The Secret Trick to Calculate Cohen’s d in SPSS - It Can Be Done (11-9)

Cohen’s d is the most widely reported measure of effect size for t tests. Although SPSS does not calculate Cohen’s d directly, there are two ways to get it. First, you can enter your SPSS output into a downloadable calculator that will give you the effect size. Second, (at 8:30) you can use a little known workaround to force SPSS to calculate a Cohen’s d for you.
This video teaches the following concepts and techniques:
Cohen’s d effect size for t tests
IBM SPSS Statistics software
Link to a Google Drive folder with all of the files that I use in the videos including the Effect Size Calculator for t Tests and datasets. As I add new files, they will appear here, as well.
drive.google.c...

[@rolanliteratus7757]

You, sir, is lit. I've never had as much fun and excitement with statistics until now. Thanks to you and your awesome videos.

[@ResearchByDesign]

Thank you so much for the comment. It is gratifying to know that I am, in fact, lit. :o)

[@Isuppose12]

This is the loveliest video for Stats that I have ever watched :) Thank you so much!

[@LeonAruba]

I agree soo much with you! Made my day.

[@ResearchByDesign]

Wow, thank you! And I hope you find many more on the channel !!

[@magdacj1]

Thank you so much for this crazy jump of Cohen's relationship (the mathematician and the musician). I was thinking the same at the same time as you! Fantastic! I would need to learn more about the effect size but I am not going to forget this lesson ;P

[@manda9074]

I didn’t come here to smile 😂😂 I came here cause I’m stressed lmao wtf 🙏🏿❤️ but thank you!!! For both

[@ResearchByDesign]

Very happy to hear that. Glad that we can bring a smile with our silly videos (silly and serious, of course). Wish you the best in finals week!

[@mr.fancyraccoon6936]

i was just here preparing for a biostatistics exam and i click randomly in the middle of the video and i get a song about not worrying about my p being small. I love the internet.

[@ResearchByDesign]

...and if I ever post the rest of the song, you will love this little slice of the Internet even more. Bottom line: when comparing groups, you don't want a small p, you want a large d...Cohen's d, that is. (Love your icon, BTW)

[@mr.fancyraccoon6936]

@@ResearchByDesign Never shall those your escape me. Know that you've made a particularly fancy raccoon mildly more competent at statistics.

[@manhattannewyork8660]

the best teacher I have ever seen!

[@ResearchByDesign]

Such a wonderful comment...thank you so much!

[@craftychinurse5941]

LOVED the song! Thank you for a little piece of joy!

[@ResearchByDesign]

Maybe someday I will post the entire song...a buddy helped me do about half a dozen of them but they don't have videos. We'll see...

[@lauramaria1348]

lol this made my day

[@shawsr7]

Brilliant presentation, especially the music.

[@taladiv3415]

2:05 That's just great!

[@Click-hr3bs]

Truly, I'm blessed for that musical

[@ResearchByDesign]

I am deeply, deeply moved knowing how you appreciate the seriousness and weight of our musical explanations. :o) (It's great that someone does) thanks for commenting.

[@admirebetera6544]

Wow! I enjoyed the lesson. How can I forget this when it is accompanied by a nice song. kkkk. Thanks!

[@ResearchByDesign]

Glad you liked it!

[@kentsfu]

Thank you so much for sharing the Excel of effect size. It's really helpful.

[@ResearchByDesign]

You're very welcome!

[@stayfocused4500]

Thank you very much for sharing the tricks. It is very clear. I also like the song!

[@MichaelTeagueFF]

Love the song. Made me laugh

[@ResearchByDesign]

Thanks for the comment about the Cohen's d song. There is more to it and we keep saying "we have to make a music video", but have not made time to do it, yet. Hopefully, some day...

[@PrettyishSwan]

@@ResearchByDesign sooo where can we get this song?? really like to download it hahaha

[@hyacinthenyandwe8644]

So funny and educative at the same time🤣🤣🤣

[@JuanDelaCruz-ekmg]

Thank you for making these informative videos. I just have one question. There are instances where we are assuming unequal variances which does not use the pooled standard deviation. What happens to the Cohen's d formula?

[@maritstoffers9349]

This is f*cking amazing!! Many thanks!!!

[@digitalmarketingandanalyti4819]

Heheh lovely song, nice voice .. thanks for the video ..

[@jonesgirleleven]

this is the best thing ever made. thank you.

[@ResearchByDesign]

Thank you for saying so...much appreciated!

[@blufaer1e]

This song is amazing. Is there a place I can stream or download it?

[@annew7043]

Is the effect size song available anywhere else on youtube? I'd love to forward the song itself to fellow nerds! :)

[@Maguii_BalV]

Can I find the Leonard Chohen's d "I'm a fan" album on Spotify??? please say yes!!!!!! I'm doing my thesis and my p value was low, and that song made me feel better since by calculating effect size I'd be able to explain why I have that result.... for real, thanks!!!!

[@ResearchByDesign]

Thanks to your question, I looked into Spotify, but I would need a music agent or promoter or something to get on there. I am working on a video with the Leonard Cohen's d song, because so many people have requested it. I also have a script about how to publish non-significant results that emphasizes effect size and power analysis. I'll get those finished so you can enjoy the entire song...it gets much funnier as it goes along.

[@Kimarievy]

Hi, why don't you square the standard deviation in the example problem when the formula you provide squares the standard deviation? (Independent samples formula)

[@tessagarcia-collart4787]

This is perfect, thank you!

[@armandokaito3862]

Very well explained. Thank you!

[@ResearchByDesign]

Glad it was helpful!

[@tiasada]

Will this work for calculating Cohen's d for an ANOVA as well?

[@ResearchByDesign]

For ANOVA, report eta squared (for one-way ANOVA) or partial eta squared (for more complex ANOVA designs). Look for my video "12-7 Effect Size for One-Way ANOVA" for how-to.

[@amenazehraali414]

Thank you so much... This is great!

[@younesaissaoui8442]

Thank you for this excellent tuto. Could you please show me how to calculate the Effect size and the standardized response mean for repeated measures in SPSS? PS: I am not a statistician.

[@ResearchByDesign]

Great question. Repeated measures effect sizes require a different formula to account for the correlated nature of the variables. I can't replicate the formula here but I have course notes I could share with you. Search for me on the Missouri State University website, drop me an email, and I will send you the notes. Best!

[@rafaelbernardon]

Thanks a lot! Awesome, helped a lot.

[@maryamsabzevarzadeh2318]

Thank you!!! it was so helpful :)

[@ResearchByDesign]

You're so welcome!

[@laurengourlay296]

Great vid - do you have one for repeated measures?

[@ResearchByDesign]

Yes, there is a video on repeated measures t test. For calculating Cohen's d on repeated measures, you should use a correction for the correlation of the groups. I have a spreadsheet for that...I think that it is in the links below the video

[@zzero9438]

LOOOOOOVE THIS

[@iaoiadnaa2541]

How do I find the song about tge effect size? 😅

[@stephenmarshall5529]

I went to Rstats Institute and couldn't find the calculator. Hope you can link or make it easier to find.

[@ResearchByDesign]

No worries...the old link is gone but there is a link to a Google folder in all of the video descriptions. You can find the calculator through that link. Right-click to download. Best!

[@malvane8061]

so much double entendre in your bands song

[@stephenmarshall5529]

In my class problem I found p>.05 so we couldn't reject the null, but using your procedure (spss25) I got -1.5, a strong effect. I'm sure I did something wrong, so I wanted to use your Rstats calculator. See my below comment.

[@ResearchByDesign]

Finding a large effect but non-significant results is most likely to occur when you have a small sample size. Small n will also affect this technique that I illustrate in this video. So you are probably more correct than you know, but you need more subjects in your sample to find the large effect that you can tell is there (from the effect size).

[@HanySalem]

How can I find Cohen's d on RStat

[@jesusleonardosanchez1642]

Thanks man, it's amazing

[@ResearchByDesign]

Glad you like it!

[@yza4896]

high how do you calculate cohen's d for paired t-test? I am trying the same method but i don't see it works in that case

[@ResearchByDesign]

That is a great question...and one for which I need a video. There is a correction factor that you need to include for a Cohen's d when using repeated measures. The typical Cohen's d formula overestimates the effect size.

[@FurqanKiAwaz]

it's off topic sorry but, can we use mediation in OLS Regression? how it works

[@ResearchByDesign]

Yes, you can use mediation (and moderation) in regression. I would suggest looking at "PROCESS" by Hayes (processmacro.org). It is an SPSS plug in...the "how to" question requires a whole book, however... good luck!

[@oceanpacific21]

Have you released a music album yet?

[@ukasz799]

Thank You very much

[@ResearchByDesign]

You are welcome

[@Benny7F]

Video starts: 8:30

[@ResearchByDesign]

Thanks...there is a run up to the actual technique.

[@drMshayisa]

I love the song, is it available online?

[@ResearchByDesign]

I had to go back to the original files to look this up...the song is called "Cold World" and it is royalty-free music from Digital Juice

[@rishikamalik4752]

wow that song was surreal

[@avandyke143]

Ha! Loved this!!

[@renedavidtacurireino7340]

Genial muchas gracias 😃😉

[@ResearchByDesign]

De nada. Gracias por ver los videos.

[@bubblezpop77]

hello! is there a way to do this for linear mixed models?

[@ResearchByDesign]

If there is a way to do it with mixed linear models, I don't know it off hand. Sorry.

[@tuedang1144]

the song is implying about my thesis. Weird! Thank God...

[@zompopo69]

Thank you. It could have been great if it worked. It doesn`t for me either. And I have tried with different variables. For example: group A with n=24 DV mean = 10.33, SD=6.42 and goup B with n=26 DV mean = 16.96, SD=9.93. With your method using SPSS gives me a d of -.610. But in online calculators give -.793. And with the formula it gives me -.786. Looks like or I am doing something wrong or this method is not so precise.

[@ResearchByDesign]

I double-checked with a randomly created dataset and think I see what is happening...I have not thought through all the math, but the issue seems to be sample size (same problem Student solved with the t-test, actually). With samples smaller than 30 (you had n's of 24 & 26), there is enough error to skew the results. Non-normality of the data may also be playing a part, so check for skewness. That may be what accounts for the difference between your online calculator and the formula, too. So the take-away is that this technique works best with samples larger than 30. Remember, too that we interpret Cohen's d to only the first decimal. Try this: duplicate your data 3 times so you have the exact same data repeated three times (n= aprox. 75 in each group). Your mean will remain the same, SD will drop a little, and the Cohen's d will approach what you got from the formula. If so, then you know the issue is the sample size. Good luck and thanks for the comment!!!

[@zompopo69]

@@ResearchByDesign Wow! thank you! I'll try. And I'll tell you back. May be tomorrow, It is late in muy place now. By the way, nice song and incredibly useful chanel. Congratulations!

[@zompopo69]

Sorry, but it doesn`t work. I have tried with some more variables that meet the normality criteria and in the most of them the Cohens d was different from the one obtained with the formula. Some were similar, but some others were too diffentent. I think that I keep doing it with Excel. Thank you anyway, and please keep trying.

[@ResearchByDesign]

Bummer that it did not work...I give you credit for trying. I played around with some additional data in the mean time. I'm pretty sure that the sample size is the culprit. Because of the z score conversion, this is just not going to work unless your n is fairly large. Thanks for teaching me something!

[@tomaszlipinski1870]

You are boss thank you :)

[@ResearchByDesign]

You're welcome!

[@roldancardona5335]

can we download your excel calculator for effect sizes?

[@ResearchByDesign]

I updated the link to the effect size calculator. I think that it is working now. The link is to a google folder with files and data sets for the course

[@albertoolivero9740]

How can I do with categorical variables?

[@ResearchByDesign]

It is not so easy with categorical variables because Cohen's d is about standard deviations. Obviously, there is not an SD with nominal categories. Knowing what to do would require examining your entire research design. Wish you the best.

[@ResearchByDesign]

This is the closest thing to effect size with categories: kzbin.info/www/bejne/qYupaJmng7SYbbs

[@kevinshen3221]

8:30 for those only want to know how to use spss

[@ResearchByDesign]

Thanks for that note...I edited the description to include this time code. Some people already know the theory and want to skip right to the practice. Thank you

[@erinprunskus2704]

This is hilarious!

[@user-ti3zg5kz6e]

New version just click Cohen d box

[@erinprunskus2704]

Loving the inappropriate banana :-)

[@ResearchByDesign]

What? Where? I had no idea. How did that happen?!? :o)

[@uliahshafar8599]

Lyiric: an effect size an effect size oh sorry I thought this is a youtube music.