point-biserial correlation?

#1
Hi!

I am doing my thesis about life-events (both positive and negative), social support and anxiety.

I want to find out which types of negative life-events that are associated with anxiety. The negative life-events are nominal variables and anxiety are scale variable.

First I did a pearson correlation with all items measuring negative life-events (nominal variable) and anxiety (scale variable), but this seems to be wrong? Is point-biserial correlation the correct analysis to do, in order to find out the association?

Thank you in advance :)
 
#2
Hi, the point-biserial correlation should be used if one of the two variables is dichotomous, the other is continuous. Does the negative-life-events-variable really have only two categories? And the anxiety variable is true continuous or rather ordinal?
 
#3
Thank you for your answer!

The life-event variable is either scored as positive or negative based on the score on the Likert-scale. The likert scale range from 1 = Very good to 6 very bad. In SPSS, scores from 1-3 are scored as positive events, whereas scores from 4-6 are scored as negative events.

The anxiety scale is also scored on a Likert scale, were the total anxiety variable is a scale variable.
 
#4
Hi,

so you have two ordinal variables which you want to correlate. In this case, you could use the Spearman correlation coefficient or, alternatively, Kendall's tau.