Posted: February 26th, 2022

The Keystone of the Experimental Method

Control: The Keystone of the Experimental Method

Research Methods in Psychology

You will Learn:

Control Achieved Through Participant Assignment and Selection

Randomization

Control Achieved Through Experimental Design

Control as Related to the Logic of Experimentation

Control achieved through participant assignment and selection

Random Selection from a large subject pool produces a representative sample

Asking for volunteers from a large subject pool who are informed of the nature of the study before volunteering does not produce a representative sample

Control achieved through participant assignment and selection

Random Number Tables may be useful in selecting subjects from a larger pool of potential subjects.

Volunteers gather in one room and you assign them a number.

Using the random number table you then select those who’s numbers show up in the table.

100 volunteers, use 2 digits (11, 74, 26, 93, 81, 44, etc.) until you have the number of subjects you want.

Control achieved through participant assignment and selection

With the total number of subjects randomly selected from the larger pool of volunteers you can now randomly assign subjects to the different groups using the random numbers table again.

Three Groups? Use 1,2,3 only.

First subject = group 1

Second subject = group 1

Third subject = group 2

Fourth subject = group 3

Control achieved through participant assignment and selection

Elimination Procedure

Equating Procedure

Use Random Number Table to assign subjects to groups, but assign equal numbers of subjects with same characteristics to each group.

Eliminate Subjects In Order to Achieve Control

Results Then Apply Only to Those With Same Characteristics as Sample

Control achieved through participant assignment and selection

Counterbalancing Procedure is another good way to assign subjects to groups

Incorrect: First person = group 1, Second person = group 2, Third person = group 1, etc.

Incorrect: First male = group 1, First female = group 1, Second male = group 2, Second female = group 2, Third male = group 1, Third female = group 1, etc.

Correct: First male = group 1, First female = group 1, Second male = group 2, Second female = group 2, Third male = group 2, Third female = group 2, Fourth male = group 1, Fourth female = group 1, Fifth male = group 1, Fifth female = group 1, Sixth male = group 2, Sixth female = group 2, etc. (two by two)

Randomization

Random Sampling

Random Assignment

All potential participants have the same and equal chance of being chosen for participation.

University Participant Pools are composed of General Psychology Students (mostly freshmen, mostly young).

University Participant Pools are not the best sources for participants.

Assigning randomly sampled subjects to groups is also a random process.

Equating and Counterbalancing are not Random Assignments.

Random Assignment is only way to ensure all groups are equal on all characteristics (known and unknown).

Differences are not eliminated, but affect both groups equally.

Control Achieved Through Experimental Design

Experimental Design

Like a blueprint experimental design tells us how to “build” our study

Two Functions

1. able to determine the effect of an independent variable on a dependent variable

2. able to rule out confounds as alternative explanations

Control Achieved Through Experimental Design

Two types of variables

Independent variable (IV) – the factor or condition that is manipulated by the experimenter

Dependent variable (DV) – the variable that is measured, thought to be changed be the IV

Have at least two groups.

Experimental group – A group that is exposed to all experimental conditions including the independent variable or variable of interest

Control group – A group that is subjected to conditions almost identical to the experimental group but for whom the independent variable is not manipulated.

Participants must be assigned to groups randomly so that all participants have an equal chance of being assigned to any of the conditions.

Seek to avoid extraneous variables

Extraneous variables – any variables other than the independent variable that are likely to influence the dependent variable

Types of Experimental Design

Posttest-Only Control Group Design

Pretest-Posttest Control Group Design

Randomly assign people to groups

Group A: Give them treatment, then Measure

Group B: Gets no Treatment, then Measure

Does not rule out some of the Threats to Internal Validity

Randomly assign people to groups

Group A: Measure, then Treatment, then Measure again

Group B: Measure, then No Treatment, then Measure again

Controls Threats to Internal Validity

But does pretest introduce a confound?

Types of Experimental Design

The Solomon Four-Group Design

Randomly assign people to groups

Group 1: Pretest, then Treatment, then Posttest

Group 2: Pretest, then No Treatment, then Posttest

Group 3: No Pretest, then Treatment, then Posttest

Group 4: No Pretest, then No Treatment, then Posttest

Now the effect of a Pretest, if any, can be evaluated.

Solomon Four Group Design

Experimental Design

Four Characteristics of True Experiments

1. Random Assignment to Groups

2. Must have at least Two Levels of the Independent Variable (IV): Control Group and Experimental Group

3. Controls for Threats to Internal Validity

4. Compares two alternative Theories (but not always maybe)

Control as Related to the Logic of Experimentation

Random Selection and Random Assignment ensures groups are equal before the experiment begins.

A pretest may be given to double-check.

Is Experimental Design adequate to Test the Hypothesis?

Treat All Groups Exactly the Same, except for the Levels of the IV.

Interpretation of Results

You Have Now Learned:

Control Achieved Through Participant Assignment and Selection

Randomization

Control Achieved Through Experimental Design

Control as Related to the Logic of Experimentation

Order | Check Discount