Pearson's Correlation

madeleine.jonsson · Post by **madeleine.jonsson** » Wed Jul 20, 2022 4:04 pm

Hi!
I am currently looking into how I could do a correlation analysis of my research data and have been suggested to use Pearson's correlation analysis. My data set consists of 47 different analyses with a different number of replicates (n8 - n72). Looking at Pearson's equation, it would not be possible to perform the analysis as correlation is done pairwise. However, Jamovi is able to generate these Pearson's r-values. Could you please help me understand how that works? What algorithm is used? Are any values discarded? Or how does Jamovi compensate for the different variable lengths?

Thankful for your help,
Madeleine

Bobafett · Post by **Bobafett** » Thu Jul 21, 2022 1:23 pm

I'm not quite clear on what you mean by replicates?

Correlations will only be performed when participants have a pair of scores. Anyone not meeting this criteria is automatically excluded from the analysis.

If you are asking if the variables themselves have different lengths (e.g. they are composed from questionnaires with a different number of items) then this is immaterial to the analysis.

madeleine.jonsson · Post by **madeleine.jonsson** » Fri Jul 22, 2022 1:22 pm

Thank you for your response, Bobafett!
By replicates, I mean variables - which are of different lengths in my data set. So, do I understand you correctly: if the variable lengths are not matching, then any residual variables are excluded from the analysis? In the example below, would the values for xa3 and xb3 be excluded from the correlation analysis?

Code: Select all

     x       y 
a    xa1     ya1
     xa2     ya2
     xa3     - 
b    xb1     yb1
     xb2     yb2
     xb3     -

Post by **jonathon** » Tue Jul 26, 2022 4:59 am

hi,

correlation is performed on pairs of values ... if one or both of the values are missing, then those are excluded from the analysis.

jonathon

jamovi