PT 551 Correlation and Regression

significance tests, correlation coefficient, slope

what statistics can help us describe how well two variables correlate

scatter plots

A graph with points plotted to show a possible relationship between two sets of data.

curvilinear relationship

a relationship between two variables whereby the strength and/or direction of their relationship changes over the range of both variables Im

correlation coefficient (r)

The strength of covariation between two variables

-1 to +1

possible values for correlation coefficient (r)

perfect negative

what kind of relationship is indicated by r = -1

0-.25

GENERAl correlation coefficient values for little to no strength

0.25-0.50

GENERAl correlation coefficient values for fair to moderate strength

0.5-0.75

GENERAl correlation coefficient values for moderate to good strength

0.75-1

GENERAl correlation coefficient values for good to excellent strength

significance

Tests of _____________ estimate the probability of observing a given correlation by chance

critical r value

If our calculated r value exceeds the this value then we reject H₀ and assume that we have a legitimate correlation.

small

Would a large or small sample require a more perfect correlation in order to reject H₀

n-2

Degrees of freedom (df) for correlations calculation

decreases

As sample size increases, what happens to the critical r values

Pearson

Correlation coefficient for two continuous variables (both interval or ratio); most common

Spearman

Correlation coefficient for two continuous variables (one or more ordinal)

Point-Biserial

Correlation coefficient for one continuous variable and one dichotomous variable

Phi

Correlation coefficient for two dichotomous variables

t-test

what is a point biserial correlation coefficient equivalent to?

2x2 test

what is a phi correlation coefficient equivalent to?

false

True or False? The CI95% for r is symmetric like the CI95% for means

coefficient of determination

the degree to which the variability in our dependent variable can be explained by variability in our independent variable

0 to 1

coefficient of determination ranges

correlation matrices

When a researcher reports correlations for more than 2 variables measured in a study

extraneous variable

In an experiment, a variable other than the IV that might cause unwanted changes in the DV.

spurious correlation (5% prevelance)

an apparent but false relationship between two (or more) variables that is caused by some other variable

interpolation

estimating a value within the range of measured data

extrapolation

Estimating a value outside the range of measured data.

MCID

when looking at slope of correlations, what should you be comparing it to?