SPSS: Calculating a Correlation between a Nominal and an Interval Scaled Variable
ฝัง
- เผยแพร่เมื่อ 5 ก.พ. 2025
- This short video details how to calculate the strength of association (correlation) between a Nominal independent variable and an Interval/Ratio scaled dependent variable using IBM SPSS Statistics. In particular, we measure the Eta statistic for the strength of association.
Thank you! I got thoroughly stuck in my dissertation and you may have un-stuck me. I searched most everywhere for this particular piece of information. Thank you, again.
This was so helpful. I have been trying to figure out how to correlate the interval and nominal data for two for two of my dissertation questions. I am very grateful. :)
its a very nice video that solves my question which makes me feel panic for so long! Thank you!
Thank you for this. It took me many days trying to search for this stuff
Hi Muhammad. Thank you for your nice comment. Regards. Jonathan.
how do we calculate p value for this test
can't thank you enough for this!
Thank you so much for this explanation! Simple and very helpful. Definitely saved my life :)
Good day. Thank you for continuing to educate us all in the uses of statistics in research. One thing i want to know is, what statistical tool in determining the relationship between nominal and ordinal variables. thank you
Hi Jay-R. This following playlist provides an overview of all the correlation coefficients, based on the scale of measurement associated with both variables. I hope this helps:
th-cam.com/play/PLJy0LHDLpgHGQkyy0D8t93mOWil-Z7ETW.html
Regards. Jonathan
Thank you so much for this video. It helped me so much with some of my dissertation data analysis.
omg this is what i need thank you so much
Hello. I hope this question be answered.
If the eta value is calculated, how do we know that the correlation is significant? I mean, how to calculate the p-value of eta coefficient?
Hi Randy. The typical approach is to carry-out a between subjects ANOVA. The significance of that test is equivalent. You would really do this using software, maybe through SPSS or with R. I hope that helps. Jonathan
Thank you so much for this explanation, this is very helful to me. I would like to know that if eta correlation can be negative in SPSS?
Thank you
So I have 2 variables: the first is a scale and says how much hours of sport an athlete did, the other is a nominal one and says whether or not they got injured. My Scale is my independent variable here. But this should still work right? And I can square the Eta and that will tell me how much the variance can be explained by the difference in hours of sport?
thank you
Thanks so much. Great video and wonderful explanation.
Thank you. It was excellent 👍
Thank you for the video. But I would like to ask you if it is possible to obtain the p-value of the ETA statistic? If not, how could I check if the ETA statistic is significant?
Did you find out?
Same question. How?
This is very helpful. Thanks! What is the equivalent technique if a categorical variable has > 2 categories?
Hi Brian.
In that situation you would be dealing with a multichotomy and an interval/ratio variable. The appropriate correlation is known as the eta correlation coefficient. Here is a link to a playlist in which I cover all the possible correlation variants. I think a little into video part 4 should be what you are looking for.
th-cam.com/play/PLJy0LHDLpgHGQkyy0D8t93mOWil-Z7ETW.html
I hope that is helpful. Regards.
Jonathan
I want to find correlation between education levels and electricity bills. The electricity bill is continuous while the education level (primary, middle, secondary, higher secondary and graduation) nominal. Should i go for checking the assumptions of linearity, normality etc or i use only Spearman's correlation directly? Thank you sir.
Hi Waqar. yes, you are correct. The Spearman Rank Correlation would be most appropriate. Here is a video playlist that should help:
th-cam.com/play/PLJy0LHDLpgHGPf1AqPIQJThoQLmeqBIWu.html
Regards.
Jonathan
Thank you, very informative and precise!
hi sir, you have explained the concept so well but just as a feedback i have one suggestion that please try to keep your rate of speech delivery little slower than current because all you have explained here is important information and one has to keep every word in mind during the analysis.
Thank you
Hi Manish. Thanks for the constructive suggestion. I will certainly try to slow things down for my next collection of TH-cam tutorials. Kindest regards. Jonathan.
Thank you sir for this valuable information.
Thank you. This is really helpful
Thank you so much, your video has greatly helped me.
Thanks and great video. I would like to know if our scale variable is not normally distributed then this test can also be applied or another non parametric test is there instead of this one?
So what about when you want to figure out the relationship between the nominal variable and the scale (for example, no of serious relationships related to a particular personality trait). Like, you interview 150 people, ask them how many serious relationships they had in their lifetime (if any... :) ), and then you also ask them some more questions to establish the strength of each of the big five personality traits. Then you want to correlate the number of serious relationships (nominal value, lets say 0 to 6), to an extroversion trait (that is a scale). How would you analyse that in SPSS? Would love an answer, if you are still reading this post, that is.....
How can you determine if it's statistically significant? I have a lot of small associations in my SPSS output and want to determine if they're statistically significant or not.
Does this apply to gender (nominal) and Likert scale data (ordinal)? I'm trying to work out the exact method to work out how gender affects likert scale choices in my survey.
Thank you!
but how we will know that whether it is significant or not?
Thank you for this helpful video. I would like to ask sir if there's a reference for the values indicating the strength of correlation like in the Pearson R and Spearman Rho?
www.psychology.emory.edu/clinical/bliwise/Tutorials/SCATTER/scatterplots/effect.htm
You explained very well
Hi, how can I know when to reject or accept the null hypothesis?
May I know how to get P value of this ETA test? Thanks
maybe do a 1-way or 2-way ANOVA
Great video. Two questions: can I use the ETA for a continuous variable which is not normally distributed as well? And does it work for all nominal variables (e.g., dichotomous, ordinal)?
Hi CH.
First, and I'm sure it was just a typo, the ordinal variables are not nominal - as they have an order associated with them.
With respect to the real nominal variables, they are broken down into two cases, those - as you suggested - that are dichotomous and those that are multichotomous. Here is a link to a playlist I had created that runs through all the possible combinations of variable types and their appropriate correlations (although there are other variants):
th-cam.com/video/EpOqsKCIHNU/w-d-xo.html
I hope this helps. Regards.
jonathan
How about partial correlation which X and Y are numeric variable while controlling Z which is nominal data ? Please help.
Thanks for a great video. How can I perform a causation analysis with the same type of data set?
Thank you Johnathan very helpful
Thanks for the great video. But how should one fine the p value???
...but how do we know the significance level? in other words, how likely is it that the null hypothesis of no difeerence between genders is correct?
How would I Analyze the relationship between the [Marijuana] and [sex] variables, and include a statistical significance test and two appropriate measures of association.
Hi, will this change anything if my nominal variable is composing of multiple options? Like let's say for gender we have Male, Female, non-binary, aesexual, etc...?
Does ETA work for a scale independent variable e.g. sleep hours and a nominal dependent variable e.g. existing cardiovascular disease? If not, what would be the test?
very helpful indeed, thank you
With no hope, because there is no questions answered here yet. Can I get the example data, so I can try it myself
Thank you for the video. But i followed exatcly the same steps and the output window says "Directional Measuresa
a ETA statistics are available for numeric data only.
Can someone help? Thank you.
@@ImranKhanCardiac you might have put in a variable that isn't numeric. If one of your variables contains string values (words) then it won't work
Unfortunately this technique requires "Nominal" *Numeric* variables, i.e; ones with value labels defined @2:15. It won't work with String (truly nominal) variables. This makes it much more annoying to use, and it sounds like Spearman should be selected instead of Eta for non-dichotomous "nominal" variables.
But is it significant statisticly?
Thank you so much for this video, it was very clear and helpful. I would also like to know how to calculate the correlation between a nominal variable (with more than two categories, like industry types) and a continuous variable
Did you find how it was done?
could you provide a reference for the values indicating the strength of correlation?
thank you!
Where is the p-value?
where is the P-value. We can't say it's associated unless we know that it's statistically significant, to begin with ?
You can tell by direction. What p-values does is to tell you whether the association is statistical valid or not based on the information (data) you have used.
2 genders, male and female. That is a contentious assumption to make these days!
"Zero indicates female, one indicates male" - This is sexism (prejudice or discrimination based on a person's sex or gender)!
it's not sexism, it's coding haha !
@@laurencarterYT 0