College of Staten Island Student Satisfaction Rating

By Sam Michalowski, Director, Office of Institutional Research and Assessment, College of Staten Island/CUNY  

The following data visualization is an initial foray into trend analysis of student satisfaction at the College of Staten Island. Our IR office conducts the Noel-Levitz Student Satisfaction Inventory (NLSSI) every two years in the spring semester in order to meet a university performance management goal of measuring and improving student satisfaction with academics and non-academic services. We sought a representative sample of undergraduates by stratifying and randomly sampling course sections by course level and department. Through outreach from the Provost’s office, a web-based registration system for those whose courses were selected, and an in-class collection method, we obtained a 10% sample in both collections thus far (Spring 2011, Spring 2013). Additional years worth of data will help confirm the trends we see across these two years. While largely representative of the demographic distribution of the undergraduate population, we intend to eventually weight the data (i.e., non-weighted data is presented here). 

This discrete area bar chart display was created in Tableau. It displays the percentage of responses in each of the seven satisfaction categories (from Very Satisfied to Not Satisfied at all) for the item “The quality of instruction I receive in most of my classes is excellent.” The distribution is broken out by collection year, gender, and class level. For clarity, satisfaction levels are portrayed using “stoplight” colors with Neutral represented by cool blue. Shades of green denote some degree of satisfaction and shades of yellow through red equal some level of dissatisfaction. Thus, one can quickly ascertain trends in general and specific satisfaction levels across the two time periods. For example: Among freshmen, female general satisfaction with quality of instruction has decreased, while that of males has increased; general dissatisfaction with this item has increased slightly for juniors and seniors; sophomores appear to take a more neutral stance on this item; etc. 

The use of interactive dashboard software streamlines the data analysis process because a client no longer has to sift through dozens of charts in an immense data report. S/he can instead create and print PDF displays for questions s/he wants to focus on. The visual “pipes” into the title (the underlined text) whichever satisfaction item is selected in the drop down menu below the color legend. We are working on developing a suite of interactive dashboards of NLSSI data including this one for relevant clients and distributing them either by Tableau Reader or through Tableau Public.  




To add a comment, Sign In
Total Comments: 9
JR posted on 7/17/2014 10:22 AM
What a great way to show so many sub-categorized trends compactly and with easy readability! This would have to be a very useful visual and easily understood information for upper administration as well as faculty and department chairs.
Tim posted on 7/17/2014 10:34 AM
I think this visualization does what you want it to. The way the formatting is set up (with the class rank as the column headings and gender as the row headings) emphasizes the distinction between each class year. The only facet that might be interesting to identify is that the sample of freshmen from 2011 should be representative and comparable to the juniors of 2013; along with sophomores of 2011 and seniors of 2013. I'm not familiar with Tableau, but that might be something to point out in your descriptions, or accompanying visualizations. Overall, this shows how these groups of students' satisfaction has changed over time.
Terry posted on 7/17/2014 11:12 AM
I think what strikes me the most is the total look of the charts. They all flow together and present a very nice visual effect.
Jeff posted on 7/17/2014 1:25 PM
These results vary a surprising amount. The graphs help to illustrate that, but it also raises questions of statistical significance. At the very least, what would help in interpreting this data would be to somehow show the sample sizes.

Another suggestion would be to compute an average score for each data point -- weighting each satisfaction level from 1 to 7 (or 0 to 6). You could then even compute T-tests on this average to begin to interpret whether the changes are statistically significant.

The basic problem is that 2 data points don't make a trend. This graphic approach to survey analysis will become more meaningful once you do your next survey and have a third data point that will help to visually illustrate the noise in the data.
Robert posted on 7/18/2014 4:24 PM
This is an interesting display. General male satisfaction has grown or remained stable through each of the class levels. Whereas, the opposite is true of females with the senior class being the most stable. Just as interesting for me is the class level progression trends. For example, the Freshmen from 2011 to Juniors in 2013 decrease for females in a meaningful downturn, but female Sophomores from 2011 to 2013 Seniors increased in general satisfaction. The male Freshmen and Sophomores from 2011-2013 as Juniors and Seniors respectively both increased in general satisfaction. For me, the trend in class across time is as important as the trend within class across time. These charts are great for examining the the same class level across time, and could have some potential for looking at class level progress across time. Good job.
Marc posted on 7/22/2014 3:48 PM
We do the NL survey every Fall. I am absolutely stealing this, Sam.
Sam posted on 7/24/2014 1:51 PM
Go right ahead Marc!
Liz posted on 7/30/2014 9:55 AM
I really like this - it is a great way to present a lot of data at once and transform it into information, don't you think? I think these are good examples of how pulling back from the actual numbers and seeing the forest, not the trees, helps to show the insight. I've passed this on to my group also. THANKS!
Betsy posted on 1/5/2016 10:22 AM
Nice work, Sam. The color scheme works well to help reduce the confusion with a 7-point scale. It also leads to interesting questions about specific sub-groups, assuming there are enough of them: e.g., Why greater extremes for Junior males, or the increasing "meh" for Freshmen females when Freshmen males are showing improved satisfaction?