- Thread starter om0303
- Start date

you can express this in terms of Cohen's d, if you wish (mind that d isn't an effect

size here, but only a characterization of the sample, using a common effect size

measure; true "effect size" refers to populations), or maybe it makes sense to express

this in terms of the original scale (something like "it changed just 0.5 points on a

20-point-scale"). And then you state that "the observed mean change was not

statistically significant (p=...)".

HTH

Karabiner

(The variables are change in health anxiety, and resilience)

'There was weak, positive correlation between the two variables,

As well as speaking about the effect size, how would I interpret what I have found if the relationship was weak and not significant?

o variables,* r*(47) = .142, n = 49, *p* = .33; however the relationship was not significant (*p = *.332)'

As well as speaking about the effect size, how would I interpret what I have found if the relationship was weak and not significant?

As well as speaking about the effect size, how would I interpret what I have found if the relationship was weak and not significant?

You have NOT FOUND a weak relationship between the variables. What you have at hand is just some sample sample

data, and these show an association larger or smaller than the true association, that in the population . Now you use

your sample to infer statements about a population. The inference is: may I assume that H0 is true (H0: "there is

absolutely no association in the population (r=0), and the r=0.14 coefficient in my sample is completely due to

chance), or may I reject that H0. The emprical evidence based on r=0.14 & n=49 was not enough to reject H0,.

With kind regards

Karabiner