Does Sample Size Actually Matter?
Audience: high-schoolundergraduate
Tags: bayes-theoremprobabilitystatistics
An urn contains 3 red balls and 3 blue ones. Imagine selecting a random ball from the urn and permanently removing it without revealing its color. The video discusses a neat little probability puzzle going over how we can use the 5 remaining balls in the urn to determine the color of the removed ball. On the way, we'll discuss Bayes' Theorem, the usefulness of sample size in tandem with signal strength, and how misleading relative probabilities can be.
Analytics
Comments
Slightly longer pauses to allow the concepts sink in. A longer exploration of the signal strength vs sample size relation. I thought that was a really interesting takeaway, but you kind of sped through it.
I had never understood how to quantitatively determine the strength of evidence. This video is very well created and easy to follow.
A nice problem exploring some fundamental statistics and probability results. Although it is important to note that sample size does I fact matter.
Great visuals and pacing, and a friendly narration!
Really good topic and well explained! Genuinely think this is very good at what it does, I would have personally liked a bit more explanation on signal strength, but that’s personal preference here. Main reason I haven’t rated it higher is that there’s nothing particularly “wow”ing me; but it is just generally very good!!
I thought every section was useful and added value to the explanation.
Very good video! The problem was well introduced, with slick animations accompanying the explanation. I did not need to look up or research anything beyond the video to understand it - it was a self-contained explainer.
Audio and video quality are solid, 1080p60, with a crisp, easily understood voice-over.
I also appreciate the YouTube chapter markers, very useful for re-watching.
I actually learned about the concept of “signal strength” thanks to this video. I knew about the concept of signal-to-noise ratio in the context of audio processing, but didn’t know it was used outside of that. Well done!
This video is outstanding in all fronts. It is well motivated, clear, and delves deeper into a very common example in statistics and introduces a much needed nuance into that example. Well done.
About as good as SoME videos I’ve seen previously, which is all the past winners ;) if I’d have to bet on a winner, this could be a top contender. Nothing to improve really, hope you can find similarly weird+specific phenomena to cover.
Hah! I saw this one before joining the voting. It was a good vid and I watched it through all the way!
Very good explanation, and the animations are very good. The video is very easy to follow but it requires some knowledge about Bayes Theorem and what normal distribution is, which high school student might not know about, so maybe spending some time to explain those, or refer to other videos in which those are explained would be great.
Nice animations and clear presentation. Subscription recommended. The topic might be a bit too advanced for high school students; perhaps it is more suitable for university.
This is a really interesting introduction to signal strength which isn’t talked about as much in introductory statistics courses. It’s a bit hard to follow without getting out pen and paper but this is really beautifully animated and presented!
The explanation is clear and the presentation is great. However the scope seems rather narrow and focused and can be improved in the innovative aspect. It would be lovely if you could draw more connections to the real world applications since the idea itself is more on the practical side.
Very intuitive explanation. It would have been nice to have a connection to other measures of information.
Perfect math explainer with wonderful visuals and voiceover. The intro, motivation and problem laid out at the beginning was understandable and interesting. The way the problem has been adressed, the questions that have been asked and the invitation to think about the problem are pedagogically very valuable. The presentation of the solution was also clear and entertaining. It’s a simple problem with an interesting and unintuitive yet simple result. There’s nothing for me to complain about.
You might want to spend a bit more time on the equations before changing them around, so students can have the time to read through them and recognize each term (the sequence at 7:35 is very intimidating) Framing the situation like an adversarial game can also be slightly confusing, especially since you forgo the “guess as quickly as possible” issue right after we played the game. Everything else (the animation, color coding the probabilities…) is very clear!
This was great! I don’t know much about probability, so this was a tough puzzle fir me. But you broke it down in a super clear way.
Very good animations and I liked the fact that you suggested the viewer to pause and solve the problem on their own (which I didn’t). It also taught me about signal strength, which is something that I haven’t thought about so far and I was usually concerned with sample size only. In addition to that the video was well animated and easy to follow.
This video serves a great explanation to the Bayes’ Theorem discussing with an example to understand the crux of probability in general.
It would have been nice to have examples of what the signal strength would look like with more possible outcomes but overall a great video
The concept, flow of thoughts and graphics, all are in sync.
It was a great video. My one nitpick is that the intro made it seem like you would answer the question, “how many balls do you have to draw to be confident in your answer to the game?”, but that was not actually answered.
I’m caught between: presentation, clarity and a-ha moment were fantastic, and the fact that after 1 watch I couldn’t tell you exactly what I learned and how it works. For me, it was so fast paced, I had to keep pausing and/or rewinding to digest what was said. The a-ha moment at 10:00 that sample size didn’t matter finally landed as “ok I get this conclusion, and I finally get the point of the video”. But then I lost the plot on why sample size does matter again - I guess because a higher sample size changes the required signal you need to have the same amount of confidence? So now my brain is confused that a low sample size is better, because a lower difference is required to have good signal.
I kinda with the intro was more like “I draw many balls. You draw 6. We come to the same conclusion, but who is more justified in their answer? Of course me, right? I did all this legwork collecting samples? Turns out it doesn’t matter. Let’s find out why”. Then I would have at least seen where the video was heading towards.
A few points, some repetition to restate the conclusion we just reached would have been very helpful. Finally at the end - what exactly did we learn and why does it matter in my life (eg. when I go on to collect data for my next survey)?