Love your work

Epic explanation ngl.

sir can you make a video on how to arrange data on excel for 2 factor rcbd for analysis on r studio.

I do this in the same way, Sir

Can you spell that package for me..

Thankyou

Thank you so much!

Thanks a lot!! I was really afraid starting with R but your video helped me see it clearer

What is the reason for using the "small y" as the "number of success" of success when we are already using "capital Y" as the script for the "random variable". Is it not confusing. What can go wrong if we use some other script.

useful - thanks

What if the groups are the same. Meaning, you got two binomial distributions from two models based on the same data that might be in two different representations. In that case the distributions to be compared are correlated. I know DeLongs method for example does a correction in terms of AUROC, is there something similar for binomial distributions? Thanks

Excellent video! Thank you!

Hello, thanks for the video. I was just wondering if replication would increase accuracy, precision or both. In control charts it's possible to distinguish systematic error from random error. I was trying to understand if replication helps to solve for systematic error in design of experiments.

given: Y~U(0, 1) ==> F_Y(y) = y @ [0, 1] and F_X( x ) = g( x ) @ [0, 1] assume: X=g^-1( Y ) @ [0, 1] F_X( x )=P( X < x ) = P( g^-1(Y) < x ) = P( Y < g(x) ) the rightmost equality is justified since g(X) is a distribution function, i.e. monotonic and non-decreasing ==> F_X( x )=P( Y<g(x) ) = F_Y( g( x )) = g( x ) since F_Y( y ) = y @ [0,1]

10:52 The comparative PMF plot shows Bin(n = 5, p = 0.01)and Po(lambda= 0.05).

Are neural networks like CNNs hierarchical models?

This is an awesome lecture. Thank you

😂

9:03 subtitle says "posterior for the mean".

2:51 Title says: Gamma CDF

Trying to summarize my understanding. P-value is the probability of getting the sampled data (and more extreme data) given it was sampled from null hypothesis H0 distribution. The alternative (Ha) distribution does *not* play any role in calculating the p-value. Next, the author simulates a quantity that resembles the point probability, P(H0 is true given |y-P95| < epsilon), where epsilon is some arbitrary nonzero constant that defines the neighborhood of P95 percentile of H0 distribution. This quantity *does* depend on the alternative Ha distribution (12:25 Bayes theorem). This is done by randomly picking H0 P95 percentile samples from either H0 or Ha distributions (according to prescribed relative frequencies of H0 and Ha) and calculating P(H0 is true given |y-P95| < epsilon) . P(H0 is true given |y-P95| < epsilon) increases: a. as P(Ha) decreases b. as |E(Ha)-E(H0)| increases

1:36 Normal-->Binomial

Thank you that was very useful!

😅😅😅😅😅😅

Hello! Thank you for the video! What if I have two vertical lines in my scatterplot when I do the linearity check? IV is gender and DV is continuous ... Looks weird :)

Hello, I'm trying to figure out if i can use LPM if I have a binary outcome and a binary explanatory variable (my # obs is less than 5000). I'd be really thankful if you could help me please. Also, i'm thinking about recursive bivariate probit, but i don't know which one is more adequate

Thank you very much! This was very helpful :)

Great Vid!

I have been searching online and my few nonparametric textbooks but cannot find and answer. In the social science literature, we see a pattern of conflicting results. There are 101 articles that conclude one point, while 175 articles conclude the exact opposite. I have conceptualized 101 and 175 as n’s for two different distributions. The null hypothesis might be “there is no size difference between the n’s.” The n’s are frequencies of two opposite results. Do you know of any nonparametric statistic that can assess this null hypothesis? It sounds as if some kind of binomial test would be appropriate. I have done any statistics for about 30 years. Do you have any videos addressing this issue?

nice video， thx！！！

Saved your STAT 587 series, thanks for sharing.

I am a little confused about your explanation after 10:20. Could you please provide additional resources and pointers to understand this better?

At 03:52, could one more reason for incompatibility be that data could have a different variance (irrespective of constancy of the variance)?

Your second statement (08:15), doesn't mention the null hypothesis. Do you think mentioning null hypothesis as well would help communicate complete information?

Between 04:50 and 07:00, you mentioned multiple hypothesis tests which different values for the mean of the null hypothesis. (i.e. data sample is fixed and the null hypothesis is variable and probably that's why you are saying that this is the inverse of hypothesis tests.) 1. Are you suggesting using the exact same observed data sample for each of these hypothesis tests? If yes, it seems you are not explicitly expressing that (probably it's obvious for experts). 2. I am assuming that here you are trying to construct a confidence interval for the observed data sample. (I am still trying to get a good grasp of confidence intervals and your answer here will help me confirm my assumptions).

@Jarad Niemi At 07:20, in the expression for Tow-sided p-value, shouldn't the inequality be reversed?

clutchh

Thank you for the video!! very helpful

thanks

very clearly explained! thank you!

Sir any reference please

SUPER INCREDIBLE

Good lecture on contrast

How can figure out the expected value of Y_bar_{1..} for this model?

Hi Dr. J before I do this analysis, do i need to test my data for normality and homogeneity?

Thanks a lot for these useful hints!

Ok It has been a day now, I wanted to give you that bit of time to respond to what I said before I presented you the truth. If the question is as original in Ask Marilyn, the question asked by Craig F. Whitaker's as follows "Suppose you're on a game show, and you're given the choice of three doors: Behind one door is a car; behind the others, goats. You pick a door, say No. 1, and the host, who knows what's behind the doors, opens another door, say No. 3, which has a goat. He then says to you, "Do you want to pick door No. 2?" Is it to your advantage to switch your choice?" The correct answer is no as your odds are 33.3% and that does not change as the question does not have any mention of the host always opening a door after you picked one nor that you knew before hand that the host would do so, which means you also do not know the motives for the opening of the door nor the option to change your pick of doors and as such the opening or the door and the option to pick the other door does not change your odds at all. Here is where you and everyone else who has been taken in by the deception fails, you see, It is not the opening of the door by itself nor the option given to change your choice, it is the knowledge that you will always be shown a door that has a goat and then be asked if you want to pick the other door, that gives you the 66.6% odds. Without the knowledge your odds will always remain 33.3%. So in short if you want the answer to change the door to be correct you first do what Marilyn did when she was confronted by her mistake, which is misrepresent the question, lie about what the question was.

I watched your video and like all the other videos I have watched today after I first heard of the "Monty hall problem" you are wrong. I am the most intelligent man I know to exist, I know you will find that arrogant and I am fine with you mistaking my confidence for arrogance. You will no doubt doubt that I am correct in you being wrong, but I know it to be a fact and I would not blame you for that mistake as I have become aware that most the world, that know of the "Monty hall problem" Seem to have fallen for this and don't seem to see the flaw in the claims they make.

great

The video is so informative and helps to understand the topic deeply