Hypothesis Testing in 17 Seconds

[@jbstatistics]

Thanks! I must admit that I truly enjoyed making and watching this one. But I had absolutely no idea what others would think of it, so I'm glad at least one person likes it!

[@jbstatistics]

I'm glad I could help out a bit. It might sound funny, but there's a couple of levels to this video, and I spent some time thinking it out. I chose a skewed distribution for one of the distributions precisely so that people could see that there are different distributions (not all test statistics have a normal distribution). I'm glad that it helped you a little bit in this way. Cheers.

[@jbstatistics]

Right now I'm just posting to this channel and don't have a website. But I may set something up in the not too distant future. (Mainly so these are all a little more organized.) I do try to be concise, and get to the point while still being true to statistics, so your comment makes me happy. Thanks!

[@jbstatistics]

Different test statistics have different sampling distributions. Many test statistics have sampling distributions that are approximately normal (under the null hypothesis), but there are many other distributions (such as chi-square, F, and t, to name a few). When testing the null hypothesis of equality of variances, we often use the ratio of sample variances as the test statistic. (continued in other response)

[@jbstatistics]

(continued from other response) Under the null hypothesis of equal population variances (and normally distributed populations), this statistic has an F distribution. The F distribution has some right skewness. The second distribution given in the video is an F distribution. Yes, we most definitely take the shape of the distribution into account when constructing rejection regions or calculating p-values.

[@MrAntihumanism]

i love this, you're fucking brilliant. this simplifies a somewhat misunderstood aspect of hypothesis testing that just goes right by kids.

[@52hands]

Thanks a lot for the explanation! I have only encountered symmetric distributions before, hence my confusion. But it's clear(er) now. :)

[@MrAntihumanism]

I think an instructor could base the entire final revision (certainly useful at my institution, ivy league kids who dont attend except the final few classes) based off your lectures. Thats how good they are.

[@MrAntihumanism]

Its great, do you have a website or just a youtube channel? I know for certain undergrads would find this useful, more so than other channels because you get to the point without missing any important information. This makes for a great final recap.

[@junal27]

excelent set of videos, this one is excellen too, you are helping me a lot with my statistics inference college course

[@paposbstn]

can you please describe what happened in the video, Please I really need this for my project.

[@52hands]

The second curve is very skewed. Does the formula for a hypothesis test for such a curve look different than for a sample that is assumed to be standard normal distribution? Does one have to take into account the shape of the distribution when accepting or rejecting the hypothesis? I would assume that a value of 0.1 means different things for samples with different distributions. Is this correct?

[@manavsainii_]

It follows the f distribution hence the skewness.

[@johnnyroc]

Why don't they ever label the axis. I'm guessing the horizontal axis is observations and the vertical axis it probability going from 0 to 1. I sort of get it vaguely but still super confused. Back to the drawing board. I hate my econometrics class so much. Just 6 more days of torture.