle-aged or elderly population had chosen to retire elsewhere and moved away, affecting the median age via absence. A large university is definitely not the only plausible reason for such a trend.

We could also present lists of alternate causes for lower estimated median incomes and higher poverty rates. They can often appear with large universities, and sometimes communities with larger universities have lower estimated median ages and incomes and higher poverty rates than communities with smaller ones. But high poverty rates, low estimated median incomes, and low estimated median ages also appear in communities without large universities, in all different degrees and combinations, because there are other factors at play.

The takeaway from this post? If a community with a large university has one of those statistical hallmarks, the student population can be said to account for some of it. But we can’t make sound, generalized predictions about a community’s estimated median age, estimated median income, or poverty rate based solely on whether or not there’s a university there, or attribute these factors solely to a university’s statistical impact. It’s not that simple.

*A brief public service announcement: we realize that our two datasets are not from the same year. The American Community Survey data is from the 2010-2014 5-Year Estimates dataset, and the university enrollment data, from U.S. News & World Report’s College Rankings and Reviews, is all dated 2016. This is a clear case of “do as we say, not as we do” – for the purposes of a loose analysis, we’re making the assumptions that enrollment at these universities has neither doubled nor seriously dropped in the last two years, and that the items taken from the ACS for each of these geographies have also made no hugely dramatic shifts. But don’t take the analysis in this post as having serious statistical rigor (it doesn’t), and if you’re attempting an analysis with serious statistical rigor, make sure your data periods match up.

[1] U.S. Census Bureau. (2016). “How the Census Bureau Measures Poverty.” <http://www.census.gov/topics/income-poverty/poverty/guidance/poverty-measures.html>. (Accessed 29 August 2016).