Understanding the p-value

[flukemate]

"p is low, null must go" <-- love it

[fhealer1]

Well put together , thanks for the video, it was very helpfull

[SuperMrLahlou]

helped greatly, cheers!

[DaisyExodus]

I wish all my lectures were like this...

[UCMSCI]

@DaisyExodus Thanks. We are busy remaking some of these so we aren't breaking copyright (!) And making some more. We are also putting them into an iPad app, along with intereactive quizzes to help people learn. SO thanks for the encouragement. We really appreciate it.

The new videos will be at CreativeHeuristics Youtube channel, and the app will by AtMyPace:Statistics. Tell your friends.

[LastDodosaur]

youtubing stats stuff while trying to cram for test. See pheonix wright in thumbnail. ... ... ... It's 3 AM.

[efb420]

But how do you work out the p-value?

[UCMSCI]

@efb420 If you look at my response to @mmmcr00 above, you can see how. Generally the computer package works it out, and we interpret it.

[palui]

The p value seems to be calculating a probability AFTER the data is in. In other words, it's producing a probability after the random event concludes. But after the data is in, there is no probability anymore. It seems the probability that I received my data, or something more extreme, after the data has been received, is 1. Seems odd.

[IceAges14Aces]

@palui The P-Value is technically the probability of extreme or more exteme of getting an observation like the one actually observed in your study/experiment when the null-hypothesis is true. For instance, suppose I ran 20 studies and our alpha level is 0.05 level. What does this mean? It means that if you run the study 20 times you would get a statistically significant result at the P=0.05 significance level only by pure luck.........

P=0.05 means 1 in 20.

[palui]

@IceAges14Aces I had figured out my error after I had made my comment. Thanks for your reply. However, your reply isn't worded correctly. The p value can be described as: In all possible random repeated samples of size n from a population for which the null were true, the proportion of times for which your test statistic was as large or larger than the value realized in your observed data is the p-value. p=0.05 has nothing to do with running a study 20 times.

[IceAges14Aces]

@palui I was giving you an example of what P=0.05 means, not that P=0.05 is always 20 studies.

You're definition of the P-Value is correct by the way.......

[TradingTutor]

Death note!

[PhoenixWright1000]

Feenie Wright in this vis makes it 10x more awesome.

[KRKbert]

Are you a Kiwi? Sounds like it!

[UCMSCI]

@KRKbert Yus.

[Yngwiaever]

Helen?

That's a guy.

[JoeDaPlumber]

I hate finals. This vid helps a lot.

[mysteryg]

0.2578 is the p-value I got when I manually tried to calculate this. I used the sample deviation and looked at the Z table, I thought of looking at the t-table but I wouldnt know about that because I dont have an alpha.

[UCMSCI]

@babish82

Well done. What you say is correct in essence, though a bit dodgy in terminology (even by my standards). You've got it. Just remember the probability is of getting that result, even though we act as if it is the probability that the alternative hypothesis is true.

[babish82]

@UCMSCI Thanks a lot...actually, I'm a man of medicine, and statistice is a completely new thing for me, thus I had to put in a lot of thought in it, in my own way...but thanks a lot...you're an angel...

[babish82]

OK...this is what I've understood, and kindly correct me if I'm wrong:

If p-value is low, there is a LOWER probability that your alternative hypothesis (Ha) tested on the basis of chance or luck, which means, there is a LOWER probability that Ha was incorrect...and thus a HIGHER probability that your Ha was correct, and as Ha was correct, Ho is incorrect, in other words:

Low p-value, low probability of luck or chance in Ha, Higher credibility for Ha (and thus lower favor for Ho)

right?

[mmmcr00]

why is it that the smaller the p-value the greater the evidence against Ho??

[UCMSCI]

@mmmcr00

Because a small p-value is a small probability, which means that it is unlikely to get this value just by chance if the null hypothesis were true.

The smaller the p-value, the less probability that it was by chance - thus more evidence that the null hypothesis can't be true.

You may wish to watch the video again - it is a very strange concept.

[mmmcr00]

but HOW do you get 0.24?????? they all just say the p-value yet never bother to explain HOW they got it.....howwwwwwwwww???

[UCMSCI]

@mmmcr00

This video is about understanding the p-value. Actually getting one is quite a different story. In this example we are using Excel.

It's a bit tricky in Excel as it doesn't have a test for a mean compared with a value.

Make one column with the 5 numbers: 73, 70, 65,68,69. Then make another column with 70,70,70, 70, 70. Then use the data analysis addin and do T-test two sample assuming unequal variances, and look at the p-value for the one-tailed test. (you asked!)

[puttjattda101]

@mmmcr00 get the casio fx-9750 gii and ull be set

[phidelio82]

Good explanation but why this riduculous animation? It serves no purpose and is just confusing...

[p0sta]

thanks you, this is really helpful to me (and many others as I can see =D )

[ForEsmee]

This should be shown in school. So easy to understand and so important if you are involved in statistics

[HanyauntukAllah1]

Now i understand,I like the wording: "there is no enough evidence to reject the null hypothesis" rather than the usual "not statistically significant."

[guitargeekette]

MELLO! Oh my gosh this is awesome! If math was taught like this I'd love it a little more.

[MiyaviDas]

Can you explain, when using excel, when it is necessary to use 1-p (i.e. 1- the pvalue given by excel) instead of p? I understand that it depends on the null and alternative hypotheses you are testing, but am still confused.

[jeroenve]

P is low, H0 must go

easy to remember

thx

[Tentenlaw]

That is something I like to ask. Do you know how to calculate the p-value by hand? Any formula? I know how to do it on a calculator but not by hand. Also, from my calculator, I get a t value.

Comparing this value with the critical t-value, is there anyway I can determine whether I can reject or fail to reject my null hypothesis WITHOUT the p-value? What does that t-value tell me?

[rishime7]

Null hypothesis is the hypothesis which is tested for possible rejection, under the assumption that it is true. i believe helens null hypothesis shud be that choco nutties are not as they shud be.

[Maoribrotha]

I think you're accent is great...sounds kiwi to me...

[caaantttiii76]

THANK YOU !! you made us pass our EXAM YAYY!!! great explanation :)

[mriwilis]

Fantastic, clear, and funny!

[NeroIstaffer]

OMG That's not helen, it's mello!! :D Anyway thanks a lot for this simple and easy to understand fun demonstration!

[libreallemagne]

how cool! thank you ! and now the t-test

[midnightroz]

This was quite helpful. Thank you. On a side note, it's amusing that Mello (a boy in the anime) turned into a girl in this video.

[tufafaku]

Great work.

[neusdadt]

this is a life saver. thanks. but our stupid university doesn't allow excel during tests so i still dont know how to manually get the p value! :(

[MoreMusicXx]

@neusdadt the way to find the p-value is:

z= (expected - observed)/SD

hope this helps!

[Faisdragon]

damn... i in love with u :P and also because of the accent

[MrGreenGrinch]

Thank you very much, this helped a lot. I'm struggling with statistics and this was a lovely change from all the jargon i'm used to from numerous text books.

5 Star :)

[A21Joker]

I love the accent!

[robooJack]

Very nice.

[superdaveXtreme]

What's a statustic?

[pepbean]

My Stat's class were taught the P-Value in August I never actually learned it until today when I watched your Video...THANKS I don't feel dumb anymore!

[79supersarah]

I saw your talk at INFORMS this afternoon and loved it. Very cute and informative.

[opiates]

LOL Death note...

[frankg56]

This is great stuff - maths as it should be....

practical, interesting , informative and with a touch of humour

[zaynahmed]

can you or someone here explain how you got the value you did for the p value

[UCMSCI]

It's a bit tricky in Excel as it doesn't have a test for a mean compared with a value.

Make one column with the 5 numbers: 73, 70, 65,68,69. Then make another column with 70,70,70, 70, 70. Then use the data analysis addin and do T-test two sample assuming unequal variances, and look at the p-value for the one-tailed test. (It's easier in a proper statistics package).

[ovrjoyed]

Great Job!! I understand!!

[UCMSCI]

That is great news.

[LBMiller79]

I tried to learn...now I'm dieing of laughter once you mentioned "Choco Nutties"

...and i have my AP Exam tomarrow

[vinzbrain]

Ho: This video isn't cool

(p<0.05)

[UCMSCI]

lol - you obviously got it

[TerrawindX2]

Ho: This video is awesome

H1: This video isn't awesome

(p>0.05),

As p>0.05, There is no significant evidence to reject the null hypothesis, therefore we fail to reject the null hypothesis. :)

[yzy010858]

@vinzbrain to be precise alpha>=0.05

[vinzbrain]

@yzy010858

isn't it that you fix alpha to whatever value you want and then test p against your alpha?

If that is true then I am not sure if you are being very precise

[UCMSCI]

@vinzbrain I'm afraid statistics is far from precise. In fact the choice of alpha is a remnant from when p-values were difficult to calculate. Really we look at the size of the p-value and judge from that.

[lavedon]

Thank you!

[frankXwhite]

p is low, null must go, haha i will appreciate it in my next exam, thank you

[IbanezRG7321]

lmao nice touch with the death note scenes xD

[UCMSCI]

Thanks - most of our other videos use deathnote pictures too.

[TIMWISDOM]

cheers mahlady!!

[waqasy]

I liked it

the other videos ive seen have a riddic. amount of pretentious - smarter than thou feel.

=)

[moxinghbian]

This is a disaster

This an entertaining way to express P value to those who already knew what  P value is.

This is NOT an entertaining way of explaining P value, Worst explanation ever. Your visuals are not aiding the explanation, instead confusing the audiences with cliches.

[UCMSCI]

I'm sorry this didn't work for you. It is designed to be viewed as part of a course in statistics, and generally people need several different explanations to grasp this rather difficult concept. So I guess it is more about reinforcing the concept than an introduction to what the p-value is. You may prefer a more mathematical approach, for instance. My students really appreciated the "p is low, null must go", which I picked up from another Youtube video.

[dc2k404]

w00t! P-Values baby!

[sidiqmk]

Thanks a lot this p value thingy is becoming clearer and clearer to me..

[brendan1222]

Most entertaining explanation of P values I have seen! You have been able to explain a very difficult concept in a straight forward and interesting way. Cheers

[thesheeple]

Nice video, these always help eh ! cheers UC