Correlation and Regression Flashcards
correlation
Measuring the association
between 2 variables
how do we graph correlation?
scatterplots plot the individual pairs of scores not means
positive correlation
Variables change in the same direction. The more studying, the higher the exam score.
negative correlation
variables change in the opposite direction. The more hours playing
video games, the lower the exam score.
no correlation
A change in one variable
is not related to a systematic change in the other. Studying is not related to exam score
how do you compute a correlation
pearsons r. When you have two variables that are on interval or ratio scale
pearsons correlation coefficient
r will always range from -1.0 to 1.0.
* The sign ( + or - ) tells us the direction of the relationship.
* The absolute value tell you the strength
* The strongest positive correlation is 1.0.
* The strongest negative relationship is -1.0
* If there is no relationship, the correlation
coefficient will be close to zero.
the closer the data is to fitting a straight line
the stronger the correlation
to compute a correlation, you must first do what
transform your raw scores into z-scores, because it puts everything on the same scale
What does the absolute value of the correlation coefficient tell us?
the strength of the relationship
The correlation coefficient tells us
the strength and the direction of the relationship.
statistical hypothesis of pearsons r
Ho: r = 0
Ha: r = 0
What is r squared
If you have a correlation between
two variables (e.g., r = .80),
squaring the r (r2 = .64) tells you
the amount of covariance between those variables.
what is covariance?
The amount of shared variance
between X and Y.
when 2 variables are correlated and we know one variable can we predict the variability of the other variable?
we can predict or “account for” a certain proportion of the variability of the other variable.
pearsons r what can it be due to
Correlations between two variables in your sample can be due to:
1.) chance factors (sampling error)
2.) They can reflect real relationships among your variables.
significance test
- Used to determine the likelihood of obtaining a certain r by chance.
- It will depend on the size of r and n.
- If it is unlikely that the r could have occurred by chance, we infer that it is a real relationship that would hold up if we tested the entire population.
curvilinear relationship
An increase in one variable leads to an increase in another—up to a certain point. After that point, an increase in x leads to a decrease in y.
What effect do the outliers have on a correlation scatterplot?
Can inflate or deflate the apparent
correlation between two variables.
Outliers are going to have a larger
influence in a small sample than in a large one.
Many different techniques can be used to diagnose and treat outliers in a sample.
What does Pearson’s r measure?
only measures linear relationships
what does a regression analysis do?
When we have two variables and they are correlated, a regression analysis gives you a more accurate level of prediction.