Tuesday, April 09, 2019

New Video: How To... Calculate Spearman's Rank Correlation Coefficient (By Hand)

Last December I got a message on my YouTube channel as follows: "wrong formula! rookie mistake". Naturally I was concerned about this - I often get messages claiming that there are errors on my videos. The video was about how to calculate Spearman's Rank Correlation Coefficient by hand - it is a relatively easy calculation to do and is based on the difference in the rankings of two variables. However, the method I demonstrated had what are called "ties" (two or more values which are the same) - the formula that I used in the video should not be used when there are ties. A rookie mistake indeed! The video was only published in June 2018, but it proved popular as it racked up just over 16,000 views in less than 6 months - very good for a new video. I immediately took the video down from my channel - I did this by making it "Private" so that the 16,000 views would still count in my overall figures. This is cheating of course, but hey - they were still views!

My New Home Studio.
Yesterday I published an updated version - this time with no ties. Some data analysts/statisticians say that if there are only a few ties in a large data set that it does not matter, but the purists say to use a different formula - Kendall's Tau is the recommended one. I hope to do a video of this soon. 

The long delay in replacing the incorrect video was partly due to time pressure, and partly due to me wanting to experiment with my new GoPro camera. I set it up like a document reader and after many trials, I finally got the settings right. I also learned how to control this with my iPad - great for getting the size and settings such as resolution right. I could also start/pause/stop using the iPad. I feel that the quality was really excellent - the sound and video were fine. The GoPro picked up my voice OK, and it was able to focus on the worksheet no problem. I don't have sound proofing in my home office, which could be a problem in future. The GoPro microphone is so good that it is picking up noise from other rooms and from the road outside my house. I have to make recordings during quiet time.

So - finally corrected, here's my new video on how to calculate Spearman's Correlation Coefficient in data that does not have ties...

No comments:

Post a comment