Correlational & Experimental Methods

* Definitions
* Examples
* Comparisons
* Conclusions
* Designing our own studies

Correlational Method

Definition:
* A technique in which two or more variables are measured so that the relationship between them can be assessed statistically
* In other words, are X and Y associated?
* Can we use X to predict the level of Y or Y to predict the level of X?
* Random sampling is necessary to make results generalizable

Correlational Method (cont'd)
* Correlations are quantified using a scale from -1 to +1
* A positive correlation means that X and Y are associated such that the higher X is, the higher Y is and vice versa
* A negative correlation means that the higher X is, the lower Y is and vice versa
* The closer a correlation is to zero, the weaker the association is (when r = 0, there is no association at all between X and Y)
* Correlational Examples
The correlation of the size of a baseball teamís payroll with the number of games it wins is + .77
The correlation of how confident an eyewitness is and her accuracy is + .03
The correlation of how important school performance is to a student and his self-esteem on days when he is rejected from

Correlation and Causation
* Correlational designs and analyses tell us about association (whether X is related to Y), but not causation (whether X causes Y)
* At least three possibilities when X is correlated with Y:
X causes Y
Y causes X
Z causes both X and Y

Interpret These Correlations

* Owning pets is positively correlated with good health
* Pepsi consumption in any given month is positively correlated with street violence
* A realistic view of the self is negatively correlated with happiness
* Hours spent watching violent TV is positively correlated with violent behavior on the playground

Experimental Method

Definition:
* A technique in which participants are randomly assigned to different conditions so that the causal influence of some variable(s) on
other variable(s) can be examined.
* In other words, does X cause changes in Y?
* Does W cause changes in Y?
* Do W and X interact to cause changes in Y?

Experimental Method (cont'd)

* Independent Variable (IV): the variable a researcher manipulates to see the effect this has on other variables
* Dependent Variable (DV): the variable measured by a researcher to see the effect of manipulating the IV
* Probability level (p value) is used to determine the likelihood that changes in the IV influenced the DV

Experimental Design Considerations
* Internal Validity: Everything except the IV must be exactly the same between conditions to conclude that the IV caused changes
in the DV
* External Validity: Extent to which the results of a study can be applied and generalized to other people, settings, and situations
* Random assignment to conditions is necessary, otherwise IV is not the only difference between groups

Experimental Examples

To test the effect that happiness has on helping behavior:
* IV: compliment one group of subjects and insult the other group of subjects
* DV: measure how many papers they pick up when a confederate walks by them in the diag and drops an entire binder on the ground

Experiments often have more than one IV and/or more than one DV

Importance of Manipulation

* We can infer causality from experiments because the only difference between groups is the IV we have manipulated
* If the IV is not manipulated by the experimenter, the study is not a true experiment
* What does that mean for studies investigating gender, race, socioeconomic status, self-esteem, etc?

Correlational Method
* Allows for the study of variables we simply canít manipulate (race, gender)
* Allows for the study of variables we ethically canít manipulate (effects of smoking on health)
* Allows us to examine the relationship between several variables at once

Experimental Method
* Allows us to draw conclusions about causality
* Affords us a great deal of control over the variables being studied
* Allows us to look at the interaction of 2 or more IVs on the DV

Correlational Method
* Tells us nothing about the direction of causality

Experimental Method
* We might be tempted to simply correlate a slew of variables instead of planning more controlled studies with specific hypotheses
* Laboratory studies can be artificial and lacking in mundane realism
* Lab studies can require elaborate cover stories and may require us to deceive participants

Checklist for a "Good" Design
* specific hypotheses
* operational and conceptual variable definitions
* random sampling/assignment
* internal and external validity
* mundane and experimental realism
* informed consent
* replication (and eventually meta-analysis)
* use of various methods, including correlational, experimental, and observational

Go Back to Overheads and Activities Main Page
Go Back to 005 Main Page
Go Back to 007 Main Page