1
Quasi-experimental Design
Jackson (2012) Chapter Exercises
#2. The psychology professor has two sections of students that were not randomly
assigned. The treatment will be weekly quizzes. The desired outcome is improved student
learning. I will assume that student learning is measured by the total score on all exams not
including the weekly quizzes. This design can be diagrammed with the following notation:
N X O
N O
Based on these assumptions I would recommend a nonequivalent control group posttest only
design for this quasi experiment.
If student learning is measured by scores on major exams through the course, however,
the notation changes, as would the recommendation. The following notation represents an
alternative possibility for this design:
N X O X O
N O O
Based on these assumptions, I would recommend a nonequivalent control group time-series
design for this experiment. Overall, this later design is a stronger design because of the multiple
observations and points of comparison between the two sections.
#4. Usually the nonequivalent control group design uses intact groups. Since these
groups are not randomly selected, the major confound for this design is selection bias. Selection
bias arises when the control and experimental groups are not comparable before the study and
gives an alternative explanation for any differences in the posttest. In this design there are a
couple of social interaction threats that are possible. Students in the experimental section may

Quantitative Research 2
find out that students in the control section are not getting quizzes and react negatively, causing
resentful demoralization, or react competitively, causing compensatory rivalry. Either case
would tend “to equalize the outcomes between groups, minimizing the chance of seeing a
program effect even if there is one” (Trochim & Donnelly, 2008, p. 171).
A single-group design’s worst confound is that there is no comparison group, therefore, if
student scores are high or low, it is impossible to tell if they are that way because of the
treatment, or because of some alternative explanation. With only a single group to measure, all
of the single group threats to internal validity are possible. In this particular study the most
likely would be a history threat, where an event or set of events could affect the outcome more
than the treatment.
#6. Three reasons a researcher might choose a single-case design include, (a) when a
single person or condition is of interest, (b) when replication of results is essential, and (c)
situations in which error variance needs to be eliminated (Jackson, 2012). Clinical trials often
are interested in the effect of a treatment on a single individual. One study was interested in
mitigating or eliminating auditory hallucinations and delusions as a result of schizophrenia, and
evaluated the use of “an innovative rational-emotive cognitive treatment” (Qumtin, Belanger, &
Lamontagne, 2012, p. 114). From this treatment there was a noticeable and immediate reduction
in depression and anxiety and an increase in the patient’s quality of life that extended through the
12-month follow up. A discussion regarding replication of empirical results resulting from
single-case designs in psychology and education was written by Kratochwill and Levin (2010)
along with suggestions for improving the credibility of these designs by using randomization.
Error variance results from differences between participants in a group. In a single-case design

Quantitative Research 3
there is no group, hence no error variance; making the determination of outcomes resulting from
the independent variable much less complicated.
#8. Single case designs can be implemented in a number of ways. In a reversal design
the focus is on a single participant, a single behavior, or a single situation, and involves the
independent variable being applied and removed one or more times to assess its impact. By
evaluating the behavior without the treatment applied and then comparing the behavior after
treatment, a determination can be made regarding the treatments effect. Succeeding periods of
non-treatment are called the reversal and can demonstrate what happens to the behavior with a
return to the baseline. A reversal design is very similar to a within-subjects group design
because each subject experiences both the control and experimental condition. Greenhoot (2003)
summarized that the reversal of an effect because of the removal of a treatment provided “strong
evidence for a causal link between the independent and dependent variable” (p. 98).
In a multiple-baseline design multiple participants, behaviors, or situations may be
involved. In these single case studies the measures of interest are individualized and descriptive
statistics are not used to aggregate them. The multiple-baseline design measures the effect of the
application of a treatment at differing times for multiple subjects, or can be used with a single
subject by applying the treatment in conjunction with different behaviors or in different
scenarios.
Part I Assignment Question Answers
Describe the advantages and disadvantages of quasi-experiments? What is the
fundamental weakness of a quasi-experimental design? Why is it a weakness? Does its
weakness always matter? A quasi-experimental design has certain advantages, including; (a)
the ability to “draw slightly stronger conclusions than . . . with correlational research” (Jackson,

Quantitative Research 4
2012, p. 342), (b) contributing original research to the body of knowledge in a field (Ellis &
Leavy, 2011), (c) allowing real-world research in the field as opposed to more controlled
conditions (Jackson, 2012), (d) may involve an independent variable that is nonmanipulated
(Hoadley, 2004), and (e) can be used with intact groups (de Anda, 2007). The disadvantages of
using a quasi-experimental design are that (a) it “limits internal validity in a study” (Jackson,
2012, p. 342), (b) does not establish a causal relationship between variables (Jackson, 2012),
while (c) probabilistic equivalence cannot be assumed (Dimitrov & Rumrill, 2003). The
fundamental weakness of a quasi-experimental design, and what makes it “quasi”, is that it does
not involve random assignment of participants to experimental conditions (Greenhoot, 2003).
Nonrandom assignment weakens internal validity, diminishes the ability to establish cause-and-
effect by eliminating the expectation that groups are equivalent. This weakness does not exclude
quasi-experiments from a researchers arsenal because of the advantages mentioned above. These
advantages make quasi-experiments the most used form of quantitative research (Ellis & Leavy,
2011). There may be times when subjects cannot be randomly assigned to groups, there is only
one subject of interest, or the topic of research either exists or it does not; in all of these cases a
quasi-experiment is a viable option for conducting the research while an experiment is not
(Greenhoot, 2003).
If you randomly assign participants to groups, can you assume the groups are
equivalent at the beginning of the study? At the end? Why or why not? If you cannot
assume equivalence at either end, what can you do? Please explain. If participants are
randomly assigned to groups from the same population of interest to the study, the groups can be
said to be probabilistically equivalent at the beginning of a study (Trochim & Donnelly, 2008).
Random assignment does not ensure groups are exactly the same, but that differences between

Analysis and Evaluation of Quasi-Experimental Designs in Educational Research

Study Now!

Document Details

Related Documents

The Statistics of Inheritance

Estimation and Hypothesis Testing

Normal Distribution - Amount of Sleep

SPSS Cumulative Assessment Instructions

Probability and Statistics Assignment

Lab Activity 5 Sampling Distributions

Hypothesis Testing � Comparing Two Groups

Applying Hypothesis Tests Discussion #8

STAT 250-004 Data Analysis Assignment 4

Inferential Statistics Week 2 Solution

Company

Explore

Study Tools