September 1, 2017
As students of all ages return to local classrooms, we return to education for September’s blog post. If you’ve spent any time exploring the available indicators and Regional Dashboard datasets, you’ll have noticed that there are many related to education in both sections, and that both sections include school enrollment data – from two different source types providing two different sets of figures. This month, we’re taking a look at these datasets, why they’re different, and which source is best suited to what questions.
One of the first decisions to make with any analysis using enrollment data is which of the available datasets to use. Our College & University Enrollment indicator uses data directly from the University of Illinois and Parkland College, while the enrollment data on the Regional Dashboard: Education page is sourced from the American Community Survey. While they both pertain to enrollment, there are differences in what exactly is being measured, which lead to marked disparities in the final figures: over the entire 10-year period of interest, total enrollment based on the local data is greater than total enrollment according to ACS estimates, in some years by over 15,000 students.
Source: U.S. Census Bureau; American Community Survey, 2005-2015 American Community Survey 1-Year Estimates, Table S1401; generated by CCRPC staff; using American FactFinder; <http://factfinder2.census.gov>; (28 August 2017)., Parkland College. Website. Summary Table A1: Annual Enrollment and Completion Data. FY 00 – FY 15. <http://www.parkland.edu/about/IAR/annualenrollment.aspx>. (Accessed 6 March 2017)., Division of Management Information. University of Illinois at Urbana-Champaign. Website. UIUC Student Enrollment. Years 2000-2016. <http://www.dmi.illinois.edu/stuenr/#class>. (Accessed 6 March 2017).
The locally-sourced data from the University of Illinois and Parkland College, as would be expected, measure only the enrollment for the University of Illinois and Parkland College, respectively. This means that individuals that live in Champaign County, our area of interest, who are enrolled in colleges, universities, or programs that are neither the University of Illinois nor Parkland College (e.g., students that commute to other institutions, or that are enrolled in online classes or degree programs from other institutions) are not counted, but that students who are not residents of the county and commute to campus are.
The ACS estimates include, in theory, all individuals whose primary residence is Champaign County and that are enrolled in an educational institution of some kind. Institutions are categorized by level of education (e.g., elementary school, high school, undergraduate college, graduate or professional school), and by whether they are public or private (which is beyond the scope of the analysis of this blog post). Other items of interest, such as whether the program is on a campus or online, whether the respondent is a full- or part-time student, and the location of the program, are not recorded in the ACS. Some detailed education data, such as program type and time status, are collected by the U.S. Census Bureau’s Current Population Survey (CPS).
Certainly, many students are counted in both enrollment datasets, but the respondent pools are not identical. And at first glance, it would seem that, by counting all enrolled college students at any institution that are residents of Champaign County, the ACS number should be greater than the local number, which is restricted to only University of Illinois and Parkland students. But as evidenced by the data, that is not the case. We can posit a few factors that might account for some of this disparity.
As we noted above, students who live outside of Champaign County and commute to the University of Illinois or Parkland College would be counted toward each institution’s enrollment statistics, because they are students of the institutions, but would not be counted toward the Champaign County ACS estimates, since they are not county residents.
Human error in ACS responses
The ACS does not collect data about full- or part-time enrollment, and its higher education response options are limited to “College, undergraduate” and “Graduate, professional school,” which are both readily associated with the pursuit of degrees or certificates. It is possible that some part-time students, enrolled in a very small number of classes and/or not pursuing a degree program but taking classes for professional development or personal interest, may not choose either of those two ACS options, and so may appear in institutional enrollment statistics but not in ACS estimates. Recording students in Census Bureau programs is challenging in general, as we discussed in our September 2016 blog post. All respondents should respond to surveys where they maintain their primary residence, and in the case of full-time, non-commuting college and university students, this would be their school address, rather than their previous home address. But surveys are not always completed correctly, and this may have some impact on ACS estimates as well.
These are only hypotheses: without significant further study, we can’t be sure to what degree commuter students and ACS response errors do, or don’t, affect the disparity between local and ACS enrollment data. (For those interested, the U.S. Census Bureau does release in-depth studies of how respondents interpret and respond to specific survey questions, as well as working papers on a variety of other topics, but at the time of publication, we don’t know of any recently released studies that deal with the ACS education questions.)
When we are unsure why datasets relate to each other as they do, it is even more important to select the one that best suits your project as one of your first steps and stay consistent with it throughout the process. For these two datasets, we sum up the difference as follows:
Local, institutionally-based enrollment data does not care where students live, only that they are enrolled in the institution. ACS estimates of enrollment in a given area do not care where students are enrolled, just where they live.
We recommend that, when you’re interested specifically in the student body at either of our two local institutions, you opt for institutional data. However, if you’re undertaking a project that includes analysis of any additional county-level data, we recommend using ACS estimates, especially if any of your additional data is also U.S. Census data: local data is ideal when possible, but consistent and comparable data is always necessary.