Thank you for showing us how to interpret the correlation. My teacher did not show us, and I was confused about how to interpret the correlation of two different levels of data. Your explanation was very helpful!
@lexcle so agree with you lol ... thanks for posting this youtube stats guy... really helped at 11 44pm in the nite... no stats teachers on call this time of day lol
Yep, Pearson' r is mathematically equivalent to the formal point-biserial correlation. Go into google books and type 'point biserial correlation Pearson correlation' and you should plenty of references. Phi in the 2x2 Pearson chi-square case is also mathematically equivalent to Pearson's r.
Hey I have a question. If you have a control and experimental group (both of which let's assume are in the same school). How can I inhibit their interaction?
Do you mean you have nominal data with three ore more categories (e.g., Canada, USA, Australia), or that your variable has three or more levels that are ordered somehow (Agree, Neither Agree/nor Disagree, Disagree)?
@@how2stats I'm two years late but hoping you respond...I am also looking for the answer to this question..both with regards to nominal data with three or more categories (university departments) and ordered data (groups at different levels of education)...thank you!
Pearson's is remarkable. It can handle dichotomous IV and continuous IV. People call it point-biserial correlation in that case, but it's really just a Pearson correlation.
How about a nominal variable and an interval variable? For instance: 1 = North 2 = South 3 = East 4 = West And then the other variable is population? So I am asking does population correlate with location? My hypothesis is that the north has a higher population. Thanks
I think in that situation, we would do Nominal by Interval Eta. A measure of association that ranges from 0 to 1, with 0 indicating no association between the row and column variables and values close to 1 indicating a high degree of association. Eta is appropriate for a dependent variable measured on an interval scale (for example, income) and an independent variable with a limited number of categories (for example, gender). Two eta values are computed: one treats the row variable as the interval variable, and the other treats the column variable as the interval variable. Go to Analyse>Descriptives>Crosstabs, then click Eta. I hope it would work.
Seriously? All you do to run a point-biserial correlation is run a regular Pearson? I sure hope this is right, cause it's what I'm gonna say in my thesis defense.
It doesn't matter for the actual analysis, as SPSS doesn't take note of that information. It's more for the user's benefit, or of you use one of SPSS's guided approaches to analysis (which I don't).
![Point-biserial Correlation [Simply explained]](http://i.ytimg.com/vi/-nmPEsYzaWo/mqdefault.jpg)
Thank you for help, I was so confused on how to calculate my data but you made it so easy!
Best stats channel, hands down.
Thank you for showing us how to interpret the correlation. My teacher did not show us, and I was confused about how to interpret the correlation of two different levels of data. Your explanation was very helpful!
Thank you so much for your presentation. I understand this technique clearly.
this man was my lecturer at uni, nice man
Thank you very much! I love it!
Thanks! Video definitely works better for us as compared to notes and writing~
Very useful. Thank you very much. :)
thank you so much for this video... :)
Thank you so much!!!
@lexcle so agree with you lol ... thanks for posting this youtube stats guy... really helped at 11 44pm in the nite... no stats teachers on call this time of day lol
Thank you !
Yep, Pearson' r is mathematically equivalent to the formal point-biserial correlation. Go into google books and type 'point biserial correlation Pearson correlation' and you should plenty of references.
Phi in the 2x2 Pearson chi-square case is also mathematically equivalent to Pearson's r.
Thanks!
thanks so much for this video, i got a better of my data now :) its just a bummer that i cant use you as reference :D
You just use the regular Pearson correlation on the relevant variables.
yes. thanx
thank you
Terimakasih pak
Hey I have a question. If you have a control and experimental group (both of which let's assume are in the same school). How can I inhibit their interaction?
In the video, both variables are key in as an interval scale.
Is that what you have to do?
It's a useful video, i was wondering how can i calculate point-biserial correlation coffecient in SPSS, thank you..
Great vid. I don't suppose you know of a way to run a rank-biserial correaltion (ordinal vs. nominal)?!
ETA is equivalent too??
How would you interpret a positive ? higher numbers of brain volume correlate with higher value of group variable?
The group with the higher number has greater brain volume.
What if you had non-dichotomous data, what statistical tool should you use?
Do you mean you have nominal data with three ore more categories (e.g., Canada, USA, Australia), or that your variable has three or more levels that are ordered somehow (Agree, Neither Agree/nor Disagree, Disagree)?
@@how2stats I'm two years late but hoping you respond...I am also looking for the answer to this question..both with regards to nominal data with three or more categories (university departments) and ordered data (groups at different levels of education)...thank you!
Do you still need to know this? I have worked out a way using Excel. Easy once you set the spreadsheet up!
what if the data not normal distribution?
thank you
+Khilyat Zaini Your best option is likely to use bootstrapping: ruclips.net/video/9VjzPnoUBJQ/видео.html
@@how2stats how about rank biserial though?
Why do we use Persons and not Spearmans. Isn't the data non linear but monotonich ?
Pearson's is remarkable. It can handle dichotomous IV and continuous IV. People call it point-biserial correlation in that case, but it's really just a Pearson correlation.
How about a nominal variable and an interval variable?
For instance:
1 = North
2 = South
3 = East
4 = West
And then the other variable is population? So I am asking does population correlate with location? My hypothesis is that the north has a higher population.
Thanks
I think in that situation, we would do Nominal by Interval Eta. A measure of association that ranges from 0 to 1, with 0 indicating no association between the row and column variables and values close to 1 indicating a high degree of association. Eta is appropriate for a dependent variable measured on an interval scale (for example, income) and an independent variable with a limited number of categories (for example, gender). Two eta values are computed: one treats the row variable as the interval variable, and the other treats the column variable as the interval variable. Go to Analyse>Descriptives>Crosstabs, then click Eta. I hope it would work.
Kamran Ahsan This was exactly my question ;) thx. :D I knew I want Eta but I couldn´t find it! ;)
So wait, to perform a point-biserial correlation in SPSS, you do the exact same thing as you would to perform a Pearson correlation in SPSS?...
+Paul Partlow Yes, exactly.
+how2stats why not call it a pearson? lol
It is a special case of the Pearson’s product-moment correlation
Seriously? All you do to run a point-biserial correlation is run a regular Pearson? I sure hope this is right, cause it's what I'm gonna say in my thesis defense.
Variables are set up incorrect. It should be one scales other nominal
It doesn't matter for the actual analysis, as SPSS doesn't take note of that information. It's more for the user's benefit, or of you use one of SPSS's guided approaches to analysis (which I don't).