Performing a One-way ANOVA in Excel with post-hoc t-tests

Tuorial describing how to perform a one-way ANOVA on groups of data and then carry out post-hoc t-tests between groups to determine significant differences.
This is part of a series of tutorials designed to help research scientists in the use of certain software applications commonly used in scientific laboratory work.
You can find the entire set of tutorial videos here: ehealth.kcl.ac....
The screencast videos have been made by the author (Dr James Clark, King's College London) in response to common questions raised by students on BSc and MSc courses and are recorded using Camtasia Studio. The content is targeted at students of all levels of undergraduate and postgraduate education as well as professional research scientists.
If you wish to link to this video on another web site please make sure you credit the author and provide a link to the blog site (shown above)
©2013 James Clark, king's College London. All rights reserved.

[@wardag5370]

I've been struggling with this for ages, thank you so much.

[@DoryVideo]

Glad it helped

[@firstname4337]

great video, great explanation -- I could have used this 3 days ago LOL youtube is just now recommending it to me after I spent the last week seraching for stuff like this

[@archaerial131]

Thanks for the simple explanations. I've been trying to figure out how to compare two groups within an ANOVA for a while. I feel dumb not realizing you do a simple t test.

[@non_ye]

You were extremely helpful with clear explanations. A huge thumbs up to you :)

[@riciy]

thank you so much, your calm and clear way of explaining really helped me with my thesis, I never was a very bright user of statistics (more the practical kind of guy) so it helped me big times..Greetings

[@doreencalo8736]

Thank you! Your lecture is so much easy to understand. GReatly appreciated.

[@DoryVideo]

My pleasure

[@brycesoraru9528]

Very helpful! Thank you for the clear explanation.

[@DoryVideo]

My pleasure!

[@taymaysie9221]

where tf were you when i was doing my stats paper!!! ur amazing!!!

[@noat5090]

very nice video. but how you assume that groups have equal variances? and please show us how you make group B bar in this graph ?

[@adrianegeesaman9157]

Very helpful and straightforward. Thank you for posting!

[@wangfluenza]

I just passed and exam thanks to you!

[@DoryVideo]

Great job!

[@GaryHutsonVBA]

Wouldn't you expect Group B to be significant, as the sample size of Group B is higher than both other groups. Then, you are assuming equal variances although they are not. Group B is twice as high as Group A! Surely, a T test with assuming unequal variance would be the best way to go? Please fill in the gaps please? Thanks

[@mouradmadouni8277]

Thank you very much. It's very helpful and clear.

[@DoryVideo]

My pleasure!

[@personaldevelopment1802]

very Helpful and easy to understand, thankyou Sir

[@yaqubmir]

what when we use two comparisons, do we have to divide 0.05 by two for Bonferroni test. Thanks in advance!

[@vinuthaksmysore]

hi what if we have too many groups to do the post test say 50-100 groups?? How to do it then? any other methods

[@ABunchORandomCrap]

thankyou from 2019 this saved me

[@DoryVideo]

Happy to help

[@drga4977]

Thank you so much for your help>> I have been trying to download SPSS since 3 days and never succeeded... :)

[@nikavorster]

THANK YOU SO MUCH!! Makes so much sense now !

[@Buuvei24]

very helpful explanation. thank you sir

[@DoryVideo]

Happy to help

[@cemilbayram]

Cannot find "Data Analysis" button under the Data section. Should i download it?

[@DoryVideo]

Yes as the video says you have to enable it first

[@drahmedshawkyphd2726]

Thank you for your professional explanation. Can i perform it among 7 different sites? Thanks

[@eldalea1158]

Thank you so much! You are a life-saver.

[@anushreesanyal3147]

Crystal clear. Thanks!

[@DoryVideo]

My pleasure!

[@arc.center9517]

Thank you so much ..It is really very helpful

[@DarkSokiX]

Do you actually have to compare group b with group c if the graph tells you that group a and group b are significantly different, and that group c is smaller than group a? Thank you for the response!

[@mgb2172]

The test is what tells you if the difference is SIGNIFICANT. Seeing it on the graph shows there's a difference, not that the difference is significant

[@glendaedwards5006]

Thanks so much for this well detail video, definitely using now to do my final year project

[@DoryVideo]

My pleasure

[@Buuvei24]

how did you find value 0.0167?

[@sparkles9729]

This is helpful, but how do I run the post- hoc tests if I have 5 groups?

[@DoryVideo]

I’m not sure if excel can handle that many groups.

[@md.masumbillah8222]

nice explanation sir.

[@DoryVideo]

My pleasure. Glad I could be of help to you

[@Onion176]

Very clear explanation, appreciate it!

[@thelazycat_]

Why do you not a F-test to know if the variance is different, and why not paired two samples for means? thank you

[@reetobrata]

excellent help. Thanks a lot!

[@nathaliearbousoff3476]

There are actually 4 tests performed if you include the anova. Shouldn't the correction change 0.05 into 0.05/4 instead of 0.05/3? If anyone can help me on this question, thanks

[@ralphanthonyespos9417]

No. He was referring to the number of pair-wise comparisons as the denominator for the Bonferroni correction. In this case, it's 3 (A-B; A-C; B-C). Generally, it's Comb(n, 2) where Comb(n, k) is the combination function for n taken k, and n is the number of elements (groups).

[@greysonnleee]

would u please tell about how 0.0167 comes out?

[@DoryVideo]

Not sure what you mean there, that’s the P value as a result of the ANOVA

[@noha1878]

He divided the significant level (0.05) by the number of comparisons which are 3 in this example.

[@heyitsysaaa]

Thank you for this!

[@DoryVideo]

My pleasure!

[@musomuzo]

If I was going to do a T-test for each one to check if they are significantly different ANYWAY, wouldn't I just skip the ANOVA step? Seems unnecessary..

[@lynnlyfish]

You would increase your chance of error by running a bunch of t-tests.

[@nathaliearbousoff3476]

but the ANOVA is an additional analysis, how doesn't it increase the overall alpha risk?

[@AustinRickard]

What kind of post hoc test is this?

[@jncook4312]

Why would you select "t-test assuming equal variances" when the result show that the variances are in fact, unequal?

[@vasilinakapetanaki6839]

Thank you very much!!

[@kar2194]

Great video thank you. Could anyone help me please... ? I am confused when to use Equal variance or Unequal Variance during selecting the t-test

[@shahiruladelie4287]

Can i use one tailed t test

[@DoryVideo]

You CAN do one in excel, of course, just adjust the parameter in the function when performing a t-test; but for data with more then two groups you cannot reasonably use a t-test.

[@ruzaiq100]

Great video, thanks!

[@DoryVideo]

my pleasure, goad you find it helpful!

[@AlejandroIllustratorCC]

wow! thank you so much!

[@toastyminimallows]

Thank you!

[@DariusDevlin001]

Fantastic! Thanks

[@DoryVideo]

My pleasure. Please subscribe as I’m adding more videos all the time

[@alanalo983]

thanks so much!!!

[@r.paramadina5532]

Thanks~

[@DoryVideo]

You’re welcome!

[@abrarbio1828]

Thank you :)

[@danielelacaita5307]

I love you

[@DoryVideo]

Why thank you!

[@JesusRodriguez-zx3em]

Jesús Robiragz Moirruf 💋 Ayúdame BodoMs-Los -Los

[@azadehtajmirriahi3872]

Thank you so much!

[@trangdo6229]

thank you so much!

[@DoryVideo]

My pleasure!