# Where Cohen went wrong β the proportion of overlap between two normal distributions

Iβve received many emails regarding the percent of overlap reported in my Cohenβs d visualization. Observant readers, have noted that I report a different number than Cohen (and other authors). For instance, if we open p. 22 in Cohenβs Statistical power analysis for the behavior sciences, we see that Cohen writes that d = 0.5 means a 33 % non-overlap, i.e. a 67 % overlap. However, if you go to http://rpsychologist.com/d3/cohend and set d = 0.5, you get 80 % overlap. How come?

Letβs calculate this ourselves using Monte Carlo integration. Doing it this way will make it easy to see what weβre doing. What Iβll do first is just sample uniform random data, and then save the points that fit under each normal curve. Letβs assume we have a treatment group and a control group, then each point will represent one patient. Hereβs the code.

And hereβs the distribution of points that we generated.

Now if we overlay the two distributions we get

Also, letβs color the individuals in the treatment and control groups that are in the region that overlap. Weβll also label the regions and show percentages.

What weβve done to get percentages is simply to count the observation and divide them by the total amount. We see that we have 10 + 10 = 20 % non-overlapping observations. The overlapping region is more densely packed with observations, since both groups contribute an equal amount of observations that overlap. The proportion of the total amount of observations in the overlapping region is 40 + 40 = 80 %.

Now, letβs plot Cohenβs interpretation of overlap, which is the proportion of shared area covered by both population. Since weβre now thinking about areas, and not frequencies, weβll have to remove half of the observations from the overlapping region. Hereβs how it will look.

We see here that the shared area is 67 %, which is a different interpretation of overlap. Clearly, I believe this interpretation to be a bit strange (since Iβm not using it on my site). To me itβs more intuitive to think: if we have two populations, one that received an active treatment, and one that received a control, then 80 % tells us that 80 % of the total observations actually overlap. I believe thatβs more intuitive than reporting that the shared area is 67 %. Whatβs interesting is that Cohen even used a frequency interpretation in his description (the distributions are βequally numerousβ).

# References

Iβm not the only one to note this discrepancy, Grice & Barrett (2014) used the same reasoning in their article βA note on Cohenβs overlapping proportions of normal distributionsβ.

Written by Kristoffer Magnusson, a researcher in clinical psychology. You should follow him on Twitter and come hang out on the open science discord Git Gud Science.

Share:

Published January 23, 2017 (View on GitHub)

### Buy Me A Coffee

A huge thanks to the 152 supporters who've bought me a 361 coffees!

Jason Rinaldo bought ββββββββββ (10) coffees

I've been looking for applets that show this for YEARS, for demonstrations for classes. Thank you so much! Students do not need to tolarate my whiteboard scrawl now. I'm sure they'd appreciate you, too.l

JDMM bought βββββ (5) coffees

You finally helped me understand correlation! Many, many thanks... π

@VicCazares bought βββββ (5) coffees

Good stuff! It's been so helpful for teaching a Psych Stats class. Cheers!

Dustin M. Burt bought βββββ (5) coffees

Excellent and informative visualizations!

Someone bought βββββ (5) coffees

@metzpsych bought βββββ (5) coffees

Always the clearest, loveliest simulations for complex concepts. Amazing resource for teaching intro stats!

Ryo bought βββββ (5) coffees

For a couple years now I've been wanting to create visualizations like these as a way to commit these foundational concepts to memory. But after finding your website I'm both relieved that I don't have to do that now and pissed off that I couldn't create anything half as beautiful and informative as you have done here. Wonderful job.

Diarmuid Harvey bought βββββ (5) coffees

You have an extremely useful site with very accessible content that I have been using to introduce colleagues and students to some of the core concepts of statistics. Keep up the good work, and thanks!

Michael Hansen bought βββββ (5) coffees

Keep up the good work!

Michael Villanueva bought βββββ (5) coffees

I wish I could learn more from you about stats and math -- you use language in places that I do not understand. Cohen's D visualizations opened my understanding. Thank you

Someone bought βββββ (5) coffees

Thank you, Kristoffer

PΓ₯l from Norway bought βββββ (5) coffees

Great webpage, I use it to illustrate several issues when I have a lecture in research methods. Thanks, it is really helpful for the students:)

@MAgrochao bought βββββ (5) coffees

Joseph Bulbulia bought βββββ (5) coffees

Hard to overstate the importance of this work Kristoffer. Grateful for all you are doing.

@TDmyersMT bought βββββ (5) coffees

Some really useful simulations, great teaching resources.

@lakens bought βββββ (5) coffees

Thanks for fixing the bug yesterday!

@LinneaGandhi bought βββββ (5) coffees

This is awesome! Thank you for creating these. Definitely using for my students, and me! :-)

@ICH8412 bought βββββ (5) coffees

very useful for my students I guess

@KelvinEJones bought βββββ (5) coffees

Preparing my Master's student for final oral exam and stumbled on your site. We are discussing in lab meeting today. Coffee for everyone.

Someone bought βββββ (5) coffees

What a great site

@Daniel_Brad4d bought βββββ (5) coffees

Wonderful work!

David Loschelder bought βββββ (5) coffees

Terrific work. So very helpful. Thank you very much.

@neilmeigh bought βββββ (5) coffees

I am so grateful for your page and can't thank you enough!Β Β

@giladfeldman bought βββββ (5) coffees

Wonderful work, I use it every semester and it really helps the students (and me) understand things better. Keep going strong.

Dean Norris bought βββββ (5) coffees

Sal bought βββββ (5) coffees

Really super useful, especially for teaching. Thanks for this!

dde@paxis.org bought βββββ (5) coffees

Very helpful to helping teach teachers about the effects of the Good Behavior Game

@akreutzer82 bought βββββ (5) coffees

Amazing visualizations! Thank you!

@rdh_CLE bought βββββ (5) coffees

So good!

Amanda Sharples bought βββ (3) coffees

Soyol bought βββ (3) coffees

Someone bought βββ (3) coffees

Kenneth Nilsson bought βββ (3) coffees

Keep up the splendid work!

@jeremywilmer bought βββ (3) coffees

Love this website; use it all the time in my teaching and research.

Someone bought βββ (3) coffees

Powerlmm was really helpful, and I appreciate your time in putting such an amazing resource together!

DR AMANDA C DE C WILLIAMS bought βββ (3) coffees

This is very helpful, for my work and for teaching and supervising

Georgios Halkias bought βββ (3) coffees

Regina bought βββ (3) coffees

Susan Evans bought βββ (3) coffees

Thanks. I really love the simplicity of your sliders. Thanks!!

@MichaMarie8 bought βββ (3) coffees

Thanks for making this Interpreting Correlations: Interactive Visualizations site - it's definitely a great help for this psych student! π

Zakaria Giunashvili, from Georgia bought βββ (3) coffees

brilliant simulations that can be effectively used in training

Someone bought βββ (3) coffees

@PhysioSven bought βββ (3) coffees

Amazing illustrations, there is not enough coffee in the world for enthusiasts like you! Thanks!

Cheryl@CurtinUniAus bought βββ (3) coffees

πWhat a great contribution - thanks Kristoffer!

vanessa moran bought βββ (3) coffees

Wow - your website is fantastic, thank you for making it.

Someone bought βββ (3) coffees

mikhail.saltychev@gmail.com bought βββ (3) coffees

Thank you Kristoffer This is a nice site, which I have been used for a while. Best Prof. Mikhail Saltychev (Turku University, Finland)

Someone bought βββ (3) coffees

Ruslan Klymentiev bought βββ (3) coffees

@lkizbok bought βββ (3) coffees

Keep up the nice work, thank you!

@TELLlab bought βββ (3) coffees

Thanks - this will help me to teach tomorrow!

SCCT/Psychology bought βββ (3) coffees

Keep the visualizations coming!

@elena_bolt bought βββ (3) coffees

Thank you so much for your work, Kristoffer. I use your visualizations to explain concepts to my tutoring students and they are a huge help.

A random user bought βββ (3) coffees

Thank you for making such useful and pretty tools. It not only helped me understand more about power, effect size, etc, but also made my quanti-method class more engaging and interesting. Thank you and wish you a great 2021!

@hertzpodcast bought βββ (3) coffees

We've mentioned your work a few times on our podcast and we recently sent a poster to a listener as prize so we wanted to buy you a few coffees.Β Thanks for the great work that you do!Dan Quintana and James Heathers - Co-hosts of Everything HertzΒ

Cameron Proctor bought βββ (3) coffees

Used your vizualization in class today. Thanks!

eshulman@brocku.ca bought βββ (3) coffees

My students love these visualizations and so do I! Thanks for helping me make stats more intuitive.

Someone bought βββ (3) coffees

Adrian HelgΓ₯ VestΓΈl bought βββ (3) coffees

@misteryosupjoo bought βββ (3) coffees

For a high school teacher of psychology, I would be lost without your visualizations. The ability to interact and manipulate allows students to get it in a very sticky manner. Thank you!!!

Chi bought βββ (3) coffees

You Cohen's d post really helped me explaining the interpretation to people who don't know stats! Thank you!

Someone bought βββ (3) coffees

You doing useful work !! thanks !!

@ArtisanalANN bought βββ (3) coffees

Enjoy.

@jsholtes bought βββ (3) coffees

Teaching stats to civil engineer undergrads (first time teaching for me, first time for most of them too) and grasping for some good explanations of hypothesis testing, power, and CI's. Love these interactive graphics!

@notawful bought βββ (3) coffees

Thank you for using your stats and programming gifts in such a useful, generous manner. -Jess

Mateu Servera bought βββ (3) coffees

A job that must have cost far more coffees than we can afford you ;-). Thank you.

@cdrawn bought βββ (3) coffees

Thank you! Such a great resource for teaching these concepts, especially CI, Power, correlation.

Julia bought βββ (3) coffees

Fantastic work with the visualizations!

@felixthoemmes bought βββ (3) coffees

@dalejbarr bought βββ (3) coffees

Your work is amazing! I use your visualizations often in my teaching. Thank you.Β

@PsychoMouse bought βββ (3) coffees

Excellent!Β  Well done!Β  SOOOO Useful!πΒ π­Β

Dan Sanes bought ββ (2) coffees

this is a superb, intuitive teaching tool!

@whlevine bought ββ (2) coffees

Thank you so much for these amazing visualizations. They're a great teaching tool and the allow me to show students things that it would take me weeks or months to program myself.

Someone bought ββ (2) coffees

@notawful bought ββ (2) coffees

Thank you for sharing your visualization skills with the rest of us! I use them frequently when teaching intro stats.Β

Andrew J O'Neill bought β (1) coffee

Thanks for helping understand stuff!

Someone bought β (1) coffee

Someone bought β (1) coffee

Shawn Hemelstrand bought β (1) coffee

Thank you for this great visual. I use it all the time to demonstrate Cohen's d and why mean differences affect it's approximation.

Adele Fowler-Davis bought β (1) coffee

Thank you so much for your excellent post on longitudinal models. Keep up the good work!

Stewart bought β (1) coffee

This tool is awesome!

Someone bought β (1) coffee

Aidan Nelson bought β (1) coffee

Such an awesome page, Thank you

Someone bought β (1) coffee

Ellen Kearns bought β (1) coffee

Dr Nazam Hussain bought β (1) coffee

Someone bought β (1) coffee

Eva bought β (1) coffee

I've been learning about power analysis and effect sizes (trying to decide on effect sizes for my planned study to calculate sample size) and your Cohen's d interactive tool is incredibly useful for understanding the implications of different effect sizes!

Someone bought β (1) coffee

Someone bought β (1) coffee

Thanks a lot!

Someone bought β (1) coffee

Reena Murmu Nielsen bought β (1) coffee

Tony Andrea bought β (1) coffee

Thanks mate

Tzao bought β (1) coffee

Thank you, this really helps as I am a stats idiot :)

Melanie Pflaum bought β (1) coffee

Sacha Elms bought β (1) coffee

Yihan Xu bought β (1) coffee

Really appreciate your good work!

@stevenleung bought β (1) coffee

Your visualizations really help me understand the math.

Junhan Chen bought β (1) coffee

Someone bought β (1) coffee

Someone bought β (1) coffee

Michael Hansen bought β (1) coffee

ALEXANDER VIETHEER bought β (1) coffee

mather bought β (1) coffee

Someone bought β (1) coffee

Bastian Jaeger bought β (1) coffee

Thanks for making the poster designs OA, I just hung two in my office and they look great!

@ValerioVillani bought β (1) coffee

Thanks for your work.

Someone bought β (1) coffee

Great work!

@YashvinSeetahul bought β (1) coffee

Someone bought β (1) coffee

Angela bought β (1) coffee

Thank you for building such excellent ways to convey difficult topics to students!

@inthelabagain bought β (1) coffee

Really wonderful visuals, and such a fantastic and effective teaching tool. So many thanks!

Someone bought β (1) coffee

Someone bought β (1) coffee

Yashashree Panda bought β (1) coffee

I really like your work.

Ben bought β (1) coffee

You're awesome. I have students in my intro stats class say, "I get it now," after using your tool. Thanks for making my job easier.

Gabriel Recchia bought β (1) coffee

Incredibly useful tool!

Shiseida Sade Kelly Aponte bought β (1) coffee

Thanks for the assistance for RSCH 8210.

@Benedikt_Hell bought β (1) coffee

Great tools! Thank you very much!

Amalia Alvarez bought β (1) coffee

@noelnguyen16 bought β (1) coffee

Hi Kristoffer, many thanks for making all this great stuff available to the community!

Eran Barzilai bought β (1) coffee

These visualizations are awesome! thank you for creating it

Someone bought β (1) coffee

Chris SG bought β (1) coffee

Very nice.

Gray Church bought β (1) coffee

Thank you for the visualizations. They are fun and informative.

Qamar bought β (1) coffee

Tanya McGhee bought β (1) coffee

@schultemi bought β (1) coffee

Neilo bought β (1) coffee

Really helpful visualisations, thanks!

Someone bought β (1) coffee

This is amazing stuff. Very slick.Β

Someone bought β (1) coffee

Sarko bought β (1) coffee

Thanks so much for creating this! Really helpful for being able to explain effect size to a clinician I'm doing an analysis for.Β

@DominikaSlus bought β (1) coffee

Thank you! This page is super useful. I'll spread the word.Β

Someone bought β (1) coffee

Melinda Rice bought β (1) coffee

Thank you so much for creating these tools! As we face the challenge of teaching statistical concepts online, this is an invaluable resource.

@tmoldwin bought β (1) coffee

Fantastic resource. I think you would be well served to have one page indexing all your visualizations, that would make it more accessible for sharing as a common resource.

Someone bought β (1) coffee

Fantastic Visualizations! Amazing way to to demonstrate how n/power/beta/alpha/effect size are all interrelated - especially for visual learners! Thank you for creating this?

@jackferd bought β (1) coffee

Incredible visualizations and the best power analysis software on R.

Cameron Proctor bought β (1) coffee

Great website!

Someone bought β (1) coffee

Hanah Chapman bought β (1) coffee

Thank you for this work!!

Someone bought β (1) coffee

Jayme bought β (1) coffee

Nice explanation and visual guide of Cohen's d

Bart Comly Boyce bought β (1) coffee

thank you

Dr. Mitchell Earleywine bought β (1) coffee

This site is superb!

Florent bought β (1) coffee

Zampeta bought β (1) coffee

thank you for sharing your work.Β

Mila bought β (1) coffee

Thank you for the website, made me smile AND smarter :O enjoy your coffee! :)

Deb bought β (1) coffee

Struggling with statistics and your interactive diagram made me smile to see that someone cares enough about us strugglers to make a visual to help us out!πΒ

Someone bought β (1) coffee

@exerpsysing bought β (1) coffee

Much thanks! Visualizations are key to my learning style!Β

Someone bought β (1) coffee

You can sponsor my open source work using GitHub Sponsors and have your name shown here.

Backers β¨β€οΈ

#### Questions & Comments

Please use GitHub Discussions for any questions related to this post, or open an issue on GitHub if you've found a bug or wan't to make a feature request.

## Webmentions

There are no webmentions for this page

(Webmentions sent before 2021 will unfortunately not show up here.)

## Archived Comments (4)

D
David Fawcett 2019-07-30

Thanks for explaining that. I agree, your way makes more sense.

P
peter_schaeffer 2019-03-12

Hi, I tried this using Wolfram-Alpha. I entered

Integrate 1/((2*pi)**.5)*e**(-(x+.25)**2/2) - 1/((2*pi)**.5)*e**(-(x-.25)**2/2) from -30 to +0

This gives a value of 0.197413. This is the area between the left-hand normal curve (x+.25) and the right-hand normal curve (x-.25). The Integration stops at zero because that is where the two curves cross.

Does this look right to you?

S
strboul 2017-10-27

That's really smart.

K
Kevin Vo 2017-05-02

You mean that in terms of number of data point, the overlapping zone contains 80% of all data points. But when we consider it as an area under the curve, their shared area is 67% of the total area. It's very interesting to know that. And thanks for your OVL formula, I used effect size to visualize PISA data:
http://bl.ocks.org/ndvo2710...

Designed and built by Kristoffer Magnusson. Powered by Gatsby.