Gateway DFWs by Department: Five-Year Trend

Developed by Ken Wendeln, Kelly School of Business
Submitted by Steve Graunke, Assistant Director for Survey Research and Standard Reports
Indiana University-Purdue University-Indianapolis
 
The audience for the presentation includes university administrators, deans, and Gateway course coordinators and instructors who all have a stake in improving retention and graduation. Gateway course DFW grades have a strong correlation to first--year retention and, in turn, to 6--year graduation rates at IUPUI. The purpose of this slide is to get the audience’s attention and curiosity to dig deeper into the detailed DFW trend data provided in tabular form for each of the designated Gateway courses. Administrators and deans quickly get the implications of the ‘Cost of DFWs’ to students for grades that have no academic value as well as the loss of tuition for non--retained students.
 
That leads to the ‘spider chart’ that shows the ranked DFW% (and actual numbers) for each department for AY07 (combined fall’06 and spring’07) in blue. The red line and markers shows the DFW percentage improvements for AY12. The yellow markers highlight where improvements have NOT been made. The ‘spider’ was chosen for its visual interest, impactful but simple rankings and comparisons, and unencumbered space for the labels and supporting data.
 
The goal is to encourage ac1on by all the course coordinators and instructors and support from the deans and administrators. The ‘exemplars’ (orange symbols for Psychology and Learning Communities) demonstrate that specific actions supported with data can lead to significant and sustained course DFW improvements. The result of improving DFWs at IUPUI has been a 16 percentage--point improvement in fall--to--fall retention and a 12 percentage--point improvement in 6--year graduation rates.
 

 

 AIR Comment Board

 
To add a comment, Sign In
Total Comments: 9
 
Leslie posted on 6/10/2013 10:10 AM
Thank You for Sharing
Nicole posted on 6/18/2013 8:48 AM
I just saw this chart used in another presentation and I really like it. This is a great way to show improvements (or lack of) in data points.
Liz posted on 6/26/2013 10:02 AM
Clarity: I really like the spider chart and I think this is a great use of this chart. This isn't a standard go-to chart, but I think you have done a great job finding something that works with your data.

I had a couple of quick comments related to just looking at the chart as a stand-alone. It is not clear what the yellow symbols are from the chart, though this is in the text. If these are exemplars, I think the same symbol is sufficient, otherwise it adds another layer of visual change that seems to be unnecessary. I also at first thought the red line was DWF rate in F11/12, but the text says it is the DWF percentage improvement. I'm assuming this means it isn't the percent improvement, but am somewhat unclear.

Accuracy: This is a nice strong clear presentation. It is easy to see the line trends and the labeling helps.

Fit and Appropriate Display: I think this is a great way to display these data and will be trying this myself for our DWF rates.

Aesthetic Appeal: I have a just a couple of suggestions on tweaking the aesthetics – see what you think. I would spend a little more ink on the spider chart and make the title a smaller font. The large title seems to dominate the page. I would also try to consolidate the left and right labeling – the change in DWF and AY semesters & DWF. You would probably bring these together. Perhaps you could tie in “by department and class” into the title and drop this label. Not that the chart needs to be a great deal bigger, but it may help to consolidate some of these bits.

Thanks for sharing your chart!
Randy posted on 7/11/2013 1:40 PM
I really want to like this chart because it is different than the same old bar chart and because it is about a topic that I've researched and worked on for year. This is a very important metric for colleges to track.

But I wonder if others find that it really works for these data. When I see a circular chart I'm automatically looking for the relationships between the elements and why they appear where they do on the chart. I can't find the logic in the order of the elements (why does COMM 635 fall between PHIL 247 and Soc 365?)

I may just be missing the point!

Also, Mary Ann Coughlin has been trying to teach me for years that two points does not make a trend! Is this really a trend report (see title) or is it a longitudinal look at two points in time?

Again, I really want to love this chart. It did catch my attention and grab me to look at it in depth.
Irene posted on 7/11/2013 6:12 PM
This chart immediately grabbed me because it is so different and evocative of melting ice sheets in Antarctica, the shrinkage being a good thing in this case! So I was interested in digging deeper. However, it did take me a moment to figure out what was going on.

Nice work in trying something different!
Eric posted on 7/15/2013 12:34 PM
Steve, thank you for submitting this! I have never used this type of graph, but it certainly provides us with another way to view the data.
Maria posted on 8/15/2013 12:59 PM
This is maybe the best example of the spider chart I have ever seen. Usually this kind of charts are not very evocative. I liked the comment about the "melting ice in Arctica" - yes, I had this association too, and it's very appropriate. I so like thos chart that I will use this idea, thank you for sharing! One critisism: I still don't know what this chart is about, because I can find nowhere what the DFW is.
Gary posted on 9/3/2013 5:44 PM
I don't know whether to consider this plot a success or not because I don't know the creator's full intentions, or the context that it was meant to be effective within.
Pros: Looks different and may therefore be more effective in getting readers to consider the data.
Cons: Polar coordinates make it really difficult to read accurately and make comparisons along the y axis. The use of a line suggests a trend, and is dissonant when combined with a categorical variable like class. Finally, the sagging gridlines meant to evoke a spider web are a bit too cute of an embellishment for my taste, and might make it difficult for some readers to treat as a serious informational graphic.
I think Ken probably knew his audience well enough to decide that the pros outweighed the cons. But there are other creative and informative ways to build upon traditional graphics that distort less - for example, finding a way to scale the plot elements to N's, which might explain some of the huge and tiny changes between the two times being compared.

Thanks for sharing!
Pam posted on 9/16/2013 1:30 PM
I agree with Randy's note and I think I have a way to improve this visual with respect to the organization of the courses. If the courses were organized by area and the area indicated in some color scheme or label further interpretation would be possible. For example, sociology, psychology, political science, and anthropology are all social science classes so placing them next to each other with a label around the outside that says Social Science.

I also don't think the dots should be connected. These are separate DFW rates by class so they are not a trend line, but separate data points. Perhaps connecting the dots between the 06-07 data and the 11-12 data with different colors indicating an increase or decrease in DFW would be a more useful way to show the rates. I think the yellow does are trying to indicate increases in DFW rates, but that' isn't labeled.

Recently, I produced a list of DFW's for classes in the past 3 years. I calculated the frequency and percentage of DFW's out of total enrollments in each class. Then I used conditional formatting to highlight higher DFW frequency and higher percentage. The list could be sorted by both frequency and percent column and the conditional formatting helped the user visually see the frequency and percent of DFW for each class. My colleagues responsible for tutoring are now using this list to target tutoring to classes with high DFW frequencies and percentages. While this is a cool representation of data, sometimes simple can also be useful.