PSY 3010 CHP 8 & 10

Quasi-Experimental Design

Quasi-Experimental Design

A research design that at first glance appears to be experimental, but either lacks random assignment or a control group.

-Usually used in field settings
-The IV is measured before DV, so there is no directionality problem

-Moderate internal validity: lies in between non-experimental and a true experiemental

One-Group Designs

Type of Quasi-Experimental research in which the sample lacks a control group. Only one group is used.
-One group posttest only design
-One group pretest posttest design
-Interrupted time-series design

One Group Posttest Only Design

No control group. DV only measured once. Ex: A group is given treatment (IV manipulated), and then their attitudes/scores on a construct are assessed (DV)
-Low internal validity due to no control group
-Weakest type of quasi-experimental design

One Group Pretest-Posttest Design

No control group. Similar to a within-subjects design, where the DV is measured in a group before and after treatment.

Interrupted Time-Series Design

No control Group-type of one group design. The IV and Dv are measured on many frequent occasions. Then treatment is implemented. Afterward, the variable is measured again on a couple of occasions to see the effect on the DV.

History

Current events that occur between the first and second measurement of the DV. Can potentially explain the change in the DV between occasion 1 and 2

Maturation

Change in DV could be due to the population developing a trait that they would regardless of the treatment.
-Ex: Overweight children enter a weight loss program. Most of the children seem to be losing weight. However, this could be due to maturation. Children might lose weight as a result of growing, not the treatment.

Testing

Explains the change in DV over time. When the same test is administered multiple times, people might change their answer.

Instrumentation

Characteristics of a measuring instrument change over time
-ex: when researchers code behavior, they may be less careful in measuring behavior when they get fatigued
-ex: if a survey is administered with a lot of questions, participants may get fatigued by the end and may be less careful in reporting their answers

Regression to the Mean

Explains that when one scores super high on a test/construct in relation to their average score, their next measurement on that construct will regress closer to their mean score.

Spontaneous Remission

Explains that any change in DV is because people might randomly improving without any treatment at all

Nonequivalent Groups Design

Quasi-experimental study in which no random assignment is used.
-Posttest only nonequivalent groups design

-Pretest-posttest nonequivalent groups design

-Interrupted time-series design with nonequivalent groups

-Pretest-posttest design with switching replication design

-switching replication with treatment removal design

Posttest Only Nonequivalent Groups Design

Quasi-experimental study in which no random sampling is used. Differs from one-group posttest only in that there is a control group. One group is given the treatment, and then the DV is measured once in both groups.

Pretest-Posttest Nonequivalent Groups Design

Quasi-experimental study in which no random sampling is used. Differs from one-group pretest-posttest design in that there is a control group. DV is measured in both groups. Then, treatment is given to one group and then the DV is measured again.

Interrupted Time-Series Design With Nonequivalent Groups

Quasi-Experimental Study in which no random sampling is used. Differs from interrupted time-series design in that there is a control group. Construct/variables are measured in intervals in both groups before treatment and again in intervals after treatment.

Pretest posttest design With Switching Replication Design

Nonequivalent Groups quasi-experimental design in which no random sampling is used. Control group is present. Variable is measured before treatment in both groups (pretest). After one group receives the treatment and the DV is measured in both groups, the control group then receives the treatment. Then the DV is measured again.

Switching Replication With Treatment Removal Design.

Nonequivalent Groups Quasi-Experimental Design in which no random sampling is used. Control group is present. Variable is measured before treatment in both groups (pretest). After one group receives treatment and the DV is measured in both groups, the conditions of the participants switches, so that the experimental condition now becomes the control condition and receives no treatment. The control condition becomes the experimental condition and now receives treatment. Then DV is measured again.

Group Research

Focusing on studying a large group of people. Uses inferential statistics for generalizability.

-Pros: Ensures that the treatment works for the majority of the population
-Cons: Does not consider the individual or small individual differences

Single-Subject Research

Usually quantitative research. Studies the behavior of a group of small participants (usually 2-10). Uses visual inspection using plots and graphs. Uses judgement and percentage of non-overlapping data to determine whether treatment had an effect.

-Pros: Focuses on each individual and making treatment effective for each individual. Focuses more on individual behavior rather than group means. Usually occurs in real life conditions and participants act as control and treatment condition (within groups).
-Cons: Findings may have a harder time generalizing to the population.

Types of Single Subject Research

-ABA Design
-Multiple Treatment Reversal Design
-Alternating Treatment Design
-Multiple Baseline Across Participants
-Multiple Baseline Across Behaviors
-Multiple Baseline Across Settings

Steady State Strategy

Participants wait until the variables measured are consistent in participants before changing the condition.

ABA Design

Applied Behavioral Analysis Design, also called the ABA reversal Design. A participant’s baseline on a variable is measured. Then a treatment is introduced and the variable is measured again. Then the treatment is removed and the variable is measured again.

Reversal Design

A design in which a baseline is measured, then treatment is administered, and then treatment is removed and baseline measured again.

Baseline

A participant’s score on a variable before treatment.

Multiple-Treatment Reversal Design

Similar to ABA reversal design, but introducing multiple treatments.
A: Baseline
B: Treatment 1
C: Treatment 2
A: Baseline

Alternating Treatment Design

Research design in which multiple treatments are used and altered in their use. Treatments switch from one to the other quickly along with the order on a regular schedule
A: Baseline
B

C
A
C
B

Multiple Baseline Design

multiple baselines are either established for one participant or one baseline is established for many participants.

-Multiple Baseline Participants
-Multiple Baseline Design Behaviors
-Multiple Baseline Design Setting

Multiple Baseline Across Participants

AB Design (no removal). Take a baseline from each participant and introduce the condition at
different times. Reduces the likelihood that changes are coincidental.

Multiple Baseline Design Across Behaviors

Taking multiple baselines for the same
participant for different dependent
variables. Introduce the treatment at
different times for each dependent
variable.

Multiple Baseline Design Across Settings

Taking multiple baselines for the same
participant in different settings

Visual Inspection

Statistics are often not used. Instead, participant scores are plotted and scientists may make a judgement based on the how much IV impacted DV ex; super low in one high in another

Level

Amount of DV changes from one condition to another

Trend

In observations, DV tends to Increase/decrease in across observations.

Latency

Time it takes for DV to change after condition change shorter changes mean it is more likely for effect to occur.

Percentage Of Non-Overlapping Data

Percentage of responses in treatment conditions that are more extreme than the most extreme in relevant control conditions for one individual. What is the highest response/the lowest? The highest response can tell us something.

Converging Evidence

Examine the pattern of flaws running through the research literature because the pattern can either support or undermine the conclusions we wish to draw.