Hypothesis Testing

[ipconfigearth102]

thanks for the video.. i am bfm student.. just got interested in this now as u showed such a good video... can u explain more in detail?? or just send the link...

[TheSpikeystuff]

great video thanks

[smuggecko]

some really good stuff here

[racastilho]

Nice, but one thing still puzzles me. When we test the hypothesis that the mean of a population is *exactly* a given real number, shouldn't we be able to reject it always? I mean, what is the probablity of a continuous random variable to be *exaclty* any given number? Zero, is it not???

[Tribal1693]

@racastilho Penis

[jaredmills06]

You know it is a good tutorial when you can watch it at 1 am and still understand it. Well done, and thanks for uploading :)

[cathex]

Brilliant tutorials, clear, concise and beautifully articulated. Thank you.

[TheRulaBaby]

great vid thanx a lot, but how do u calculate the p value manually?

[TheRulaBaby]

LMAO ur examples the same one in my book same figures and everything :D

[detestableguy]

if this is aa two-tail test then don't you think df needs to be n-2?

[tijani927]

Hi, Congratulation

Well I like you demostration but my own problem is very diffcult for to test.

here is my email ahmedhm1@yahoo.co.uk please kindly let me have yours so that that you can help to test my Hypoth.

[missypunkrocker]

7 people are stats lecturers.

[Seshom]

upload the exel doc =D

[kyuubinaruto2345]

@Kissiq1618 most of the time no, its just people prefer to have a sample size of 30 so that way its safe to assume that u can use a normal distribution and use the z-test or t-test. it doesnt matter its just 30 is consider a "good" size for sampling.  Hope this help :p

[ampikine]

Brilliant video. Thanks a lot!

[VerticleIce]

I have a horrible professor who doesn't understand how to teach. Your videos are so helpful. Thanks!!!!!

[ITWebSolutionLLC]

@Hulkmania - degrees of freedom are how many variables exist. To find this out, you take N-1, where N=the number of variables in the study.

[ZeroSumRuler]

I can't "like" your video enough. It taught me how to run statistical tests on my thesis data! THANK YOU!!!!!!!

[arshadahmad1]

thanks for the video. it really helped.. you should give more real life example to understand the hypothesis

[shhawkward]

i m supposed to do an assignment in which we have to find statistics and apply all these theories/formulas learnt to make a conclusion. Can someone tell me where can I find statistics online? Preferably, one with large samples. Thx ! :D

[CellarDoor367]

@shhawkward NIH.gov has a lot of large statistics sets.

[TheMesho666]

I want to know how did you get the 18.5 hypothesis? could you clarify this more please,

[Hulkmania316]

what exactly is degrees of freedom?

Thank you for your answer David

[Jloslefthand]

@Hulkmania316 Umm lol I'm late but n-1= degrees of freedom.

[omenukwa1]

this is great my teacher had me sooooo confused, but he understood it

[bhu4153]

hey question, we're using the t test because our sample size is less than 30?

[MDMike1988]

@bhu4153 t is used in stead of standard normal (z) because of the fact that the population deviation (σ) is unknown.

[Jloslefthand]

@bhu4153 You use the t-test because sigma is unknown. Use the z-chart when sigma IS known.

[vitalstrike82]

Thanks....

[TheMissChilli]

FANTASTIC !!! Thank you!

[truncale74]

Very nicely explained and many thanks! =)

[AzxE10]

Very good video, thanks!

[pedropaulo1784]

David,

Maybe it'd complement your explanation the fact that you can get the t value from a two-tail t table that corresponds to the 5% significance OR get it from a one-tail t table that corresponds to the 2,5% significance. The output is the same.

Regards,

Pedro

[trisky1234]

duuuuuuuhnominatornominatornom­inatornominator

sorry, heheh. nice tutorial!

[TheDentist27]

"I can reject with 98.7% confidence"

My old stats prof would say "No! You can reject with 95% confidence. The p is post-hoc."

I have no idea who is right. But I think effect size should come in here somewhere?

[trisky1234]

@TheDentist27 my textbooks says 90%, 95%, or 99% but I guess you could always reject with another value in theory :)

[TheDentist27]

@trisky1234 You select the 90%, 95% or 99% before you run the test.

If you collect 100 samples from the same population, and set alpha at 95%, then 5 times out of 100, you will fail to reject the null. You will also find P values between 95-99. So to take one of these p values, say one that is 97.5, it would be a mistake to say that you can reject null with 97.5% confidence, because you know that the other 99 times will show different p values.

[spand3244]

Really great. I hope to see how exactly the p-value has been calculated. I dont wanna use excel's own function.

[lmpeders]

Please correct me if I'm wrong and explain why, because when I look up the critical t values, I get critical t = 1.703 when a=5% and v=27. When a=2,5%, I get critical t=2.052. Where's the error?

[InternetLoveMachine]

t values mean two tailed test. Basically instead of finding one a at 5%, you have two a's on each side. At 5%, you'd have each a at 2.5% (Iboth have to equal the significance value)

hope that helps, got my final today on this...

[lethe123]

thank you David, now i got the idea how I can work this out in my upcoming test!

[Souliee]

Thank you David, this will help me on my upcomming exam in math-stat.

[Melodyv03]

my collage friend told me to wacth this and im only in 6th grade and i dont understand a bit of this thing

[neoking77]

the turtle's finger that is

[neoking77]

i swear it looks like that finger is flipping me off!

[makomark]

Thank you very much for that, very clearly explained, and found it great how you link the 3 tests or measures together in the explanation.

[Krish1309]

Excellant Tutorial. You explained things in easy manner by using excel. I rate your postings is good and easy to understand

Could u pl advise me your other postings in youtube related to statistics

[cusanusnicolas]

For comprehensive Randomized Block Theorem for better design and interpretation of RB designs, search Yahoo for "Randomized block foos" - I hope to have a utube presentation for it soon.

[ashumittal79]

Excellent Tutorial.... Thanks a ton buddy

[jimbot72]

excellent, thank you.

[finanzist]

Thanks a lot ! Great explanation

[m3Rlinus]

Thanks a million! Short, simple, straight to the point. Thanks to your tutorial now it all makes sense!

[caviper1]

Can you show us how to test hypotheses under Bayesian assumptions?

[yc3p0]

Can anyone tell me the formula for computing the p-value with your t-value, d.f., and # of tails? Is there an exact formula for that withouth using a spreadsheet?

[smartones888]

many many thx...great explanation!

[RednaxNL]

men;) you can explain things better in this 8.55 min. than MY MATHTEACHER at school tried to do in three weeks!:P

[wubs23]

@RednaxNL just go to your teacher and ask him to explain it fully because you don't understand. and if he makes a step you don't understand, just ask him why. he wants to explain, that's his job. but you have got to ask the right questions ;)

[alatif6]

you are indeed the most helpful person on youtube so far!

[brymanbonner]

Thanks very much for the video, David!

[jcnichols3]

Hi David. I just ran across you videos and think they are GREAT. I do have a question. I built a spreadsheet and the test blew up if I used a hypothesis greater than the mean. Is an absolute value appropriate?

Thanks again

Nick

[pompomjangkrik]

Good video... I wish u were my lecture.

[givingstars]

is standard error the same as coefficient of variation?

[brymanbonner]

givingstars-

The coefficient of variation, or CV, is the standard deviation of a set of observations divided by the mean value of the same set. The equation is:

CV=sample standard deviation/sample mean

(Sorry I don't have the fonts available to give you the proper representations of the variables.)

[brymanbonner]

(Continued)

The standard error of the sample mean (from a sample generated by a population with standard deviation σ), on the other hand, is either expressed by:

σ(sub sample mean)=σ/square root of n

when we know population standard deviation σ and where n is the sample size

or by:

s(sub sample mean)=s/square root of n

when we don't know σ and need to estimate it using the sample standard deviation s.

Hope this helps, although my response is so late!