Coorelation Flashcards
Measuring correlation
Using “r”
The closer you are to 1.00 the stronger the relationship
The closer you are to 0 the weaker the relationship
0=no correlation
+0.87
Positive=direction 0.87=magnitude (strength)
Correlation
Can explore the relationship between variables
Pearson “r”
R=+1.00: perfectly positive
R=-1.00: perfectly negative
R=0.00: no relationship between variables
Four Assumptions of Pearson “r”
- Interval or ratio data, need to calculate mean because using deviation scores
- Has to be linear relationship, a curvilinear relationship will produce r=0
- Radom sample
- Data is normally distributed
Spearman rs
Skewed data possible
Ordinal data
Uses D (x-y) ΣD=0
Strength of Correlation
R=|.20|~ weak relationship
R=|.30|~ moderate relationship
R=|.40|~ strong relationship
R=|.60|~ very strong relationship
Linear Regression
Prediction-used to predict the criterion and determine if variables are related
y’=”y prime”
Linear Regression Line
allows us to find the formula for the linear regression line
this is the straight line that minimizes the average error from each data point
y’=bx+a
b=slope
a=y-intercept
Calculating Regression
Df=N-2 (2 variables in correlation)
Always use =.05
You have homoscedastic data (both x and y variables are normal)
|r| > 0 (rcv) Significant Correlation
|r|=0 Non-significant Correlation-no relationship correlation
Ex:N=82 r=|.43|
df=82-2=80
rcv=.217 Significant
Calculating Regression
1st find b (the slope)
Then find a (the intercept)
Ways to determine line
- (mean of X, Y)
- (0, a) a=mean y - (b)(mean of x)
- (X, Y’) y’=bx+a
Error
On average, how far off are our predictions
r^2
Proportion of variance accounted for
1-r^2
Proportion of variance not accounted for
