New Freshman Cohorts

By Steven White, Data Research Analyst, North Carolina State University

This visual display shows how students’ SAT scores and first-term GPAs relate to graduation success. Each graph displays three dimensions of data in a two-dimensional space. This display is not intended for the casual observer; rather, it is intended to facilitate complex discussions about enrollment planning and student success.   

Utilizing Tableau software, the visual display compares 10 years of freshmen cohorts by several different SAT or first-term GPA bands, which were then overlapped with the new freshman six-year graduation rate.  



To add a comment, Sign In
Total Comments: 7
Tim posted on 6/19/2014 12:58 PM
When I first looked at the visual display, though remarkably organized and engaging, I knew that it was going to be complex and take some time to learn how to read/interpret before discussions could be made. And in the first paragraph, you confirmed those thoughts, which is very good. This visual display reinforces the complexity of understanding how pre-college factors may be related to graduation. I think the choice of colors was a good way to the similarities in the displays, though keeping them separate enough to mark GPA and SAT. I was able to interpret the displays in a general way fairly quickly.
Jeff posted on 6/19/2014 1:06 PM
These are very ambitious graphs and try to tell multiple stories at once. It is possible that a simpler design could more effectively provide insights useful to just the "casual observer." So what are the punch lines that the reader should be learning from this information?

The first graph illustrates how dramatically graduation rates rise as a function of first term GPA.

The second graph reveals that graduation rates rise as a function of SAT, but much less dramatically than as a function of first term GPA.

The variations by cohort do not obviously reveal a pattern, so the insight there is there does not appear to be improvement across cohorts.

As for the bar charts themselves, these are meant to show the mix of students by GPA or SAT. Perhaps, a stacked bar chart would work better there. There does not appear to be a relationship between mix and graduation rate, so these analyses could be performed separately.
Cliff posted on 6/19/2014 1:48 PM
I would put good money on the table that, if you added a simple bi-modal set of tables by number of additive credits earned 12 months after the first date of enrollment, with the dividing line set at 20, you would have a more convincing set of relationships---and a lot more bars.
Betsy posted on 6/19/2014 2:02 PM
Yes, it's definitely complex, as well as attractive, but I'm having trouble understanding the 3-position lines toward the top of each bar chart as well as the left and right axes. Years, scores and 6-year graduation rates are all quite clear.

But where do I slide along the scale and find the place where I'm looking at 0.2% of the cohorts? I'm guessing it should be 20%, but either way, I'm confused at what I'm looking at. Does the 2012 cohort scoring 1400 and above comprise 30+% of the graduates and maybe 9% of the whole freshmen cohort?

I still have no idea with the three-point line might represent, though.
Terry posted on 6/20/2014 9:56 AM
These charts point out one of our biggest challenges, we have mountains of data and how do you relevantly present it? As Steven points out the charts are dense but with just a moment more of study, the meaning begins to stand out.
Meg posted on 6/27/2014 4:21 PM
Good display! I am definitely going to create this for our Enrollment Management team, too!
Sam posted on 7/29/2014 10:24 AM

Good stuff. Just a few things, some of which have been mentioned already:
--Multiply your dual axises (% of Cohort and % Grad) by 100 to get a natural percentage.
--Leave out the cohorts which have not aged into the 6 year graduation rate.
--Consider adding lines for the 4 year rate since that is now becoming (again) a more scrutinized metric of college performance
--Explain in a footnote how a GPA can be Null if it can also be 0
--Attempt to standardize the colors so that at least lowest and highest values of GPA have the same color as the lowest and highest of SAT. Not easily done when there are different numbers of categories. This could also be accomplished with losing the colors and going for shades of grey or shades of 1 particular color.
--Add the Totals which conforms to IPEDS or other external reporting standards. Often executives like to see that a deeper analysis conforms to top level rates/percentages when rolled up (particularly because they typically only remember these and must respond to their higher ups about them)