True Experiments

Why is this Important?

How Do You Determine Causality! You need to have control of the situation if you want to have a reasonable chance of saying factors X and Y really affect outcome Z.

Hint: Review Workshops #7, #8 and #12 right now!

• Our Goal is the True Experiment - How do you determine what factors are casual when you have so many possible independent variables?

Solution: Control over all aspects of the situation like the What, the When, the Where and the How of the Experiment.

• A True Experiment prefers Random assignment to the various levels of the IVs as it hopefully avoids confounding with other variables.

• Nuance: If the experimenter must select subjects from pre-existing groups, the study is a quasi-experiment. Confounding is more of a risk here (see Workshop #14).

What Do You Need to Conduct a True Experiment?

ü Determine the possible IVs and DVs (See Workshop #12).

ü Is this particular study going to have one or more IVs.

Ü One IV is a single factor design (see Workshop #7)

Ü Two or more IVs make it a factorial design
(see Workshop #8). Your question is whether or not you predict an Interaction. This is when one level of an IV yields different effects of another IV.

ü Is the Experiment Going to Use Each Subject Only Once or More than Once?

Ü Is the Design Categorized as a Between Subject or Within Subject Design?(See Workshop #18).

ü You Will Need to Determine the Number of Levels (or Conditions or Treatments) for your Independent Variables (see Workshop #12)

ü You Will Need to Determine the Number of Subjects You Will Need for this Design to have sufficient Statistical Power (See Workshop #4).

You can have more than one Dependent variable in a true experiment. If you do - there are advanced statistical techinques like the Multivariate Analysis of Variance (MANOVA) that will handle this for you.

If you use a repeated Measures Design, you have to take in to account order or practice effects.

Common solutions are:

Block Randomizations: Orders of conditions randomized with all occurring before any repetitions.

Reverse Counterbalancing: ABBA or ABCCBA orderings of treatment levels.

Latin Squares Techniques:

A B C D

B C D A

C D A B

D A B C

 Bottom Line

The reason you use a true experimental design is that you want to have the best chance you can to determine causality. You want to understand causality as it gives you the best information necessary to predict and/or control psychological outcomes.

Thus, you must use manipulations of the independent variables to see if your hypotheses are correct. You must be careful to avoid confounds as then your results are not useful.

Without experimental research, it is difficult to really make the case that TV is related to violence or smoking causes cancer. Look how folks argue about these topics. However, a good experimental design can convince you.