correlation

Correlation analysis asks what question

whether or not 2 variables change in a related manner

what steps are used to decribe how well 2 variables correlate

1) significance test
2) correlation coefficient
3) slope of the line

how are correlations graphed

scatter plots

how can we manipulate data that has a clear relationship that is not linear, and create a linear relationship

manipulate one or both variables by rank ordering or taking the log

correlation coefficient (r)

the strength of covariation between 2 variables

range for r

-1 to 1

ballpark value for r that indicates little to no relationship

0-0.25

ballpark value for r that indicates fair to moderate relationship

0.25-0.5

ballpark value for r that indicates moderate to good relationship

0.5-0.75

ballpark value for r that indicates good to excellent relationship

0.75-1

what variables are expected to have a higher r

simple/direct relationships

what variables are expected to have lower r

constructs

how is df calculated for r

n-2

do we reject or accept the null if the r exceeds the critical r value

reject

as sample size ____, critical r _____, and ____ r is significant

increases, decreases, smaller

correlation coefficients used for interval/ratio data

pearson

correlation coefficients used for ordinal data

spearman

correlation coefficients used for one continuous variable with one dichotomous variable

piont-biserial

correlation coefficients used for 2 dichotomous variables

phi

correlation shows ___, not ____

covariation, not matching

extraneous variables

another variable, other than the one we are studying, that can cause a correlation

spurious correlation

2 variables correlate by chance