Which statistical test to run on SPSS?? Help me pleassse!!

[char12710]

Hi there,

I am wondering if you might be able to help me on a problem I am having with my psychology project.

I am running statistical analyses on some results for a psychological experiment. At the moment I have a personality score (conscientiousness) which can range from 1-5 and an opinion score which ranges from 'definitely no' (1) to 'definitely yes' (4). I want to compare the two and see if having a higher personality score means you are more likely to say 'definitely no' on the opinion score. Which statistical test do I need to use to verify this??

Thanks!!!

 

[gianmarco]

Hi,
please note that I am not familiar with this kind of tests, nonetheless I believe that Spearman correlation could be used; see this link and the example therein:
http://www.statpac.com/statistics-calculator/correlation-regression.htm

Besides, May be that Kendall t can be' used as well:
http://en.m.wikipedia.org/wiki/Kendall_tau_rank_correlation_coefficient

regards
gm

 

[gianmarco]

(I was meaning that I am not familiar with those kind of experiments....)

gm

 

[TerriP]

Depends if you think there is causation between the variables (in which case Spearman would be appropriate) or just correlation without causation (in which case Pearsons Product Moment would be the right test).

 

[tstats]

correlation of any kind (spearman or pearson) does not prove causation. Since both variables are ordered, spearman correlation is better.

 

[DavidBill]

Hmmm.. I apologize if I'm missing something, but as I read the poster's description:

I am running statistical analyses on some results for a psychological experiment. At the moment I have a personality score (conscientiousness) which can range from 1-5 and an opinion score which ranges from 'definitely no' (1) to 'definitely yes' (4). I want to compare the two and see if having a higher personality score means you are more likely to say 'definitely no' on the opinion score. Which statistical test do I need to use to verify this??

It sounds like (s)he is looking to get the probability of a certain score on the dep variable (definitely no) rather than a general correlation between the variables across all values of the dependent variable.

If this is correct, what about recoding the dep variable as 0,1 (1=definitely no, 0=the other three values) and running a probit or logit?

(And FWIW, this sounds like observational data rather than data from an experiment.)