Correlation evaluation

[Zarathrustra]

Is there a single statistic which can reflect the different importance between similar correlations, say 0.7, based on different sample sizes such as n=10 and n=1,000?

 

[hlsmith]

Don't just look at the correlation coefficient, but also examine the 95% confidence intervals. Those will incorporate sample size data into your interpretation.

 

[Zarathrustra]

Thanks hismoth.

How do I then use this stat? In my case, I have something like 300 independent variables, all with an n=66, and with confidence intervals varying between 0.0015 and 0.50. So could I reasonably say that correlations above (say) 0.5 with a confidence interval of below 0.01 are more indicative of a meaningful relationship than one of 0.5 and a confidence level of 0.05?