Thursday, January 28, 2016

Using My Own Data in Class #339

Preparing class notes is part of every Lecturer's job. Some lecturers will use notes created by others, some will use lecture notes provided by text book authors, while other will create their own. My preference is to create my own - I find I understand the subject better, and consequently am in a better position to teach it.

I use a lot of material from textbooks and examples form the web. Creating notes for students has got easier with online resources. So - if I want to show an example of a chart, it will be easy to find a suitable example online (citing source of course!). Not being an active researcher, I have very little of my own data to use in notes - YouTube Analytics from my channel is about the best that I can do. However, last evening I introduced the topic of "Multivariate Analysis", and while preparing the notes I remembered that I did a lot of this type of analysis for my PhD between 1985 and 1987. There is no soft copy of my PhD, and only three hard copies were ever printed (I have one, the Library in Trinity has one, and God knows where the other one is!). There is also a microfiche copy in the Trinity Library. So I decided to use a chart from my PhD to illustrate the results of a multivariate analysis - here it is (scanned from my hard copy):

This shows three populations of Calliostoma zizyphinum (a marine snail) from Scanlan's Island (Co Clare), Rosmoney Bay (Co Mayo), and Audley's Castle (Co Down). I took ten shell measurements for loads of shells at each site and fed the results into a computer which generated the above printout, and it shows that the three populations can easily be separated based on their measurements. Shells to the left of the diagram are a lot narrower that those on the right. I got a bit of a buzz in using my own 30 year old PhD data in class for the first time, and explaining to students what it means.

My PhD was submitted to Trinity in January 1988 as a printout only - no online or disk submissions in those days. Inside the back cover I included a printout of all my data, which I have since scanned into a file. Now - do I dare carry out the multivariate analysis I did in 1987 on the same data again? It would be interesting to see if modern tools would show the same result. What would I do if they didn't? 

No comments:

Post a Comment