QuasiExperimental Research - PowerPoint PPT Presentation

1 / 32
About This Presentation
Title:

QuasiExperimental Research

Description:

... involve a comparison of conditions. Will not involve: The manipulation ... Random assignment of participants to conditions. Some Examples of Quasi-Experiments ... – PowerPoint PPT presentation

Number of Views:27
Avg rating:3.0/5.0
Slides: 33
Provided by: gra4
Category:

less

Transcript and Presenter's Notes

Title: QuasiExperimental Research


1
Chapter 9
  • Quasi-Experimental Research

2
Quasi-experiments are almost experiments
3
The True Experiment
  • Manipulation of a variable
  • Control of extraneous variables
  • Random assignment of participants to conditions
  • Comparison of conditions

4
Sometimes the circumstances are such that we
cannot conduct a true experiment. In those
instances we might be able to conduct a
quasi-experiment.
5
The Quasi-Experiment
  • Will involve a comparison of conditions
  • Will not involve
  • The manipulation of a variable
  • Control of extraneous variables
  • Random assignment of participants to conditions

6
Some Examples of Quasi-Experiments
  • Looking for gender differences in verbal or
    mathematics ability
  • Looking for developmental trends among children
  • Evaluating the quality of schools

7
Inferring Cause and Effect from a Quasi-Experiment
Because quasi-experiments do not involve the
manipulation of an independent variable nor the
control of extraneous variables, it makes it
difficult or impossible to infer cause and effect
from a quasi experiment. Some quasi-experimental
designs are better than others.
8
Quasi-experimental designs fall into two general
groups
  • Non-equivalent groups designs
  • Time series designs

9
Non-equivalent groups designs
When the researcher cant control the assignment
of participants to the various groups or
conditions of the study
10
1. The Differential Research Design
The researcher compares pre-existing groups that
differ with regard to the variable of interest
11
1. The Differential Research Design
The researcher compares pre-existing groups that
differ with regard to the variable of
interest Example 1 Comparing the verbal or
mathematics ability of men and women.
12
1. The Differential Research Design
The researcher compares pre-existing groups that
differ with regard to the variable of
interest Example 1 Comparing the verbal or
mathematics ability of men and women. Example 2
Comparing the children from intact and broken
homes.
13
The difficulty with differential research
It is impossible to assert that the groups only
differ with regard to the independent variable.
The text refers to this as an assignment bias.
14
2. Post-test only non-equivalent control group
design
Following treatment a treatment group is
compared to a non-treatment control group X O
(treatment group) O (control group)
15
2. Post-test only non-equivalent control group
design
Following treatment a treatment group is
compared to a non-treatment control group X O
(treatment group) O (control group)
Example evaluation of an educational program.
16
Assignment bias is also a problem for the
post-test only nonequivalent control group
design.
17
3. Pretest-posttest nonequivalent control group
design
Observations are obtained from both groups prior
to treatment, one group receives the treatment,
then observations are obtained from both groups
again. O X O (treatment group) O O (control
group)
18
3. Pretest-posttest nonequivalent control group
design
Observations are obtained from both groups prior
to treatment, one group receives the treatment,
then observations are obtained from both groups
again. O X O (treatment group) O O (control
group)
Example evaluation of a medical or
psychotherapy.
19
The pretest-posttest nonequivalent control group
design allows the researcher to check for an
assignment bias prior to the treatment.
20
Time Series Designs
A series of observations is made over time of a
single group on individuals
21
1. One-group pretest-post test design
Observations are made both before and after a
treatment O X O
22
1. One-group pretest-post test design
Observations are made both before and after a
treatment O X O Example Evaluation of a
particular educational program or therapy.
23
Time related factors that threaten validity
  • History effect external events might have an
    effect
  • Instrumentation measurement instrument might
    change over time
  • Practice/fatigue effects scores might improve
    from practice or deteriorate from fatigue
  • Maturation observed changes might reflect
    physical or psychological maturation
  • Regression toward the mean extreme scores on
    the first observation will be less extreme on the
    second

24
2. Time series design
  • A series of observations are made both prior to
    and following a treatment.
  • O O O X O O O

25
3. Equivalent time-samples design
O O O X O O O N O O O X O O O N O O O
Works well for treatment effects that are
temporary.
26
Developmental Research Designs
Designed to assess the effects of age
27
1. Cross-sectional design
Design compares different groups of individuals
that differ with regard to age
28
1. Cross-sectional design
Design compares different groups of individuals
that differ with regard to age Example
comparing cognitive abilities of 50-year-olds,
60-year-olds, 70-year-olds, and 80-year-olds
29
1. Cross-sectional design
Design compares different groups of individuals
that differ with regard to age Example
comparing cognitive abilities of 50-year-olds,
60-year-olds, 70-year-olds, and
80-year-olds Cohort effects (i.e., differences
between cohorts that are unrelated to age) are a
particular threat to the validity of
cross-sectional designs
30
2. Longitudinal designs
Sample same individuals at different points in
time.
31
2. Longitudinal designs
Sample same individuals at different points in
time. Example Compare cognitive ability of a
group of individuals when they 50, 60, 70, and 80
years old.
32
2. Longitudinal designs
Sample same individuals at different points in
time. Example Compare cognitive ability of a
group of individuals when they 50, 60, 70, and 80
years old. Difficulties with longitudinal designs
mortality (or drop-out rate) and expense.
Write a Comment
User Comments (0)
About PowerShow.com