Higher Ed Data Stories

You can look at a lot of places on this blog to find the story of women and the increases in educational attainment over time, but perhaps none is so compelling as this one.  It was very rare for women to have college degrees in the 1940's and 1950's , but even rarer to find doctors, lawyers, and dentists who were women. As you'll see below, that all changed in the late 1960's and early 1970's.  What happened? It's probably a lot of things, but you could probably do worse than to point to birth control as a major contributing factor. There are four views of the data from the Digest of Educational Statistics : View 1 shows all degrees over time to men and women; the top via stacked bars, and the bottom using line charts.  The top chart shows the dramatic increase in degrees to women; the bottom shows that in 1955-56, almost all degrees (blue line) were awarded to men (purple line.) View 2 shows the same data, presented a different way.  On the top chart, ...

Note: You should view this on a tablet or desktop. The Western Interstate Commission for Higher Education (WICHE) has just published the 9th version of " Knocking on the Door ," a look at demographic projections of high school graduates in the US.  And several organizations have already published interesting views of the data, like this one on the WICHE site and this one on Hechinger Report . As I make my case for more self-service BI, these are great examples of what I call the 80/80 rule: Eighty percent of what an analyst will give you is not what you need as a practitioner; and 80% of your questions won't be answered when someone else does the analysis for you.  So I took the data ( and allow me to complain a little bit about putting data for 50 states and DC in 51 worksheets in an Excel workbook, WICHE ) and spent a lot of time restructuring it for analysis.  Then I started asking my questions, and came up with 6 views, in an attempt to provide practitioners maxi...

One of the most popular posts on this blog has been this one , where I showed the baccalaureate college of the nation's 2011 doctoral recipients. This is an update to that, using 2014 data from the NSF . It's pretty simple: There are three views here. On the first view, you can see the undergraduate college of all doctoral recipients in 2014.  The view starts with known US institutions only, but you can add in foreign or unknown institutions if you'd like.  You can also look at a single state, the degrees awarded at the institution (for instance, if you have a student who really wants a liberal arts college, choose "Bachelor's-granting institutions"). Finally, choose the broad category of the doctorate, if you'd like, or even the specific program.  Note that the filters cascade: If you choose "Life Sciences" under Broad Category, you won't be able to find "Economics" under Specific Program. On the second view (using tabs acr...

We tend to think a lot about a college's average test scores, despite the many ways colleges can and do manipulate them for their own benefit.  After my last post on the relatively low number of students who enroll in the most selective institutions, someone asked if I could do the same for test scores.  So here they are. I've calculated very close mean ACT Composite and SAT CR+M means by taking the midpoint of the 25th and 75th percentiles.  They're almost certainly not perfectly accurate, but are very close, in all probability.  Then I've broken up enrollment to show where students attend college. The first view is based on the earlier visualization; the second is a scatter showing both the ACT and SAT averages.  The first has just three filters; the second has more, plus a "Color By" parameter that allows you to color the colleges by one of several factors. I hope this helps people think about and put score ranges in some context. (Note: IPEDS does ...

One of the very first posts I did on this blog was showing just how many "Uber Selective" colleges and universities there are (or aren't), and how many students they enrolled (or didn't.) I used it last week at a presentation at NACAC , and several people asked me if I had an update on it, so as soon as I got home, I pulled down the data and started visualizing it.  It's below, and it should be self-explanatory: Of the 1,943 four-year institutions shown, only 18 admit less than 13% of freshman applicants.  These institutions (blue bars) enroll just 82,000 students (under 15,000 of whom are African-American, Hispanic, or Native American), and only about 18,000 freshmen.  Yet they get a relatively large share of the press and attention whenever the discussion turns to college admission. This has limited interactivity: You can choose region, public or private, or Carnegie group. And of most importance: This is but a sliver of American higher education; for ins...

One of the things I hope to show people on this blog is that data is a lot more fun and interesting when you actually do something with it , rather than just present it in a spreadsheet. Here's a good example. This week, over 6,000 people who work in or around college admissions will converge on Columbus, Ohio for the NACAC Conference .  (Yes, Oktoberfest is also in Columbus this weekend , and based on my informal discussions, there may be some overlap.)  NACAC puts its attendees in a table on its website for anyone to use. But it's just data: What does a simple spreadsheet have the power to tell us?  Maybe more than you think.  Yesterday, I put the information in a visualization (first page is set up for mobile but autosized) designed to help people find other attendees.  As a side effort, I put up a chart of the most common first names of attendees, and it proved to be very popular. So last night I did a little more, and looked at most common first, and l...

Whoa, you might say as you look at this. It's way too funky for me. That's OK; I'm going to show you a new feature in the data visualization tool, Tableau , that I use that will make this all make sense. Hang on. I wondered: Do states with higher median income levels charge more for tuition?  So I began to explore. On each dashboard, median family income is displayed on the top chart, and college tuition on the bottom.  The view starts with four-year publics, but you can change it using the filter. The first dashboard shows only the rank of the states, from 1 to 5, with 1 being the high value in each. If you can't make sense of it, don't worry: Use the little box in the upper right hand corner to select any single state, and that state's data will be instantly highlighted on both the income and the tuition chart.  You can see where a state stands on both measures. The second dashboard (using the tabs across the top) shows the actual inflation-adjusted va...

A note in response to some questions from IPEDS geeks and others:  My data selection was from 2014 IPEDS data.  I used Title IV participating, US only, all sectors except administrative units.  That resulted in 7,018 institutions.  My visualization shows 6,876 because there were 142 institutions with absolutely no data reported.  I should have defined in my original post. Also, the selectivity bands are not defined: Cut points are at less than 15%,, 25%, 40%, 60%, and 75%.  All others are "Not selective/Open." College. University.  We think we know what these terms mean, and yet, any discussion of colleges in the US invariably leads to someone saying, "It depends on what you mean by college." For instance, there are about 6,900 post-secondary institutions in the US, but only 2,654 offer a bachelor's degree; they enroll 10.5 million of the 17.6 million undergraduates. Of all the institutions in the US, only 293 enroll at least 15,000 undergradua...

Note: The visualizations are not optimized for mobile.  A desktop is recommended for best viewing. From the annual College Board Trends in College Pricing comes some interesting data, which I've combined into one database for visualization, focusing on public university tuition for residents and non-residents.  This looks complex, but it's pretty simple. The opening view shows six charts: 2015 tuition for residents; for non-residents; and the premium a non-resident pays (in sticker price) across the top.  On bottom are three scatters: Resident tuition as a function of state funding per FTE student; five-year, inflation adjusted tuition for residents and not residents; and funding per $1000 of personal income and resident tuition.  Of these, I think the middle is the most compelling: Note the states that have raised tuition faster for residents than for non-residents. The chart starts with US Averages in red, against the states as gray.  Use the control in...

Note: Visualizations are not mobile friendly.  I recommend a laptop or desktop for viewing this site. Someone asked me today about what I thought higher education's biggest challenge was, and I said college costs without thinking.  And a few hours later, I still think that, with a twist: College costs for low-income students, especially at public institutions who presumably have a primary mission of educating students of all income levels in their state. To be sure, costs are too high at private institutions, and many of the trends you'll see here are carried over and amplified in the private sector; but private colleges and universities may exist for different reasons, and that can be hard to capture in a visualization like this. There are two views here, using the tabs across the top.  The first is a scattergram, arraying almost all 660 US, four-year public colleges and universities that admit freshmen (a few are missing data).  The x-axis shows in-state tui...

This is a good example, I think, of how data visualization helps you make sense of things: Even simple things like a small table of data. In this case, the table is from The College Board, showing changes in the percentage of in-state freshmen in our nation's public universities.  You can see the raw data by downloading Table 28, here . What you can't see by looking at that table, of course, is the overall pattern.  That's where a picture comes in. T here are only tw o numeric values in the table: Percentage of freshman enrollment that are state residents in 2002 and 2012.  I added a third, by subtracting one from the other.  Then I put them on a   choropleth hex map, a format I like because all the states are the same size.  On this map, orange colors show states where the percentage of in-state residents has increased; purple shows a decrease, and grays are mostly even. Be careful about  interpreting  this data. This  visualization ...

Note: I tweeted a link that was set up for mobile, and thus the visualization scrunched down to almost nothing.  If the URL has m=1 at the end, just delete it, or click on the title above to go to the desktop/iPad version. The College Board just published long-awaited concordance tables to compare new SAT scores to old, and new SAT scores to ACT. You can download the data here if you wish , or look at them visually below.  The tables in the data correspond to the tables on the visualization (that is, for instance, that Table 7 in the College Board worksheet can be viewed on Dashboard 7 here, using the tabs across the top.) For convenience, Old SAT scores are always in light gray.  Notice also I've labeled the chart when the axes are not synchronized. As this data is public, I have cited the original source, its purpose is educational, and this blog is not monetized, I believe the use of it in this format falls under Fair Use . As always, hover over the dots for ...