What Is A P-Value? - Clearly Explained 

THE ONLINE GUIDE toptipbio.com/what-is-a-p-value/

To anyone who still doesn't get this, as the video is a little convoluted: the p-value is simply the probability that the results you've obtained from the experimental group (and no, it doesn't just have to be people) is solely due to chance. Ergo, smaller p-value, smaller chance of it just being due to luck/chance.

I finished my undergraduate in mathematics this year and now I finally understand what p value means

Convoluted video, not simple at all for beginners, thank God I'm not a beginner. Simply P-Value is the percentage of Luck and False positives affecting your results instead of your experimented factors. So in an even more simpler way: P-Value % = Luck, the less % the less luck and more real effect of factors experimented by you.

Finally. I've heard it so many times and now I finally understand it! Thanks!

Finally i got an intuition about p-value, thank you, may the almighty bless you 🙏😊❤️

What a great explanation! This is a content area in which I struggle and the visuals and explanations helped me understand the topic more. Thank you!

really great explanation - before I had a problem with understanding the p-value. The example with "two worlds" is a great way to explain what it really is. Thank you!

Incredibly well explained! The first time you gave the definition for p I had no idea how to interpret it. 5 minutes later I understood the same definition perfectly.

This is the best video that I have watched in the explanation of hypothesis testing. Thanks a million for this video.

THANK YOU! I m trying to catch up with my studies and your videos helped so much! Also, it would be nice if you make more R programming tutorial as i love the way you explain things. It's really clear

I'm not sure how many videos I have watched about this topic, it has been more than 6 hours of me trying to understand it, but THIS, this is the only video that made sense to me, and I finally can say that I understand! TYSM!!

YAS! After 3 years of college as a bio student, I finally someone who can actually explain this!

Well explained. I never ignore liking and subscribing such well explained content.

I really appreciate the way you bring us the example; this really helped me a lot thankyou!!

Very lucid explanation Now I can understand what p value is atleast to some extent Thanks very much

im so grateful to have found this channel

You are a very good teacher. Kudos.

One of the best explanation, the probability would be more interesting if all colleges have teachers like you

THANK YOU SO MUCH, please keep on making the good exploitational videos.

Really efficient explanation! Thanks for sharing 👏🏼

I think this is the best video explaining p value. Straight to the point and less technical jargon

Awesome! Great job!

Finally i get the idea of p value. thank a lot

Amazing, many thanks 🙏

Brilliantly explained

wow! this is the only video that finally helped me get this thanks!

This is amazing, thank you! The only thing that would make it even better is maybe a simple explanation of how the p-value is derived in the first place, for this probability to even be identified.

I have just come to this in my social sciences degree. I will be watching this video a great many times in the next few day's.

A great review! Thanks.

This is very helpful, thanks

Great explanation!

Thank you very much for brief presentation.

Thanks so much that was a great explanation.

Excellent Thank you so much for your clear explanation.

Thanks for this!

Thank you very much it cleared my doubts!

Appreciate the explanation ❤

thank you that was so useful

Well presented

very helpful!

Wow you are a good teacher

loved it!

Many thx. It is my first understanding it.

Best video out there

The more I watch this, the more I believe that it would be better to introduce the random noise concept when you were explaining the null hypothesis. So, we formed this null hypothesis BASED ON THE ASSUMPTION that our data were observed due to extraneous factors (random noise), like the mentioned high metabolism gene. If the noise is what contributed to the difference, then we CANNOT assume that the drug worked to reduce weight. IF the observed results were due to random noise, then our p-value tells us that we can repeat this experiment 50 times and only 1 of those times we could get this same (or more extreme) result. This is very unlikely, and so we can be confident in rejecting the null hypothesis and accepting that our results weren't caused by random noise.

Incredibly perfect

Thank you very much.

thank you I understand now

Thank you very much dear

nice! good job

I was too busy looking at those lovely drawings to get it!

Good Video

Thanks!

good video

Thanks

Thank you sir

thanks for the video but still confused...watched lots of videos but non was helpful to me. your video is simpler but needs some more explanation to clarify my concepts.

great!

Boss, p-value of 0.02 is highly significant. P-value is probability of null hypothesis being true. At 0.02 alternate hypothesis gets selected.

Thank u so much

All the comments are wrong. A p-value represents the probability of observing a sample statistic at least as extreme as the one actually observed under the assumption that the null hypothesis is true.

Thank you so much! From what I understand, the smaller the p-value the closer one gets to the edge of the distribution, meaning that it is less likely we get something more extreme. I would just like to clarify a statement "The smaller the p-value the less likely we found this result purely by chance" Is this statement true because finding values at the edge of the distribution are extremely unlikely in the first place?

Nice

awesome

Please how I know the standard deviation ( at 100 trials ) of an outcome that has 78% probability of occurring ?

Can we say this : while settling for Ho (no difference), p is just the chance of an anomaly i.e. the chance that a difference may exists? If we set a threshold alpha, then were a saying that if this percentage of anomaly is gt alpha then we are not going to go with Ho?

If I got it right, it would be better written like "If this were true, what is the probability of discovering a 1 kg reduction (or more) in body weight in those treated with Drug X from our sample (Group B), compared with the placebo (Group A) BY A RANDOM CHANCE (ACCIDENTALLY)" on 3:05

Subscribed. To reduce coincidence of random sampling, in this case, would the researchers filter out people with that gene before conducting the study?

This nice video that correctly makes the point that the p-value is a probability assuming two populations are statistically behaving equal. However there is a further small print that the video does not go into: it not only says that it is _assumed_ that the populations are statistically behaving equal, but statistically equal in the sense that they are both _independent_ samples of a a very specific _assumed_ statistical model e.g. from a normal bell shaped distribution (or for the conoisseurs depending on the test: student-t, or binomial or...). It is precisely because of such assumptions that one can _compute_ the probability of an outcome at least as skewed as was found: once you make these assumptions it is math not non unlike the proverbial math exercise that asks you to compute the probability to throw 600 or more heads when throwing a coin 1000 times assuming the coin is fair and has 50% probability to show up heads. Whether the assumption of a specific distribution is warranted depends very much on the problem (read experimental setup) and the kind of questions you ask and in particular which test you use (the so called "non parametric tests" tend to be a lot less sensitive to at least the assumption of normality). In general, no statistical power tool can substitute understanding experimental/measuring setup, and tests that work brilliantly for finding minute differences in energy by testing trillions of indistinguishable electrons, may also "prove" there is a statistical difference between groups of thousends of people, except it just shows you detect a difference assuming all the idealisations and assumptions, which may likely be impossible to organise (good luck trying to find two random populations, and treating them exactly equal), and in any case given enough people you can always find differences, but the differences between individuals are much larger! Mind you, this is not a dunk on statistical testing or on p-values! They are an extremely useful tool to keep everyone honest!

Superbly explain 👍

I understand the null hypothesis, i.e no difference with control group and the group that gets a sugar pill, but I don't get how the percentage that is arbitrarily assigned . What is that assignment based on?

Thanks for the contents. in my opinion, it is easier to focus on the subject if the annoying hand and the anime is removed

Should the alpha be halved when being compared to the p-value for a two-tailed hypothesis test?

Hello there, you say that the p-value is the probability that there is a difference in the weight greater than 1 kg between the two groups - provided the null hypothesis is true. Therefore, wouldn't it be more logical to reject the null hypothesis if the p-value were large

if say p-value = 0.01, does this translate to that there is 1% chance that the null hypothesis is true and but there is 99% confidence that the null hyphothesis is not true?

Hello. A question: if i had to interpret a p value of 10%, does that make sense when i say there is 10% chance to observe the difference in the popn given that H0 is true?? For me it somehow doesn't sound right, i mean in this case we actually accept the H0, since 0.05 our threshold. Can you please help me with it? Thank you in advance

Hi, this is incredibly well explained, but I am still a bit confused if you could please clarify something to me: Given that the p-value is the probability of the alternative hypothesis given that the null is true, why wouldnt a low p-value imply that you accept the null instead of rejecting it? For example, given that there is no difference between the weights of the two groups, the probability of it actually being different is so so low that wouldnt this imply that there is indeed essentially no difference between the weights, and hence we should accept the null? Please please help me clarify this in my brain, I would appreciate it so much.

More important is how you come up with the p value. Can it be manipulated?

Is the p-value based on the idea of hypothetically repeating the experiment a bunch of times? (Which we don’t do)

I think the only part that seems counterintuitive is if it's just a tiny noise (say 0.02), why should we reject the Null? It should be the other way round. Nay?

I dont think you can draw all these perfectly so fast

The difference can be due to variables not accounted for in the experiment. It need not be “random noise”.

drug X is my favorite. the p-value of that is pretty high.

What’s the difference between a “p-value” and the “actual significance level”?

From Bangladesh 😍

I am confused. Is p value =0.02 really means 2% chance of observing the weight loss or 2% chance of observing the weight loss due to some random fluctuations and 98% certain to observe the weight loss?? If p=0.02 means 2% change of observing the weight loss, than how p

Cohen did a great paper on p-values called something like "The Earth is Round p < .05". The p-value is the probability of the DATA (not the hypothesis!) or data more extreme, ASSUMING the null hypothesis is true. That's why effect sizes are important to include, along with confidence intervals. So you get effect size E, the p-value is the probability of that effect size or one larger, assuming there is in "reality" no effect (the null hypothesis). It is p(D|H), not p(H|D)...and to understand the larger context, one needs to understand Bayes' Theorem which logically shows how one adjusts probabilities of hypotheses based on data. Bayes' Theorem is also the normative model for subjective probability change based on data, against descriptive models such as cognitive bias.

5:02 where did the 1kg (or more) come from?

Please explain that .... If two groups are identical... Thn p value just 2per ... Shows that only 2 per chance that these are not identical... Why for just 2 percnt we reject null hypotheses

it's the probability of sum of three things: (1) of an event occurring (2) of an event occurring that is just as rare (3) of an event occurring that is rarer or more extreme than 1 or 2. Boom!

I need drug x

Surely the null hypothesis should be: "There is no significant difference as a result of the pill"

Just remember that Group A will probably reduce, because of they know that they are being measured, that is exactly why we need to do this, to know how the people behave just by being measured.

All made sense until 4:40. Don't you mean at p=0.02 there's only a 2% chance the weight loss would be LESS than 1kg (i.e. closer to the null hypothesis)?

3.32, yes but why. The counter intuitive aspect is not addressed. A lawyer having smaller amounts of evidence would not lead to a conviction.

I mean if something has an effect(which is opposite to null hypo) and its proved why should there be a P value(iff null hypo is true) like it's not logical null hypo isn't true why say if null hypo is true what is probability gett9ng results

You made everything much more difficult to understand!

Did even understood a single point from the start, please start explaining in a good manner

Without statistical analysis, it's better to begin with a simple example where population mean or average in the null hypothesis is above (or below) a hypothesized value. I would avoid discussing p-value from a test about parameters like ``mean-difference" or "slope" in a video without showing any statistical analysis.