Correlational Research

Question Asked with Experimental Designs:

Question Asked with Correlational Designs:

I. Characteristics of Correlational Research

A.

B.

II. If a correlation exists between 2 variables,

example:

A. Predictor variable -

B. Criterion variable -

III. Types of correlational relationships -

2 characteristics of correlations:

A. Direction:

Positive Correlation -

example 1: positive correlation between studying and amount learned.

example 2: between price and quality in shopping

example 3: in children, between age and coordination skills

Negative Correlation -

example 1: Negative correlation between job satisfaction and # days absent from work

example 2: between amount of soup sold and temperature outside

example 3: between stress level and overall health

Curvilinear Correlations -

example: Level of arousal with performance on a task

Uncorrelated -

example: amount of creativity with shoe size

B. Magnitude:

Typically Correlations are not perfect.

1. Strong: Most of the time when variable A is high, variable B is high.

2. Moderate: An above average amount of the time when variable A is high, variable B is high.

3. Weak: A small amount of the time when variable A is high, variable B is high.

The direction and magnitude/strength of a correlation can be statistically measured using a correlation coefficient.

Correlation coefficient is a descriptive statistic!

a. It falls within the range of -1 to +1.

(1) A correlation coefficient of -1

(2) A correlation coefficient of +1

(3) A correlation coefficient of 0

b. The stronger a correlation, the closer it will be to -1 or +1.

(1) .99

(2) .10

(3) -.80

(4) -.25

Correlation does NOT imply causation!!!!

example: A study showing a positive correlation between # of hours of violent TV watching and amount of aggressive behavior in children.

Tempted to conclude that watching more violent television causes children to act more violently.

BUT . . .

2 problems prevent us from drawing a causal relationship from correlational designs

1. Directionality Problem:

2 possible interpretations

1.

2.

2. Third-Variable Problem:

example: there is a positive correlation between the number of churches in a city and the amount of crime.

Does increased religiousness cause more crime?

Does more crime cause more religiousness?

Therefore, can only conclude that there is a relationship between the variables of some kind, but cannot conclude that it is a causal one.

IV. Why use a Correlational Design??

A.

B.

C.

D.

Regression analysis

A correlation tells you how strong the relationship is between 2 variables, and if it is significant.

A regression analysis

1.

2.

Example: Using SAT scores to predict how well a person does in college.

Multiple Regression

1. More than one variable used to predict another variable.

example: