In this video, I lay out some properties of the Normal distribution that are essential to understanding basic hypothesis tests. I hope you enjoy it.
It’s taken 10 months, but now I think most of the prerequisite material to understand the basic hypothesis tests. I’ve had a script on p-values, Type-I error and power gathering dust in my materials, all because I realized there was too much I hadn’t covered and that bloated the script.
Balancing Youtube & Ph.D Research
One interesting aspect of running Very Normal is that it’s made me realize the bottleneck of my production is not me, but actually my computer. I am currently in the process of doing heavy simulation work to revise a manuscript. Not only that, but this revision has an impending deadline: this coming January. Either I get these revisions done, or I lose my chance to submit to a major journal. But the fact is, I cannot edit videos and run these simulations at the same time.
So this has put me at an interesting impasse: do I give up on submitting to a major journal to progress the channel? Or do I upload at an even slower rate?
After a week of working like this, I realized that I didn’t have to think this way. I could have it both ways. But this required some innovation in how the simulations were run on my computer. My simulation code had to be adapted to let me “pause” for a few hours to let me edit. Then, when I don’t want to edit anymore, I can restart the simulations back where they were. It’s more overhead for me to keep track of, but I can get more done.
If this was an obvious solution to you, I’d love to pick your brain on how you’d handle this type of challenge. Unlike with statistics, my coding skill only improves when I encounter a new problem that I need to overcome. I don’t have much “theoretical” knowledge of efficient code, if that makes sense.
What am I working on right now?
At one point in the future, I’d like to analyze my revenue data, probably in about 6 months. Not only is this a popular topic on Youtube, I think it would be good for motivating other educational channels to take that leap.
Many statistical analyses require an “independent and identically distributed” data assumption. One problem is that ad revenue data most certainly does not satisfy this assumption. I try to change at least one aspect of my video-making process after I upload, so the process for making them is not the same each time. Furthermore, it’ll be impossible to know what audience sees the videos, what ads are served and what else is going on on Youtube at the time.
That being said, I still think it would be a good case study for anyone trying to approach analyzing their own data.
That video will come in a few months when I have more data, but for now, I’m working on another project concerning the paper I found after the 8 statistical innovations video. I couldn’t get it out of my head, so I’m excited to dive into that paper for y’all.
What am I enjoying right now?
Audiobook: The Courage to Be Disliked by Fumitake Koga and Ichiro Kishimi
Series: Doona on Netflix (kinda?)