Experimentation and Between Groups Design

How do you assure equivalence between groups?

Groups Must Be Equal

Extraneous Variables/ Confounds:

Experimenter bias

Demand characteristics

Instrumentation effect

Subject attrition

(systematic subject attrition)

• Random assignment - each participant is equally likely to be assigned to either group
• (remember: this is different from random selection)

• Matching pairs - identifying pairs so that the experimental groups are equivalent on one important characteristic (uses pretesting)

• Same people do both conditions (within-subjects design)

One definition of an experiment:

Experiments allow researchers to make causal statements.

If a subject variable is used as an independent variable to explore the effect on a dependent measure = correlational study.

Study in which groups are not randomly assigned can be correlational = looks at relationships.

Correlational studies do not allow researchers to make causal statements.

Why?

Participants in the experimental and control condition have equivalent personal characteristics, especially with respect to the dependent variable, before they take part in the experiment.

~*Random assignment*~

Between-Groups Design

(or between-subjects or independent-groups)

"The performance of participants in one or more groups is compared to the performance of participants in another group" (Schweigert, 2012, p.99).

Most common type of experimental design

Not all studies using between-groups design are experimental!

"An investigation in which the researcher manipulates an independent variable to see if there are any differences in the dependent variable" (Schweigert, 2012, p. 97).

Requirements for an experiment:

1. Manipulation of the independent variable

2. Groups being compared must be equal

3. Design must be free of confounds

Single-blind procedure: Either the participants don't know which experimental condition they are in or the experimenter doesn't know.

Double-blind procedure: Neither the participants nor the experimenter know which condition the participants have been assigned to.

Criteria for an experiment

1. Manipulation of the independent variable.

2. Holding all other variables constant.

3. Participants in the experimental and control condition have equivalent personal characteristics, especially with respect to the dependent variable, before they take part in the experiment.

3a. Random assignment.

If all of the above criteria are not met - quasi-experiment.

Experimentation and Between Groups Design

Experimentation

(and correlational studies/ quasi-experimental studies)

What is an experiment?

Do these meet the criteria of an experiment?

Facebook experiment

Milkshake experiment

Pay the monkey