Please help me. I really know there are many differences between the tetrachoric and the pearson correlation, but I can't write in Rstudio. Actually, I don't know how to programming. Can you give some examples about programming about multiple choices in education reasearch?? Thanks a lot!
I'm a little confused by "at the same time". If you mean the correlation between more than two variables you can search for joint cumulants or the coefficient of multiple correlation (two slightly different things) but I suspect that's not what you mean. If you want the correlation between a binary variable and a numeric variable you want a point biserial correlation which is a special case of Pearson's. For unusual cases (eg. not your normal Spearman and Pearson) I would also be asking why you want correlations. The tetrachoric and polychoric are useful for generating correlation matrices from scale data in psychometrics to then do things like factor analyses but it doesn't mean I would be presenting them any time I had binary variables.
@@DrLyndonWalker I wanted to say for the case when it is necessary to create a matrix of correlation coefficients for three or more variables to do a confirmatory factor analysis for example.
@@MrLucianomattar I'm not sure if anyone has made a function to do this but if not, in the worst case you could use the appropriate functions for each pair of variables and construct the matrix yourself. If it was going to be a regular task you could write a function for it.
@@DrLyndonWalker I was curious to know if there was any method for this. I appreciate your suggestion, as actually doing an entire matrix by pairs is totally manageable.
I used this (thank you) and I have many empty cells. The program replaced the empty values with .5 (my binaries were 1 and 2). Will this give me incorrect responses or are those .5's ignored?
If you are giving the function a table then it won't matter how you code your variables, but if giving it raw data you can always recode to 0s and 1s. If you have table cells with zero counts then the 0.5 adjustment is unavoidable (note, this is cells of the table being zero, not individual observations being missing)
@@DrLyndonWalker Thanks! I gave it a table of binaries, 1 and 2, and it returned a big matrix (I have 50 variables I need to cor, its data cleaning). It does show warnings that it used .5 for blanks. It says .5 for zeros- but they are blank. I guess that .5 affects the results or I can rely on this matrix of correlations? My discipline insists on using this stat but no body can explain it so I really appreciate your video!
Please help me. I really know there are many differences between the tetrachoric and the pearson correlation, but I can't write in Rstudio. Actually, I don't know how to programming. Can you give some examples about programming about multiple choices in education reasearch?? Thanks a lot!
Thanks for the explanations. Please, would there be a method to make the correlation between binary, factors and numerical variables at the same time?
I'm a little confused by "at the same time". If you mean the correlation between more than two variables you can search for joint cumulants or the coefficient of multiple correlation (two slightly different things) but I suspect that's not what you mean.
If you want the correlation between a binary variable and a numeric variable you want a point biserial correlation which is a special case of Pearson's.
For unusual cases (eg. not your normal Spearman and Pearson) I would also be asking why you want correlations. The tetrachoric and polychoric are useful for generating correlation matrices from scale data in psychometrics to then do things like factor analyses but it doesn't mean I would be presenting them any time I had binary variables.
@@DrLyndonWalker I wanted to say for the case when it is necessary to create a matrix of correlation coefficients for three or more variables to do a confirmatory factor analysis for example.
@@MrLucianomattar I'm not sure if anyone has made a function to do this but if not, in the worst case you could use the appropriate functions for each pair of variables and construct the matrix yourself. If it was going to be a regular task you could write a function for it.
@@DrLyndonWalker I was curious to know if there was any method for this. I appreciate your suggestion, as actually doing an entire matrix by pairs is totally manageable.
I used this (thank you) and I have many empty cells. The program replaced the empty values with .5 (my binaries were 1 and 2). Will this give me incorrect responses or are those .5's ignored?
If you are giving the function a table then it won't matter how you code your variables, but if giving it raw data you can always recode to 0s and 1s. If you have table cells with zero counts then the 0.5 adjustment is unavoidable (note, this is cells of the table being zero, not individual observations being missing)
@@DrLyndonWalker Thanks! I gave it a table of binaries, 1 and 2, and it returned a big matrix (I have 50 variables I need to cor, its data cleaning). It does show warnings that it used .5 for blanks. It says .5 for zeros- but they are blank. I guess that .5 affects the results or I can rely on this matrix of correlations? My discipline insists on using this stat but no body can explain it so I really appreciate your video!